Databricks intros OpenSharing, a new standard for sharing AI

Copyright TechTarget - All rights reserved https://cyber.law.harvard.edu/rss/rss.html Techtarget Feed Generator en Sat, 13 Jun 2026 17:53:53 GMT https://www.techtarget.com/searchdatamanagement editor@techtarget.com Just as Databricks in 2021 developed an open standard that enabled enterprises to securely share data both internally and with external partners, the vendor has introduced OpenSharing so that organizations can share AI assets. Five years ago, Databricks built Delta Sharing, a sub-project within the open source <a href="https://www.techtarget.com/searchdatamanagement/news/366542953/Databricks-introduces-Delta-Lake-30-to-help-unify-data">Delta Lake project</a>, so that enterprises could share data -- without moving or duplicating it -- to fuel collaboration. Launched on June 10 and now available on GitHub, OpenSharing, which is hosted by the Linux Foundation, is an extension of Delta Sharing. The new open-source standard allows organizations to further foster collaboration across platforms, departments and with partners by sharing AI models, agent skills with specialized knowledge and workflows, and <a href="https://www.techtarget.com/searchdatamanagement/feature/Unstructured-data-needed-but-often-untapped-for-agentic-AI">unstructured data</a>. In addition, OpenSharing broadens collaborative efforts by adding support for platforms that connect to the <a href="https://iceberg.apache.org/rest-catalog-spec/">Apache Iceberg REST Catalog</a>, which enables sharing between a new set of organizations, and adds partnerships with on-premises storage partners to enable no-movement sharing of on-premises data and AI assets. "OpenSharing marks a shift from simple data exchange to a unified, governed interface for the AI and data stack," William McKnight, president of McKnight Consulting, told TechTarget. "Beyond traditional tables … this framework provides a blueprint for studying and scaling how autonomous agents interact with distributed data. This could be quite significant for data sharing." Stephen Catanzano, an analyst at Omdia, a division of Informa TechTarget, similarly called OpenSharing an important development. "OpenSharing is a solid development in the AI infrastructure landscape," he told TechTarget. "What makes it particularly important is that it extends secure, zero-copy sharing beyond structured data to include agent skills and AI models -- assets that are becoming critical in the agentic era. Previously, organizations had no standardized way to share these AI components across platforms." Based in San Francisco, Databricks was a pioneer of the <a href="https://www.techtarget.com/searchdatamanagement/news/366545117/Lakehouse-architecture-the-best-fit-for-modern-data-needs">data lakehouse</a> format for storing data. Like many data management vendors, Databricks, which from its founding in 2013 included machine learning capabilities, in recent years has focused much of its product development on enabling customers to <a href="https://www.techtarget.com/searchdatamanagement/news/366638723/Databricks-launches-PostgreSQL-Lakebase-to-aid-AI-developers">build generative AI and agentic AI capabilities</a>. <section class="section main-article-chapter" data-menu-title="Sharing AI"> <h2 class="section-title">Sharing AI</h2> Collaboration, whether within a single department, across organizational domains or with third-party partners, is an important means of speeding innovation and improving productivity. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> OpenSharing marks a shift from simple data exchange to a unified, governed interface for the AI and data stack. Beyond traditional tables … this framework provides a blueprint for studying and scaling how autonomous agents interact with distributed data. This could be quite significant for data sharing. </figure> <figcaption> William McKnightPresident, McKnight Consulting </figcaption> </div> </blockquote> Historically, collaborative efforts were informed by data. As a result, particularly during and in the immediate aftermath of the COVID-19 pandemic, many data management and analytics vendors <a href="https://www.techtarget.com/searchbusinessanalytics/feature/Enabling-collaboration-a-rising-analytics-trend">built collaboration capabilities</a> into their platforms to enable workers in remote locations to work together in a virtual hub. Now, as <a target="_blank" href="https://kpmg.com/us/en/media/news/q1-ai-pulse2026.html" rel="noopener">more enterprises build agents</a> and other AI tools to assist employees and execute certain business processes, collaborative efforts include AI. Not only are collaborative projects built on data and data products such as reports and dashboards, but they include agents and other AI assets. However, without a standard means of sharing AI assets, organizations are forced to piece together their own way of doing so, which must be done over and over again each time a department or business tries to collaborate with a department or business that doesn't use the exact same tools. OpenSharing was developed to provide a standard, repeatable means of sharing AI to enable collaboration, according to Akram Chetibi, director of product management at Databricks. "AI created a problem nobody had really solved yet," he told TechTarget. "When you look at how organizations are starting to share things across company boundaries -- not just data tables, but AI models, agent skills, prompts, and tools -- there was no standard way to do it. Everyone was cobbling together their own approach." Given Databricks' experience developing <a target="_blank" href="https://delta.io/sharing/" rel="noopener">Delta Sharing</a>, OpenSharing was a logical extension, Chetibi continued. "We'd already seen this problem play out with structured data before Delta Sharing existed, and we didn't want the AI world to repeat it," he said. "OpenSharing fixes that by giving organizations a single open protocol for publishing and consuming AI assets … regardless of which platform either side runs on." Specific benefits of OpenSharing include the following: <ul type="disc" class="default-list"> <li>An open protocol for publishing and sharing data and AI assets so files don't need to be manually copied for each collaborative initiative.</li> <li><a href="https://www.techtarget.com/searchapparchitecture/definition/application-program-interface-API">APIs</a> for discovery, authorization and access irrespective of the platforms that different departments and organizations use for managing data and AI tools.</li> <li>Support for Apache Iceberg APIs, expanding the reach of Delta Sharing to organizations that use Iceberg-native tools.</li> <li>Integration with <a href="https://www.techtarget.com/searchenterpriseai/news/366617361/Enterprises-shift-to-on-premises-AI-to-control-costs">on-premises</a> platforms and <a href="https://www.techtarget.com/searchcloudcomputing/definition/private-cloud">private clouds</a> to provide enterprises electing to keep their AI operations in more secure environments than public clouds with the same collaborative capabilities as those using public clouds.</li> </ul> As agentic AI systems proliferate, the sharing requirements to collaborate are evolving away from just <a href="https://www.techtarget.com/searchdatamanagement/opinion/Why-agentic-AI-demands-both-structured-and-unstructured-data">structured datasets</a> to include new, complex assets, Catanzano noted. OpenSharing addresses some of the challenges that have risen as a result, including the multi-cloud and hybrid infrastructures across which AI assets are spread. "OpenSharing addresses this new complexity by providing a unified protocol that works across these fragmented environments, enabling the kind of seamless AI collaboration that the current landscape demands but couldn't previously support," Catanzano said. In addition, given that agents themselves are becoming some of the primary consumers of data, <a target="_blank" href="https://www.evidentlyai.com/blog/ai-agents-examples" rel="noopener">executors of workloads</a> and collaborators across domains, a standardized way to share data and AI assets is beneficial, according to McKnight. "These systems read data and the underlying metadata like model weights and execution code. OpenSharing treats this like a single, governed package," he said. "Furthermore, modern enterprises are running fast and consequently have many lakehouses and open data formats, so organizations need a vendor-neutral, open standard to share assets without the cost and lag of copying data." </section> <section class="section main-article-chapter" data-menu-title="Standardizing documents"> <h2 class="section-title">Standardizing documents</h2> Beyond OpenSharing, in other open source development news, the LF AI & Data Foundation revealed the formation of the DocLang Specification Working Group. The group was formed to develop DocLang, an open, AI-native document format to standardize preparing, exchanging and governing document data for AI systems. Catanzano noted that most enterprise knowledge is stored in documents such as PDFs, Word files and presentation slides. Extracting such document data and operationalizing it for AI can be difficult and time-consuming, slowing AI initiatives. The DocLang working group, therefore, is addressing a significant <a href="https://www.techtarget.com/searchnetworking/definition/bottleneck">bottleneck</a> in AI adoption, according to Catanzano. "The combination of DocLang -- the standard -- with the open-source processing toolkit Docling creates a complete stack for document AI, which could accelerate enterprise AI adoption by making document understanding more deterministic and interoperable across systems," he said. "This is particularly timely as agentic AI systems increasingly need to work with unstructured enterprise documents at scale." McKnight, meanwhile, theorized that formation of a working group signals a rising emphasis on AI-native document data. "The launch … is where an industry shifts from fragmentation toward fragmentation-killing collaboration," he said. "It's the beginning of a shift in the foundations to AI-native documents and interactions." While both OpenSharing and the effort to build DocLang are individually valuable, they continue an ongoing trend of developing open standards to enable AI development, deployment and management. The Model Context Protocol, an open source set of code released by Anthropic in November 2024 that standardizes how AI models connect to an organization's proprietary data sources, <a href="https://www.techtarget.com/searchdatamanagement/feature/One-year-of-MCP-Support-a-must-for-data-management-vendors">has become widely adopted</a>. The Agent2Agent Protocol, introduced by Google Cloud in May 2025, provides a standard for agent interactions. Building and managing agents and other AI tools using open-source capabilities is valuable because it <a href="https://www.techtarget.com/searchenterpriseai/tip/Best-practices-to-avoid-AI-vendor-lock-in">allows enterprises to remain flexible</a>, according to Chetibi. That's why Databricks elected to make OpenSharing open source rather than keep it a proprietary feature within the broader Databricks platform. "Customers and partners don't want to collaborate with their data and AI assets while being confined to a single vendor proprietary ecosystem," Chetibi said. "It simply doesn't stick because it constrains innovation." Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> The open source protocol modernizes collaboration by enabling teams to share AI assets such as AI models and agent skills across domains and with external partners. https://cdn.ttgtmedia.com/rms/onlineimages/collab_a362306286.jpg https://www.techtarget.com/searchdatamanagement/news/366644177/Databricks-intros-OpenSharing-a-new-standard-for-sharing-AI Thu, 11 Jun 2026 14:05:00 GMT Databricks intros OpenSharing, a new standard for sharing AI Unstructured data is critical in the age of AI. However, just as when most enterprise data initiatives focused on building analytics tools such as reports and dashboards, unstructured data is underutilized now that organizations are building agents and other AI applications capable of autonomously generating insights and executing business processes. <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/petrie_kevin.jpg" alt="BARC U.S. analyst Kevin Petrie">Kevin Petrie </div> While structured data such as financial records and point-of-sale transactions provides key information and is critical when building analytics and AI tools, unstructured data such as text in documents and emails and audio from customer interactions <a href="https://www.techtarget.com/searchbusinessanalytics/feature/Ng-Biggest-benefit-of-AI-may-be-unlocking-unstructured-data">adds vital context</a> that structured data can't provide. Autonomous AI applications require vast amounts of high-quality, contextually relevant data to deliver trustworthy outputs. Without enough data, agentic AI tools won't understand<a href="https://www.techtarget.com/whatis/definition/business-logic"></a> an organization's unique characteristics and will make up responses. Often, those made-up outputs are so bizarre that they're easy for humans to dismiss. Sometimes, however, they are plausible enough to fool people, which can lead to misinformation that <a href="https://www.evidentlyai.com/blog/ai-hallucinations-examples">causes a business significant harm</a>. It's estimated that unstructured data now makes up <a href="https://mitsloan.mit.edu/ideas-made-to-matter/tapping-power-unstructured-data">as much as 90% of all data</a>. Unstructured data, therefore, can be the difference between an agent that doesn't have enough context to properly perform and never makes it beyond the pilot stage, and an agent that generates substantial business value for an organization. According to <a href="https://barc.com/research/harnessing-unstructured-data-for-ai-innovation/">a report</a> from research and advisory firm BARC titled "Harnessing Unstructured Data for AI Innovation: Problems, Practices, and Principles for Success," nearly three-quarters of organizations report that less than 50% of their unstructured data is discoverable and can be used to inform decisions. That means that even as investments in AI development continue to surge and organizations deploying agents have potential competitive advantages, most enterprises are still unprepared for building agents. In addition to finding that many organizations report that they don't have systems in place that enable them to discover their unstructured data for AI, the report, authored by BARC analysts and Kevin Petrie and Merv Adrian and co-sponsored by Datahub and Ohalo, found that <a href="https://www.techtarget.com/searchbusinessanalytics/feature/GenAI-demands-greater-emphasis-on-data-quality">data quality</a> problems such as inaccuracy and inconsistency hinder the operationalization of unstructured data. In a recent interview, Perie discussed the report, including the importance of unstructured data in AI development and why organizations still struggle to make appropriate use of it. In addition, he spoke about the consequences of failing to operationalize unstructured data for AI, the benefits of successfully using unstructured data to inform AI, and how organizations can go about tapping into their unstructured data. Editor's note: This Q&A has been edited for clarity and conciseness. Why is unstructured data important for agents -- what is it about agents and other AI applications that make structured data not enough? Kevin Petrie: Structured data -- in other words, <a href="https://www.techtarget.com/whatis/definition/table">tables</a> -- remain the top input for AI models overall because they contain the most easily verifiable facts. However, unstructured objects such as documents, images, emails and so on offer rich context that AI agents cannot get from tables. Unstructured data describes user intentions, stakeholder behavior patterns, company processes, customer sentiment, corporate values and myriad other factors that influence the decisions and actions of a business manager each day. Without this rich context, AI agents will be unable to reason like humans. They will fail to <a href="https://www.techtarget.com/searchcio/feature/Startup-founder-says-trust-is-biggest-barrier-to-AI-agents">generate trustworthy outputs</a>, take safe actions and deliver business value outside of a narrow range of use cases. This long tail of unstructured data represents the beating heart and conscience of a business. You have to capture and make sense of it to differentiate yourself and create true competitive advantage with agentic AI. Why do many organizations struggle to harness their unstructured data and derive value from it -- what are the problems that prevent them from tapping into unstructured data as a resource? <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> Unstructured data represents the beating heart and conscience of a business. You have to capture and make sense of it to differentiate yourself and create true competitive advantage with agentic AI. </figure> <figcaption> Kevin PetrieAnalyst, BARC U.S. </figcaption> </div> </blockquote> Petrie: Unstructured data has piled up in siloed and far-flung systems for years. We found that 52% of organizations have unstructured data in on-premises or hybrid database environments, and another 16% have it sitting in multiple cloud database platforms. Given this, 70% of organizations say that less than half of their unstructured data is discoverable and usable. Other obstacles include skill gaps, privacy requirements and immature <a href="https://www.techtarget.com/searchdatamanagement/feature/What-executives-look-for-in-a-data-quality-platform">data quality controls</a>. Most data teams have focused their initial agentic AI projects on operational tables and a selection of trusted documents. That's a safe way to start. But over time they clearly need to cast a wider net to address more sophisticated, value-generating use cases. As agentic AI becomes more ubiquitous, what are the competitive consequences of not tapping into unstructured data? Petrie: The reality is that large language models themselves do not create competitive advantage. Any enterprise can subscribe to <a href="https://www.techtarget.com/searchenterpriseai/news/366624572/Anthropic-intros-next-generation-of-Claude-AI-models">Anthropic Claude</a> and use it to make their knowledge workers more productive -- in fact, that's now table stakes to survive. But to truly differentiate your enterprise in the modern era, you need to integrate smart multimodal agents into your proprietary business processes. That requires the context that you can only get from the unstructured data sitting behind your firewall. If you cannot harness that unstructured data and tap that value, your agentic AI initiative will fail to differentiate your organization. You will be limited to lower-value use cases that your competitors will match. Conversely, as more organizations move agents into production, what are the competitive benefits of operationalizing unstructured data as a contextual source for agents? Petrie: That's where things get interesting. If you can classify, validate, and derive meaning from your customer service records, you can start to have agents prioritize and escalate complaints on <a href="https://www.techtarget.com/searchdatamanagement/opinion/Real-time-data-streaming-for-AI-invest-where-it-matters">a real-time basis</a>. If a hospital chain or pharmaceutical company can analyze more doctors' notes in less time, its caregivers might identify new methods of improving patient treatments. [Benefits include] use cases that can help improve customer satisfaction, reduce costs and increase revenue. For organizations still working to manage and operationalize unstructured data for AI, what is a blueprint -- what are the steps they need to take and the ideal technology stack they need to put together to build a system that organizes and prepares unstructured data for AI? Petrie: Merv Adrian and I recommend some specific steps for data and AI leaders to harness their unstructured data for AI projects. First, they must find, <a href="https://www.techtarget.com/searchdatamanagement/feature/16-top-data-catalog-software-tools-to-consider-using">prioritize and catalog</a> all this stuff. The more they can organize critical <a href="https://www.techtarget.com/whatis/definition/metadata">metadata</a> for mission-critical documents, and text records that humans consume on a regular basis, the better they can feed agents the necessary context to add value in business processes. Amazingly, only 38% of survey respondents have cataloged their unstructured data for AI. Second, they must extend their <a href="https://www.techtarget.com/searchdatabackup/tip/Enterprise-data-governance-Frameworks-and-best-practices">data governance programs</a> to address these unstructured objects, with smart human oversight. We found especially concerning gaps in data bias and lineage for unstructured data that will create agent chaos if left unaddressed. Only half of organizations have bias controls in place, and less than half trace lineage. Third, and perhaps most importantly, we recommend that data teams <a href="https://www.techtarget.com/searchdatamanagement/opinion/Why-data-semantics-matters-for-context-aware-systems">create an independent semantic layer</a> that can query and make sense of data wherever it lives. This is required because migration complexity, data gravity, and sovereignty concerns make full consolidation a non-starter for most AI adopters. With so many organizations -- 70% with less than half their unstructured data AI-ready -- still struggling to overcome the barriers that hold them back, how long do you think it will take for that number to reverse itself and most organizations to have AI-ready unstructured data? Petrie: That's a great question because it's hard to predict major shifts like this. I don't believe enterprises should try to reach 100% readiness, because inevitably some portion of that unstructured data will not add value for AI initiatives. But given the huge <a href="https://www.techtarget.com/searcherp/feature/Why-context-engineering-is-the-next-enterprise-software-priority">focus on context engineering</a>, I expect that most companies will have discovered and classified most of their unstructured data within the next 24 months. What is the state of AI-readiness when it comes to structured data -- are many organizations struggling to even manage their structured data for AI, or do they have a much better handle on that than they do their unstructured data? Petrie: While our survey did not investigate this, it's fair to say that structured data is overall much more ready. AI teams tend to use structured data first because it is cleaner, more organized and more accessible than any other data type. Database tables are the lifeblood of any organization, driving business functions such as finance, sales, operations, and so on. While data quality issues continue to plague most database environments, unstructured files <a href="https://www.computerweekly.com/news/366631618/Podcast-How-to-get-value-from-unstructured-data">pose a bigger challenge</a> and require more preparation. Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. AI development initiatives hinge on the quality and completeness of the underlying data, but research from BARC shows that many organizations struggle to operationalize key data. https://cdn.ttgtmedia.com/rms/onlineimages/machine%20learning_g1186820873.jpg https://www.techtarget.com/searchdatamanagement/feature/Unstructured-data-needed-but-often-untapped-for-agentic-AI Mon, 08 Jun 2026 14:03:00 GMT Unstructured data needed, but often untapped, for agentic AI The semantic layer, historically embedded in business intelligence tools, is becoming central to enterprise analytics as AI agents begin querying enterprise data directly. Without governed definitions, agents can produce answers that are consistent in execution but inconsistent in meaning. That arrangement worked as long as a person sat at the end of every query. A <a href="https://www.techtarget.com/searchdatabackup/tip/Ways-to-protect-data-platforms-from-turnover-risk">human analyst carried the institutional knowledge</a> to pick the right definition and catch a number that looked wrong. Agents carry none of that. They act on whatever definition they are handed or infer at machine speed across systems that might never have an analyst in the loop. AI has turned the semantic layer from something taken for granted into something <a href="https://www.techtarget.com/searchdatamanagement/opinion/Enterprise-data-platforms-adapt-for-GenAI-and-agentic-AI">enterprises have to decide deliberately</a>. An enterprise putting agents into production could, in theory, take a shortcut, letting the model approximate what the data means rather than defining it, but that has significant downstream consequences. An approximation no one can trace to a governed definition is an answer no one can defend when questioned. <section class="section main-article-chapter" data-menu-title="Agents changed the layer that was always there"> <h2 class="section-title">Agents changed the layer that was always there</h2> For most of its history, the semantic layer did its work quietly and in one place. It fixes what a business term means before anyone queries it, so that revenue or active customer resolves to one agreed definition rather than whatever the query writer assumes. Those definitions lived inside a company's BI tool, and that was enough for the time. The people querying the data worked closely with it, and the arrangement never had to travel. Now, <a href="https://www.techtarget.com/searchbusinessanalytics/feature/Agents-semantic-layers-among-top-data-analytics-trends">that has changed</a>. The definitions can no longer stay where they have always lived. "Having those [definitions] tied up in your BI tools doesn't work anymore, because you need AI agents to access them," said Chris Child, vice president of product for data engineering at Snowflake. The semantics themselves did not move, but who or what needs to access them is fundamentally different now. An enterprise agent works across an estate no one person holds in their head at once: object storage in one cloud, a warehouse built a decade ago, operational systems like ServiceNow and Salesforce, <a href="https://www.techtarget.com/searchdatamanagement/feature/Controlling-data-sprawl-requires-governance-discipline">data lakes federated across acquisitions</a>. Sergio Gago, chief technology officer at Cloudera, put the limit plainly. "The human, with all the domain knowledge and expertise inside the company, is able to navigate this complexity," he said. "But an agent cannot do that." The difference between the data an agent can reach and data it can correctly interpret is the gap the semantic layer fills. It is why a concept that held steady for years is being pulled into a part of the stack it has not occupied before. </section> <section class="section main-article-chapter" data-menu-title="A query that runs is not a query that's right"> <h2 class="section-title">A query that runs is not a query that's right</h2> The shortcut for building a semantic layer is to let the model write its own definitions on the fly. LLMs can produce SQL that executes, and an executing query feels like a successful answer, which is why the shortcut is easy to ship yet hard to trust. The distance between a query that runs and one that returns what the business asked for is where the semantic layer earns its place. Snowflake learned this on its own data, training its Arctic models on years of customer SQL, only to find they "were still not great at actually answering your real business questions," said Child. The models had the syntax and still missed the meaning. The turn came not from more data but from providing definitions. "We gave them access to semantic models, and they got dramatically better at answering the real question." The ambiguity runs deeper than edge cases. A question as basic as a company's customer count carries several correct answers at once -- the public figure, the internally tracked number, or the count of active accounts. No model resolves this ambiguity on its own. The number it returns will be confident and defensible in the context of whichever definition it chose, but no one can say which definition produced it. This is the failure mode that should worry a governance lead, because it <a href="https://www.techtarget.com/searchdatamanagement/feature/Why-AI-forces-securityfirst-governance">does not announce itself as a failure</a>. An agent given broad access will produce what looks like an answer, including tables never meant to be authoritative, and return it with the same confidence it brings to everything else. None of this is new work, which is the part enterprises keep missing. Defining what data means, documenting it and governing its use was always the standard practice, but human analysts absorbed the cost of skipping it, holding definitions in their heads through institutional knowledge and catching each other's errors. Agents take the definition they are handed or invent one, faster than any dashboard could expose that same data. </section> <section class="section main-article-chapter" data-menu-title="Which format are businesses using?"> <h2 class="section-title">Which format are businesses using?</h2> If the semantic layer is now the layer that matters, the contest becomes who controls its shape. A definition locked in one vendor's proprietary format is one the customer cannot take anywhere else, and a semantic layer's value is that an agent can read it wherever the data sits. That is the logic behind the <a href="https://www.nojitter.com/data-management/companies-come-together-to-standardize-data">Open Semantic Interchange</a>, the standard many vendors are backing to make semantic definitions portable across platforms. The problem it solves is duplication that compounds with each tool. An enterprise running two BI tools already maintains two semantic models, and pointing an agent at the same data demands a third. A standardized format lets a company define a metric once and use it everywhere instead of redefining it for each system that touches the data. Customers are already pushing in that direction, and not toward any single vendor. The typical enterprise now spreads its data across several platforms, Gago said, with “some data in Snowflake, some data in Databricks, some data in Cloudera,” and that sprawl is producing demand for open systems and open standards “that don’t lock them in.” A portable semantic layer is the version of that demand aimed at the layer agents actually consume. Many BI tools that once guarded their semantic definitions have signed on to the open standard, said Child, having concluded the definitions must be usable across many tools, not locked inside one. The contest over which format prevails is not settled, but the direction is. The vendors aligning to open standards become the obvious choice, and the holdouts betting on their proprietary definition being better are betting against a shift that their own peers have already joined. In the end, a definition is only worth as much as the number of places an agent can carry it. Scott Thompson is the Site Editor for TechTarget's Data Technologies group, covering data management and business analytics topics for senior enterprise data leaders. He has edited data and analytics content for TechTarget since 2021. </section> Semantic layers are moving from BI tools into the core analytics stack as AI agents query enterprise data, requiring governed definitions for consistent interpretation. https://cdn.ttgtmedia.com/rms/onlineimages/mobile_g1022892890.jpg https://www.techtarget.com/searchdatamanagement/feature/Why-enterprise-AI-depends-on-the-semantic-layer Mon, 08 Jun 2026 13:00:00 GMT Why enterprise AI depends on the semantic layer Graph technology specialist Neo4j is taking aim at a new market with the acquisition of GraphAware, an intelligence analysis vendor whose platform caters to government agencies. Financial terms of the deal between the longtime partners, which was revealed on June 3 and is expected to close during the third quarter of 2026 following regulatory approval, were not disclosed. Intelligence analysis capabilities enable users to collect, connect and analyze large amounts of <a href="https://www.techtarget.com/searchdatamanagement/opinion/Turning-data-into-a-strategic-advantage">fragmented or isolated data</a> to investigate relationships between data points and derive insights. Like <a href="https://www.techtarget.com/searchbusinessanalytics/news/252507769/Gartner-predicts-exponential-growth-of-graph-technology">graph databases</a> that similarly discover relationships across broad swaths of data differently than traditional relational databases, intelligence analysis platforms are frequently used by law enforcement, defense and cyber intelligence organizations. Based in London, GraphAware's most direct competitors <a href="https://www.computerweekly.com/news/366560657/Palantir-awarded-NHS-FDP-data-contract">include Palantir</a> and i2. The vendor's Hume platform is an AI-powered set of capabilities based on open standards and built on Neo4j's graph technology. Following the acquisition, GraphAware Hume is now part of Neo4j's graph intelligence platform, enabling Neo4j to expand beyond its established customer base to directly target government agencies and compete with leading intelligence analysis providers. In addition, it expands Neo4j's user base beyond application developers and data scientists to include analysts who generate insights and help make organizational decisions. Given that the acquisition evolves Neo4j beyond its database roots by providing purpose-built software that helps organizations deploy graph technology for complex investigative work, adding GraphAware's technology will be significant for Neo4j users, according to Stephen Catanzano, an analyst at Omdia, a division of Informa TechTarget. "The acquisition of GraphAware is very meaningful because it brings a production-ready, government-grade intelligence analysis platform that's already deployed in mission-critical environments, transforming Neo4j from primarily a database technology into a complete intelligence analysis solution," he said. "This adds immediate value." Matt Aslett, an analyst at ISG Software Research, similarly noted that the acquisition adds potentially valuable capabilities. "Neo4j already offers a platform for graph-based data processing and analytics, but with GraphAware Hume, it adds capabilities for collaborative investigation and decision intelligence," he said. Based in San Mateo, Calif., Neo4j recently launched capabilities aimed at <a href="https://www.techtarget.com/searchdatamanagement/news/366623639/Latest-Neo4j-release-aims-to-simplify-graph-technology">simplifying graph technology</a> and <a href="https://www.techtarget.com/searchdatamanagement/news/366630145/Neo4js-latest-targets-graph-database-performance-at-scale">improving the performance</a> of its graph database to handle AI workloads. Founded in 2007, Neo4j's only previous acquisition was its 2023 purchase of Distributed Technology Associates. <section class="section main-article-chapter" data-menu-title="Joining forces"> <h2 class="section-title">Joining forces</h2> Neo4j and GraphAware were closely linked long before the acquisition. The vendors were partners for more than 10 years, and GraphAware relied on <a href="https://www.techtarget.com/searchdatamanagement/news/366593101/Graph-technology-helps-battle-election-misinformation">Neo4j's graph technology</a> as a foundational layer for its AI-powered platform. In addition, organizations such as the U.S. Department of Defense, Internal Revenue Service and European Commission were among numerous joint Neo4j and GraphAware customers before the acquisition. The acquisition of GraphAware to officially join forces was motivated, in part, by Neo4j's desire to add analysis capabilities, according to Sudhir Hasbe, Neo4j's president and chief product officer. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> The acquisition of GraphAware is very meaningful because it brings a production-ready, government-grade intelligence analysis platform that's already deployed in mission-critical environments, transforming Neo4j from primarily a database technology into a complete intelligence analysis solution. </figure> <figcaption> Stephen CatanzanoAnalyst, Omdia </figcaption> </div> </blockquote> "We have been thinking about moving up the stack and providing more verticalized solutions," he said. "GraphAware Hume is 100% built on top of Neo4j, and we have partnered closely for over 10 years, which means we already had government agencies running both together in production. ... The demand was already proven rather than theoretical, and that gave us a lot of confidence in the move." Beyond officially joining forces, which simplifies using Hume in conjunction with Neo4j's platform, Neo4j's acquisition of GraphAware is aimed at adding government agency customers that have <a target="_blank" href="https://resources.data.gov/standards/" rel="noopener">unique data needs</a> served by intelligence analysis capabilities, Hasbe continued. Driven by geopolitical events such as the United States' war with Iran and rapid advances in AI technology, data sovereignty -- the concept that information is subject to the laws of the country in which it was created -- <a href="https://www.techtarget.com/searchenterpriseai/tip/How-to-navigate-data-sovereignty-for-AI-compliance">is a growing concern</a> for many organizations. Government agencies need to control their data in ways that adhere to sovereignty laws while enabling easy access so they can build and deploy AI agents and other insight-generating applications using information from complex data estates. Palantir Gotham, which was originally purpose-built for the U.S. intelligence community, is one platform that enables government agencies to connect and analyze data in real time. Neo4j's acquisition of GraphAware will enable Neo4j to provide a direct alternative. "Advances in AI and growing geopolitical tensions have turned data sovereignty from a nice-to-have into a hard requirement, and government agencies … want to own, manage and control their data, their deployment and their exit path," Hasbe said. "That pushed us to bring these capabilities in-house as a proven, open-standards alternative to Palantir Gotham." Regarding Neo4j's decision to make an acquisition to add new capabilities <a target="_blank" href="https://www.svpg.com/article-build-vs-buy-in-the-age-of-ai/" rel="noopener">rather than build internally</a>, speed and expertise were influential, Hasbe added. "Building intelligence analysis software that government agencies will actually trust takes years and a very specific kind of expertise, from accreditations through to mission-critical deployments, and GraphAware already has all of that in place," he said. Given that Neo4j and GraphAware worked together for more than decade before the acquisition and GraphAware Hume already incorporates Neo4j's technology, the two are a logical fit together, with technological integrations perhaps easier than when the companies that haven't previously partnered, according to Aslett. "The company is a natural fit that enhances Neo4j’s ability to support intelligence applications for use-cases including law enforcement, national security and financial authorities," he said. However, despite Neo4j and GraphAware seemingly being a good fit, all acquisitions have <a target="_blank" href="https://online.hbs.edu/blog/post/mergers-and-acquisitions" rel="noopener">potential risks</a>. With Hume built on Neo4j's graph capabilities, the complex technological integrations that sometimes hinder mergers and acquisitions are unlikely. However, other risks remain, according to Catanzano. "The main concern is whether Neo4j can successfully manage the transition from being a technology platform company to operating a solutions business that requires deep domain expertise, ongoing customer support and navigation of complex government procurement and security requirements," he said. </section> <section class="section main-article-chapter" data-menu-title="Looking ahead"> <h2 class="section-title">Looking ahead</h2> Following its acquisition of GraphAware, Neo4j's product development focus over the coming months will be on making its graph technology part of the AI workflow as a <a href="https://www.techtarget.com/searchenterpriseai/tip/Exploring-the-context-layer-for-AI-systems">knowledge layer for AI systems</a> such as agents, according to Hasbe. In addition, adding agents to its own platform and data sovereignty are focal points, he continued. "We are investing more AI agents as a significant part of the $100 million roadmap we announced last October," Hasbe said. "These autonomous, context-aware agents can turn raw, siloed data into intelligence people can actually act on. ... [Another initiative] is open standards and sovereignty, because that is increasingly what customers are asking for." As Neo4j builds an ecosystem of agents, integrating GraphAware Hume with its agentic AI capabilities would be wise, according to Catanzano. In addition, he noted that Neo4j's acquisition of GraphAware could serve as a starting point for adding purpose-built capabilities not only for government agencies, but for industries such as financial crime, supply chain intelligence and <a href="https://www.techtarget.com/revcyclemanagement/news/366642704/DOJ-forms-West-Coast-Strike-Force-to-stop-healthcare-fraud">healthcare fraud detection</a> as well.  "They could also expand into adjacent markets where similar investigative and relationship analysis capabilities are needed, while continuing to emphasize their open standards approach as a key differentiator that prevents vendor lock-in and enables true data sovereignty," Catanzano said. Aslett suggested that as Neo4j enters new markets and competitive situations following the acquisition of GraphAware, it should not only keep GraphAware's talent -- including founder and CEO <a href="https://podcasts.apple.com/us/podcast/michal-bachman-ceo-of-graphaware/id1531899005?i=1000733889601&l=zh-Hans-CN">Michal Bachman</a> -- but add to it to make Hume more visible. "It is clearly positioning itself to compete with Palantir’s Gotham, so [Neo4j] will need to ensure it retains and enhances GraphAware's technical and sales expertise to build on its existing success," he said. Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> The purchase adds analysis capabilities for government agencies that work on top of the vendor's graph database, expanding its target audience to include analysts. https://cdn.ttgtmedia.com/rms/onlineimages/collab_a275903017.jpg https://www.techtarget.com/searchdatamanagement/news/366643806/Neo4js-GraphAware-acquisition-targets-new-customer-segment Thu, 04 Jun 2026 15:08:00 GMT Neo4j's GraphAware acquisition targets new customer segment Microsoft is making Fabric a foundation for agentic AI. First <a href="https://www.techtarget.com/searchbusinessanalytics/news/366559533/Microsoft-launches-Fabric-adds-Copilot-for-the-new-platform">launched in 2023</a>, Fabric brought together seven previously disparate Microsoft data management and analytics capabilities, such as Data Factory and Power BI, and infused them with generative AI through early iterations of its Copilots. As enterprises have <a target="_blank" href="https://kpmg.com/us/en/media/news/q1-ai-pulse2026.html" rel="noopener">increased their investments</a> in developing agents and other AI applications, additions such as OneLake to unify data and <a href="https://www.techtarget.com/searchdatamanagement/feature/One-year-of-MCP-Support-a-must-for-data-management-vendors">Model Context Protocol servers</a> to connect agents with data sources marked Fabric's evolution toward becoming a base for developing agents that can understand the unique characteristics of an individual business. Microsoft's latest Fabric capabilities, unveiled on Tuesday during its Build user conference in San Francisco, are designed to further aid developers attempting to create AI tools that can be trusted to perform properly in production. Among them are a tool that unifies business logic to form a contextual foundation for agents and a database purpose-built for the scale of AI workloads. Collectively, the new Fabric features are valuable to Microsoft users because they directly address <a target="_blank" href="https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf" rel="noopener">problems enterprises face</a> when attempting to move agents beyond pilots and into production, according to William McKnight, president of McKnight Consulting. "Microsoft is introducing a suite of features designed to eliminate the context bottleneck by providing AI agents with a persistent, shared understanding of business data," he said. "Key updates … merge application backends, high-speed processing and semantic context into a platform capable of deploying autonomous, enterprise-scale AI agents -- the goal of many organizations today." Mike Leone, an analyst at Moor Insights & Strategy, similarly noted the significance of Microsoft's new Fabric features, particularly those that the tech giant terms ontology capabilities that help agents understand what an enterprise's data means in <a href="https://www.techtarget.com/searchenterpriseai/tip/Exploring-the-context-layer-for-AI-systems">the context of its business</a>. "The real story is that Microsoft is closing the distance between where your data lives and where agents actually act on it," he said. "What stands out is that you can … give agents a clear map of what that data means in your business and then let them act on it directly. That round trip, from raw data to an app or agent that does the work, used to take stitching three or four separate systems together." <section class="section main-article-chapter" data-menu-title="Grounding for AI"> <h2 class="section-title">Grounding for AI</h2> Many enterprises are making a push to move past experiments with AI to put agents into production. However, disorganized data that makes it difficult to discover and operationalize the contextually relevant data agents require to deliver accurate outputs remains an obstacle for many. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> Microsoft is introducing a suite of features designed to eliminate the context bottleneck by providing AI agents with a persistent, shared understanding of business data. </figure> <figcaption> William McKnightPresident, McKnight Consulting </figcaption> </div> </blockquote> Like numerous other data management and analytics providers -- <a href="https://www.techtarget.com/searchdatamanagement/news/366637142/New-Databricks-tool-aims-to-up-agentic-AI-response-accuracy">Databricks</a>, <a href="https://www.techtarget.com/searchdatamanagement/news/366641929/Google-unveils-data-cloud-purpose-built-for-agentic-AI">Google Cloud</a> and <a href="https://www.techtarget.com/searchdatamanagement/news/366643795/Snowflake-barrage-adds-more-AI-development-analysis-tools">Snowflake</a> among them -- Microsoft is now making context for AI a focal point of its product development plans for Fabric. "Context matters because as models become more capable and more available, the differentiator isn't just access to intelligence, it's the ownership," Kyle Daigle, Microsoft's developer chief marketing officer, said during a virtual press briefing before Build. "The real question every organization is asking is how to use your expertise, your data, and your way of working." New features unveiled during Build include the following: <ul type="disc" class="default-list"> <li>Microsoft IQ, an enterprise intelligence layer for AI that unifies an organization's data estate and joins it with <a href="https://www.techtarget.com/searchdatamanagement/opinion/Why-data-semantics-matters-for-context-aware-systems">semantic meaning</a> and <a href="https://www.techtarget.com/whatis/definition/business-logic">business logic</a> to empower agents.</li> <li>Fabric IQ, a feature within Microsoft IQ that grounds AI with consistent definitions, metrics and relationships.</li> <li>Rayfin, a backend-as-a-service feature for application development that runs on top of Fabric.</li> <li>New shortcuts in OneLake that make it easier to connect data across platforms so users don't have to move or duplicate data.</li> <li>GPU-accelerated workflows in Fabric Data Warehouse to improve query performance.</li> <li>A database hub in Fabric where customers can centrally manage their Microsoft databases.</li> <li>Azure HorizonDB, a new <a href="https://www.theserverside.com/tip/MySQL-vs-PostgreSQL-Compare-popular-open-source-databases">PostgreSQL database</a> in public preview that improves on the performance and scalability of Azure Database for PostgreSQL to better handle AI workloads.</li> <li>New security capabilities in preview for existing workloads in Azure Database for PostgreSQL.</li> <li>The general availability of Azure Cosmos DB Linux Emulator, a tool that enables users of Microsoft's NoSQL vector database to locally build, test and validate applications across Linux, macOS and Windows without having to do their work in a cloud environment.</li> </ul> Capabilities in Fabric IQ that allow customers to define their data once and have that definition used by every agent provided by Microsoft are among the most valuable, according to Leone. "Instead of re-teaching each new agent what a customer or an order is in your business, you set it once and they all inherit it, and that kind of reuse is hard to pull off unless you own both the data layer and the agent tooling, which few players do," he said. As Microsoft adds capabilities to Fabric, <a href="https://www.techtarget.com/searchbusinessanalytics/news/366616023/With-new-Fabric-features-Microsoft-aims-at-AI-development">the platform is evolving</a> to become a well-designed bridge for moving AI experiments into production, Leone continued. However, he noted that better data governance guardrails, an orchestration framework for multi-agent networks and <a href="https://www.techtarget.com/searchdatamanagement/opinion/Why-agentic-AI-demands-both-structured-and-unstructured-data">operationalization of unstructured data</a> could all improve Fabric. Meanwhile, from a competitive standpoint, Microsoft differentiates itself with the breadth of its data and AI capabilities, though individual tools are perhaps not as deep as those provided by more specialized vendors, according to Leone. "I'd put Microsoft right at the front of the pack on the unified data foundation idea, and the differentiation is breadth," he said. "Pulling analytics, transactional databases, a semantic layer, and now app development into one platform is a more integrated bet than most of the specialized data platforms are making, since those players tend to go deeper in their lane while Microsoft goes wider." Like Leone, McKnight called <a href="https://www.techtarget.com/searchenterpriseai/post/AI-agents-are-only-as-smart-as-the-data-that-feeds-them">Fabric IQ one of the most significant new features</a>. In addition, he noted the value of Rayfin. "Fabric IQ and Rayfin serve as the core pillars for building enterprise-grade AI, respectively solving the critical challenges of data context and deployment speed," McKnight said. "Fabric IQ eliminates the context bottleneck [and] Rayfin then operationalizes this intelligence. … Together, they allow developers to transition their experiments to production-ready multi-agent systems." Comparatively, Microsoft's data and AI capabilities are in line with those of its <a href="https://www.techtarget.com/searchcloudcomputing/definition/hyperscale-cloud">hyperscale cloud</a> competitors, he continued. "Microsoft's AI data strategy -- Fabric, OneLake, Purview and integrated vector stores -- positions it securely alongside the hyperscaler cohort, excelling in platform integration while playing catch-up in technical depth compared to specialized vendors. It excels in bundling, operational simplicity, and ecosystem coherence rather than inventing net-new data primitives." </section> <section class="section main-article-chapter" data-menu-title="Looking ahead"> <h2 class="section-title">Looking ahead</h2> As Microsoft continues to build up Fabric as a foundational layer within its data and AI platform, there remains room for improvement, according to McKnight. While its overall capabilities are competitive, specific areas such as governance, <a href="https://www.techtarget.com/searchapparchitecture/definition/interoperability">interoperability</a> with third-party platforms, workflow depth and cost transparency all need addressing. Specifically, McKnight suggested that Microsoft make Purview -- an integrated security, governance and compliance service -- more AI-ready, embrace open table formats and add built-in <a href="https://www.techtarget.com/searchenterpriseai/definition/large-language-model-operations-LLMOps">LLMOps</a> and agentic orchestration capabilities. "This would help position Fabric as the definitive, safe choice for production-grade AI," he said. Leone, meanwhile, advised Microsoft to make it easier for new customers to get started with Fabric so it evolves from a default platform for existing customers to a destination for new ones. "First, make it dead simple to start small, because most new customers aren't moving their whole data estate on day one," he said. "Let them adopt one workload, … and expand from there instead of feeling like they have to buy into the entire platform up front. Second, the faster and lower-risk it makes moving, the more Fabric turns from a Microsoft-shop default into a real destination for brand-new customers." Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> New features that feed agents contextually relevant data add breadth to the platform and keep its data and AI capabilities current in a competitive market. https://cdn.ttgtmedia.com/rms/onlineimages/telecommunications_g1189468316.jpg https://www.techtarget.com/searchdatamanagement/news/366643955/Microsoft-boosts-Fabric-to-make-it-a-foundation-for-AI Tue, 02 Jun 2026 15:07:00 GMT Microsoft boosts Fabric to make it a foundation for AI Snowflake on Tuesday unveiled an avalanche of new features aimed at helping customers build AI tools that make employees better informed and more efficient. Among others, they include a fully managed <a href="https://www.techtarget.com/searchdatamanagement/opinion/Real-time-data-streaming-for-AI-invest-where-it-matters">streaming data service</a> in Snowflake CoCo (formerly Cortex Code), which is the vendor's coding agent for developing workflows and applications, and personalization capabilities in Snowflake CoWork (formerly Snowflake Intelligence), which is a personal agent that assists users as they analyze data and build data workflows. In addition, Snowflake introduced new tools in Horizon Catalog, a data catalog that enables users to govern and discover data, aimed at securing and governing agents and standardizing <a href="https://www.techtarget.com/searcherp/feature/Why-context-engineering-is-the-next-enterprise-software-priority">the context agents call upon</a> to carry out tasks. The new features were revealed during Snowflake Summit, the vendor's user conference in San Francisco. Michael Ni, an analyst at Constellation Research, noted that by unifying capabilities in CoCo, CoWork and Horizon Catalog, Snowflake is demonstrating its evolution toward becoming a platform for agentic AI. As a result, its additions are significant. "Snowflake's release looks less like a product launch cycle and more like platform maturation," Ni said. "There are plenty of new features, but the real significance lies in Snowflake's … bigger strategic ambition as it shifts from being the Data Cloud, where the story was 'bring AI to your data', to the 'Agentic AI Platform' with the story of using trusted context to govern AI actions across the enterprise." Sanjeev Mohan, founder and principal of analyst firm SanjMo, likewise noted that Snowflake's new features collectively comprise a significant update with a new feature called Cortex Training, which allows users to customize <a href="https://www.techtarget.com/whatis/feature/Foundation-models-explained-Everything-you-need-to-know">foundation models</a>, showing the vendor's growth. "It's significant due to the breadth," he said. "Many vendors ship features in one or two layers of the stack. Snowflake shipped simultaneously across infrastructure, metadata and semantics, security and AI surfaces for both developers and knowledge workers. The Cortex Training announcement … could be a net-new revenue category. Thus far, training or fine-tuning small models has not become mainstream." Based in Bozeman, Mont., but with a campus in Menlo Park, Calif., Snowflake's data platform and AI development capabilities are designed to enable users to build AI and analytics tools on <a href="https://www.techtarget.com/searchbusinessanalytics/news/366618249/Trusted-data-at-the-core-of-successful-GenAI-adoption">a trusted data foundation</a>. Beyond introducing new features, Snowflake on May 27 expanded its partnership with AWS, signing a collaboration agreement to invest $6 billion in helping joint customers build and deploy AI. <section class="section main-article-chapter" data-menu-title="Empowering enterprises with AI"> <h2 class="section-title">Empowering enterprises with AI</h2> Snowflake <a href="https://www.techtarget.com/searchbusinessanalytics/news/366573245/Snowflake-boosting-its-commitment-to-AI-including-GenAI">was slow</a> to add AI development and management capabilities after OpenAI's November 2022 launch of ChatGPT sparked surging interest in AI development that <a target="_blank" href="https://www.gartner.com/en/newsroom/press-releases/2026-1-15-gartner-says-worldwide-ai-spending-will-total-2-point-5-trillion-dollars-in-2026" rel="noopener">continues to increase</a>. Rival Databricks and hyperscale cloud vendors AWS, Google Cloud and Microsoft all quickly added integrations with large language models such as ChatGPT -- some even developing their own -- and created development frameworks designed to simplify building AI tools. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> It's significant due to the breadth. Many vendors ship features in one or two layers of the stack. Snowflake shipped simultaneously across infrastructure, metadata and semantics, security and AI surfaces for both developers and knowledge workers. </figure> <figcaption> Sanjeev MohanFounder and principal, SanjMo </figcaption> </div> </blockquote> Following <a href="https://www.techtarget.com/searchbusinessanalytics/news/366571855/Snowflake-CEO-Slootman-steps-down-Ramaswamy-takes-over">a CEO change</a> in February 2024, Snowflake similarly embraced AI as a core part of its platform and continues to add AI capabilities that simplify using its tools as well as features that enable customers to create their own AI applications. "Our whole mission is based on the premise that we are the platform that will help organizations make every team member be more productive … through the benefits of AI, and do so being able to sleep well at night because of security, compliance and governance," Christian Kleinerman, Snowflake's executive vice president of product, said during a virtual press conference on May 26. Many of the new capabilities Snowflake revealed on Tuesday are tied to that aim. Snowflake CoCo is the interface for developers to build the AI and analytics workflows that enable business users to be more productive. New CoCo features include Datastream to bring <a href="https://www.techtarget.com/searchdatamanagement/news/252512512/Apache-Kafka-31-opens-up-data-streaming-for-analytics">real-time Apache Kafka data</a> into AI applications to keep them current and accurate, desktop and mobile versions that enable developers to work in preferred environments, Automations to autonomously execute recurring workflows, and prebuilt Skills that simplify engineering tasks. Snowflake CoWork is the AI-powered interface that enables business users to be more productive. New CoWork capabilities include User Skills to personalize insights and actions based on an employee's role, Deep Research to enable in-depth analysis across both <a href="https://www.techtarget.com/searchdatamanagement/opinion/Why-agentic-AI-demands-both-structured-and-unstructured-data">structured and unstructured data</a> and Cortex Sense to join data with business definitions and operational knowledge to provide agents with better context. Horizon Catalog is the hub that connects and governs an enterprise's Snowflake estate, enabling development and analysis. New Horizon Catalog tools include Horizon Context to provide context layer that ensures AI-driven outcomes are reliable, Semantic Studio and Semantic View Autopilot <a href="https://www.techtarget.com/searchdatamanagement/opinion/Why-data-semantics-matters-for-context-aware-systems">to build semantic views</a>, Agent Identity to give each agent a verified identity before it can access data or take action, and adaptive compute to automatically optimize compute and software resources. "Horizon Context along with Cortex Sense are probably the most valuable [new capabilities]," Mohan said. "AI agents are only as reliable as the definitions they reason from. … Snowflake's earlier Semantic Studio and Semantic View Autopilot along with the [Open Semantic Interchange] standard solved this at the platform level. Now they are taking it to a higher level -- context." Finally, to improve access to the often distributed data that informs agents and other AI tools, Snowflake is improving the interoperability of its platform with capabilities such as support for <a href="https://iceberg.apache.org/spec/">Apache Iceberg v3</a>, zero copy integrations with data sources including SAP and Salesforce, centralized governance across systems through Apache Polaris within Horizon Catalog, and Open Data Sharing to enable organizations to securely share data and AI assets with customers and partners. While each of the individual features address customers' evolving needs, the most important additions are the tools that deliver trusted, <a href="https://www.techtarget.com/searchbusinessanalytics/feature/Talend-CEO-discusses-importance-of-mining-relevant-data">relevant data</a> to agents, according to Ni. "The most valuable thing Snowflake announced wasn't another agent," he said. "It was the shared understanding that those agents operate from. Snowflake recognizes that when intelligence becomes cheap with the new LLMs, the hardest problem in enterprise AI is ensuring multiple humans, BI tools, applications, and agents operate from the same business truth." </section> <section class="section main-article-chapter" data-menu-title="Competitive standing"> <h2 class="section-title">Competitive standing</h2> Although Snowflake was once slow to react to surging interest in AI development, the vendor is now one of many data management vendors <a href="https://www.techtarget.com/searchdatamanagement/opinion/The-race-to-build-the-ultimate-data-platform">in a race</a> to provide the tools customers need to develop and manage agentic AI systems on a foundation of governed data, according to Ni. Amid that race, however, Snowflake is carving out its own niche rather than directly competing with <a href="https://www.techtarget.com/searchdatamanagement/news/366638723/Databricks-launches-PostgreSQL-Lakebase-to-aid-AI-developers">rival Databricks</a>, he continued. "While data and AI platform vendors like Databricks focus on helping developers build agents, Snowflake focuses on making agents simple to trust and scale," Ni said. "At the same time, the market is moving from agent creation to agent governance, and that's where Snowflake is making its biggest bet." Mohan similarly noted that Snowflake is taking a different approach than Databricks, which has historically catered more to <a href="https://www.techtarget.com/searchenterpriseai/feature/AI-engineer-vs-data-scientist-Whats-the-difference">data scientists and engineers</a> than Snowflake, which focuses on business users. "Databricks' strengths are in the machine learning and data engineering workflow," he said. "Snowflake's counter-play is governance-first agent infrastructure. For teams building ML-heavy systems, Databricks is still stronger. For enterprises that need AI with audit trails and consistent business semantics, and want it to work for business users, Snowflake's Summit announcements make a strong case." Regarding what more Snowflake can add to continue serving its customers as they attempt to modernize with agents, Mohan suggested that the vendor add tools that <a href="https://www.computerweekly.com/feature/Why-AI-is-forcing-enterprises-to-rethink-observability">oversee how agents behave</a> in production so that customers don't have to seek out such capabilities from competitors. "Snowflake now has CoCo for developers and CoWork for knowledge workers, but … the hardest problems are operational, like monitoring, debugging, regression testing for agent behavior," he said. "Snowflake should build native tooling here before customers are forced to stitch together third-party solutions." Ni, meanwhile, advised Snowflake to continue adding and refining features such as Horizon Context that <a href="https://www.techtarget.com/searchenterpriseai/tip/Exploring-the-context-layer-for-AI-systems">help agents understand</a> an enterprise's operations so they can perform as intended. "Horizon Context is important to helping AI understand what the business means," he said. "The next frontier is helping AI understand how the business operates. Most enterprise decisions are not driven by data alone, but by the combination of business context, process state and operational constraints." Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> A streaming data service and tools that provide agents with contextual awareness highlight the latest from the vendor as it constructs a foundation for agentic enterprises. https://cdn.ttgtmedia.com/rms/onlineimages/code_g175422126.jpg https://www.techtarget.com/searchdatamanagement/news/366643795/Snowflake-barrage-adds-more-AI-development-analysis-tools Tue, 02 Jun 2026 09:00:00 GMT Snowflake barrage adds more AI development, analysis tools Fivetran and DBT Labs are one. The merger of the two, <a href="https://www.techtarget.com/searchdatamanagement/news/366632699/Fivetran-DBT-Labs-merge-to-add-complementary-capabilities">first revealed in October 2025</a>, became official on Monday, creating a new company that combines the data integration capabilities of Fivetran with the data transformation and data modeling capabilities of DBT Labs. Although financial terms of the all-stock transaction were not disclosed, Fivetran was valued at $5.6 billion in September 2021 when it raised $565 million in venture capital funding, while DBT Labs was valued at $4.2 billion in February 2022 when it raised $222 million from venture capitalists. The combined entity will operate as Fivetran + DBT Labs with former Fivetran CEO <a href="https://www.techtarget.com/searchdatamanagement/news/366538572/Fivetrans-new-funding-a-hedge-against-economic-uncertainty">George Fraser</a> serving as the new company's CEO and former DBT Labs CEO <a href="https://www.techtarget.com/searchdatamanagement/news/252513954/Dbt-Labs-raises-expectations-for-data-transformation">Tristan Handy</a> serving as president. Given that Fivetran and DBT Labs bring separate but complementary capabilities to Fivetran + DBT Labs, their combination is logical, according to Devin Pratt, an analyst at IDC. "Fivetran moves the data and DBT makes it trustworthy," he said. "Together they cover the two things that buyers care about most right now, [which are] data quality and AI readiness." However, whether existing customers -- particularly those of <a href="https://www.techtarget.com/searchdatamanagement/news/366632586/DBT-Labs-targets-costs-with-new-data-engine-adds-AI-agents">DBT Labs</a>, which began as an open source project -- stay loyal as the new company evolves remains to be seen, Pratt continued. "The technology fit is the easy part," he said. "The real test is keeping the DBT open-source community's trust through the transition." Donald Farmer, founder and principal of TreeHive Strategy, similarly noted that Fivetran and DBT Labs are a strong technological fit. But given Fivetran's history as a closed-source <a href="https://www.techtarget.com/searchcloudcomputing/definition/Software-as-a-Service">SaaS</a> vendor "with a reputation for complex, aggressive consumption-based pricing models" and DBT Labs' open-source ethos, there could be culture clashes as the two join forces, he cautioned. "If they integrate well and operate as a single platform they can eliminate some of the complexity of the data stack," Farmer said. "And they do share many customers already. But they may be less compatible in business terms. ... Bringing these two communities together is going to be a real challenge." Ultimately, the motivation behind the merger might be an initial public stock offering, he added. "Perhaps the real driver is that neither company was likely to successfully IPO individually [and] this merger consolidates their annual recurring revenue to cross a threshold required for a successful public listing," Farmer said. <section class="section main-article-chapter" data-menu-title="Technological fit"> <h2 class="section-title">Technological fit</h2> While Fivetran and DBT Labs each developed user bases as independent vendors, as <a target="_blank" href="https://kpmg.com/us/en/media/news/q1-ai-pulse2026.html" rel="noopener">agentic AI becomes more prevalent</a> across enterprises and data management evolves to become a foundational layer for multi-agent systems, some vendors are turning their platforms into end-to-end systems for data and AI. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> The technology fit is the easy part. The real test is keeping the DBT open-source community's trust through the transition. </figure> <figcaption> Devin PrattAnalyst, IDC </figcaption> </div> </blockquote> Those that have the capital to compete -- hyperscale cloud vendors such as AWS, Google Cloud and Microsoft along with data platform providers including Databricks and Snowflake -- are expanding. As they do so, it makes it difficult for niche vendors to remain independent, which is leading to consolidation. Some formerly independent companies have opted to sell to broader platform vendors. For example, Informatica is now <a href="https://www.techtarget.com/searchcustomerexperience/news/366624960/Salesforce-to-acquire-Informatica-in-8-billion-deal">part of Salesforce</a>, Confluent was <a href="https://www.techtarget.com/searchdatamanagement/news/366636098/IBM-acquiring-Confluent-to-boost-AI-development-capabilities">bought by IBM</a>, and Dremio was <a href="https://www.techtarget.com/searchdatamanagement/news/366642794/SAP-acquisitions-of-Dremio-Prior-Labs-target-AI-development">acquired by SAP</a>. Rather than find buyers to become small pieces of larger wholes, Fivetran and DBT Labs elected to merge to expand beyond their specialties. Together, they can provide a data infrastructure layer designed to prepare data for AI, including the semantic modeling capabilities and business logic that help feed agents the contextually relevant data they require to deliver accurate, trustworthy outcomes. In addition, with DBT Labs' origins in the open-source community, the combined Fivetran + DBT Labs platform includes open standards that work across all clouds, engines and tools so that customers can use the data management architecture of their choice and avoid becoming too closely aligned with any single vendor. Pratt noted that IDC research shows that 97% of organizations want to reduce the number of products they use for data management. However, <a href="https://www.techtarget.com/searchenterpriseai/tip/Best-practices-to-avoid-AI-vendor-lock-in">only 12% want to use a single vendor</a>. Therefore, though the space for specialists is shrinking, there remains room for independent vendors such as Fivetran + DBT Labs that provide more than one niche capability but aren't end-to-end data and AI platforms. "Specialists can still thrive, as long as they slot cleanly into a core or get big enough to be that core," Pratt said. "Combining, as Fivetran and DBT have, is one way to do that rather than waiting to be bought." Farmer likewise noted that despite ongoing consolidation, there remains a place for independent vendors. In particular, independent vendors with unique engineering approaches can survive given that integration becomes the focus amid acquisitions rather than innovation. "Independents do have opportunities, especially if they can support a methodology and community," Farmer said. "When independents -- like Confluent, Dremio or DBT -- get absorbed, their engineering resources are inevitably redirected from product innovation toward integration and from the interests of their community towards alignment with enterprise sales." </section> <section class="section main-article-chapter" data-menu-title="New capabilities"> <h2 class="section-title">New capabilities</h2> Beyond the merger, Fivetran + DBT Labs unveiled its first new features. They include the following: <ul type="disc" class="default-list"> <li>Agents Schema, an open source standard for <a href="https://www.techtarget.com/searcherp/feature/Why-context-engineering-is-the-next-enterprise-software-priority">providing context to agents</a> that designates one <a href="https://www.techtarget.com/searchdatamanagement/definition/schema">schema</a> in a data warehouse or data lake that is compatible across systems as the shared context layer for agentic AI.</li> <li>DBT Core 2.0, the latest version of DBT Labs' open source Fusion engine for data transformation using <a href="https://www.techtarget.com/searchdatamanagement/definition/SQL">SQL</a> and <a href="https://www.techtarget.com/whatis/definition/Python">Python</a> code.</li> <li>DBT State, a caching layer for data pipelines aimed at enabling users to reduce infrastructure costs.</li> <li>DBT Wizard, an autonomous assistant that uses context including lineage and defined metrics from DBT projects for model authoring, refactoring and debugging.</li> </ul> DBT Core 2.0, DBT State and DBT Wizard are in various states of testing and not yet generally available. "I like DBT State," Farmer said. "In an era where CFOs are cracking down on unpredictable bills, if DBT can cut infrastructure costs … with smart caching, that's a solid -- and testable -- claim." Pratt, meanwhile, noted the potential value of Agents Schema. "The hardest part of getting agents into production isn't the model, it's giving the agent context it can trust," he said. "Agents Schema goes straight at that, as an open standard the customer owns rather than one more lock-in." Looking ahead, now that Fivetran and DBT Labs have merged, Pratt recommended that the company continue to stress and honor the openness on which DBT Labs was founded. With some vendors making it difficult to integrate with third parties, and others enabling only some <a href="https://www.techtarget.com/searchbusinessanalytics/feature/More-reasoning-interoperability-key-to-future-of-agentic-AI">interoperability</a>, Fivetran + DBT Labs could stand apart from at least some competitors by fully embracing openness. "Staying open is their biggest asset," Pratt said. "The opportunity now is to pair that openness with strong governance and automation, the things buyers value most, which would position them to keep their users and attract new ones." Toward that end, Farmer suggested that Fivetran + DBT make <a href="https://www.techtarget.com/searchbusinessanalytics/news/366556394/DBT-Labs-updates-Semantic-Layer-adds-data-mesh-enablement">DBT's semantic modeling capabilities</a> open source rather than keep it a paid feature. "They need to fully open up DBT's semantic layer," he said. "A semantic layer must integrate with outside tools to be useful." Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> The combined capabilities of the newly formed company provide the infrastructure, including data integration and preparation, for agents and other cutting-edge applications. https://cdn.ttgtmedia.com/rms/onlineimages/collab_a346805525.jpg https://www.techtarget.com/searchdatamanagement/news/366643590/Fivetran-DBT-Labs-complete-merger-to-form-data-layer-for-AI Mon, 01 Jun 2026 08:57:00 GMT Fivetran, DBT Labs complete merger to form data layer for AI With the launch of Aida and the introduction of the Starburst Enterprise Intelligence Platform, Starburst is extending AI-fueled query and analysis capabilities beyond its data management environment and into the workflows and applications where business users do their jobs. First <a href="https://www.techtarget.com/searchdatamanagement/news/366641734/Starburst-intros-AI-assistant-to-boost-analysis-exploration">introduced in April</a>, Aida -- which stands for AI data assistant -- is an AI-powered assistant that lets users explore and analyze data using natural language and is the main interface for Starburst's Enterprise Intelligence Platform. However, unlike assistants featuring basic text-to-SQL translation capabilities that enable only rudimentary questions and answers, Aida is built on a framework that allows it to access an organization's <a href="https://www.techtarget.com/searchbusinessanalytics/opinion/The-importance-of-data-products">data products</a>, such as governed datasets that include semantically related tables, so it can reason, act and observe based on <a href="https://www.techtarget.com/searchdatamanagement/opinion/Why-data-semantics-matters-for-context-aware-systems">proper context</a>. In addition, rather than accessing only centralized data or data duplicated and imported into Starburst, Aida works across distributed data environments such as multiple clouds, different data lakes and warehouses, and disparate systems and applications. By working across distributed data estates, Aida enables organizations to save the time it takes to extract, transform and load data, and avoid the cost and risk of exposure associated with moving data. Additionally, it allows organizations to keep their existing <a href="https://www.techtarget.com/searchdatamanagement/tip/Assemble-the-layers-of-big-data-stack-architecture">data infrastructures</a> rather than build new ones for AI. As a result, the Starburst Intelligence Platform is a valuable addition for the vendor's users, according to Stephen Catanzano, an analyst at Omdia, a division of Informa TechTarget. "The Starburst Enterprise Intelligence Platform represents a significant addition for existing users because it transforms their data infrastructure from a query engine into a comprehensive AI enablement layer without requiring them to rethink their architecture or move their data," Catanzano said. "Users can now … effectively shorten the path from data access to AI-driven business value." Based in Boston, Starburst provides a <a href="https://www.techtarget.com/searchdatamanagement/news/366545117/Lakehouse-architecture-the-best-fit-for-modern-data-needs">data lakehouse platform</a> geared toward connecting data generated and stored in otherwise disparate systems. The new features were unveiled during AI & Datanova, Starburst's user conference in Miami. <section class="section main-article-chapter" data-menu-title="Enabling AI"> <h2 class="section-title">Enabling AI</h2> Difficulty accessing relevant data, along with other data issues such as poor quality, are among the problems <a target="_blank" href="https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf" rel="noopener">preventing many enterprises</a> from building agents and other AI tools trustworthy enough to move into production. As a result, data management and analytics vendors such as <a href="https://www.techtarget.com/searchdatamanagement/news/366637142/New-Databricks-tool-aims-to-up-agentic-AI-response-accuracy">Databricks</a> and <a href="https://www.techtarget.com/searchbusinessanalytics/news/366642778/Tableau-repositions-for-AI-unveils-new-knowledge-layer">Tableau</a> have introduced new capabilities over the past few months designed to help customers better access the contextually relevant data that AI tools need to deliver trusted outcomes. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> The Starburst Enterprise Intelligence Platform represents a significant addition for existing users because it transforms their data infrastructure from a query engine into a comprehensive AI enablement layer without requiring them to rethink their architecture or move their data. </figure> <figcaption> Stephen CatanzanoAnalyst, Omdia </figcaption> </div> </blockquote> Now, Starburst is doing the same, but with its own approach. Rather than import data to AI, the Starburst Enterprise Intelligence Platform connects and governs distributed data so AI can run directly on that data, underpinned by the context provided by <a href="https://www.techtarget.com/searchbusinessanalytics/feature/Treating-data-as-a-product-a-method-to-grow-analytics-use">trusted data products</a>. A combination of customer feedback and market trends pushed Starburst to build the new platform, according to Jitender Aswani, the vendor's senior vice president and global head of engineering and security. Customers were attempting to build agents and other AI tools on top of Starburst's query engine but struggling with data that wasn't AI-ready, he noted. Simultaneously, Starburst recognized a similar problem in the broader market with agents failing because of problems related to their underlying data. "These two signals confirmed that our investment should focus on bridging that gap," Aswani said. "The Enterprise Intelligence Platform is … a natural evolution of Starburst's foundation in federated, governed data." Catanzano, meanwhile, noted that Starburst's extension of AI to data distributed to AI-powered analysis could distinguish the vendor from competitors such as Databricks, Dremio -- which is <a href="https://www.techtarget.com/searchdatamanagement/news/366642794/SAP-acquisitions-of-Dremio-Prior-Labs-target-AI-development">being acquired by SAP</a> -- and Snowflake. "Starburst's approach is differentiated in its commitment to true in-place data processing without requiring consolidation or re-platforming, combined with its focus on bringing AI directly into existing workflows," he said. However, whether Starburst's unique approach proves more effective remains to be seen, Catanzano continued. "The ultimate differentiation will depend on execution quality and how well these capabilities perform at enterprise scale compared to competitors," he said. Like Catanzano, Kevin Petrie, an analyst at BARC U.S., noted that <a href="https://www.techtarget.com/searchdatamanagement/feature/Data-domain-ownership-data-mesh-chart-path-to-AI-ready-data">a decentralized approach</a> to AI-powered analysis has advantages. Data consolidation is a struggle -- and ultimately a mirage -- for many companies, according to Petrie. Although they migrate and centralize certain data on one platform for an initiative, inevitably other units within the business migrate and centralize certain data on another platform for a different initiative. Meanwhile, still other data needs to remain in place due to migration complexity or <a href="https://www.techtarget.com/searchenterpriseai/tip/How-to-navigate-data-sovereignty-for-AI-compliance">data sovereignty concerns</a>. "All this means that the data inputs needed for AI projects -- especially unstructured objects like documents or emails -- often reside in multiple locations," Petrie said. Starburst's Enterprise Intelligence Platform, therefore, could be an important new option for many enterprises, he continued, adding that BARC research shows that companies with data products as part of <a href="https://www.techtarget.com/searcherp/feature/4-structural-foundations-of-reliable-enterprise-AI">a foundation for AI</a> are three times more likely to put agents in production than those that don't use data products for AI initiatives. "AI adopters need governed, enterprise-wide programs in order to get agentic analytics safely into production," Petrie said. "These new capabilities from Starburst help organizations meet those requirements by federating metadata, enforcing consistent rules, and aligning outputs on standard business definitions. Packaging these features with data products is critical." Beyond the Starburst Enterprise Intelligence Platform, Starburst unveiled the following during its user conference: <ul type="disc" class="default-list"> <li>An engine that improves the performance of the open source <a href="https://www.techtarget.com/searchdatamanagement/feature/A-look-at-Presto-Trino-SQL-query-engines">Trino query engine</a> that powers Starburst Enterprise and Starburst Galaxy.</li> <li>Support for Model Context Protocol (MCP) to connect Aida with external tools and third-party systems.</li> <li>Resilience capabilities that keep AI systems, including agents, operating without disruption when infrastructures fail.</li> <li>Icehouse Ingest -- data loading capabilities that combine the Trino engine with <a target="_blank" href="https://iceberg.apache.org/" rel="noopener">Apache Iceberg</a> tables -- to load batch and streaming files</li> <li>Icehouse LakeOps to automatically optimize Iceberg tables, observe the health of data tables and tune queries.</li> <li>A BYOC deployment option.</li> </ul> Collectively, Starburst's new features, including those that comprise its Enterprise Intelligence Platform, are seemingly aligned to enable running AI on distributed data, according to Catanzano. However, tools that add <a href="https://www.computerweekly.com/feature/Why-AI-is-forcing-enterprises-to-rethink-observability">model observability</a> and explainability capabilities would improve the platform, he continued. "Explicit capabilities around AI model observability and explainability help users understand not just what their AI agents are doing but why specific recommendations or actions were generated, which becomes increasingly critical as these systems move to autonomous decision-making in production environments," Catanzano said. </section> <section class="section main-article-chapter" data-menu-title="Next steps"> <h2 class="section-title">Next steps</h2> As Starburst plans product development, focal points include building agents that don't merely respond to questions, but instead proactively surface insights, schedule data workflows and flag data anomalies, according to Aswani. In addition, adding depth to its ecosystem by connecting to more sources <a href="https://www.techtarget.com/searchdatamanagement/feature/One-year-of-MCP-Support-a-must-for-data-management-vendors">through MCP</a> and federating context and governance across distributed data estates are priorities, he continued. "The goal across all three is the same: helping customers move from AI pilots to production without forcing them to move their data, rebuild their stack or compromise on governance," Aswani said. Catanzano, meanwhile, suggested that Starburst could continue serving the needs of users by adding industry-specific capabilities designed to speed and simplify building and deploying agents. In addition, he noted that a broader partnership ecosystem and <a href="https://www.techtarget.com/searchcloudcomputing/tip/What-is-cloud-cost-optimization-Best-practices-to-embrace">cost control capabilities</a> would be beneficial. "Enhanced cost optimization tools that provide visibility into the economics of distributed AI workloads would address margin pressure concerns and position Starburst as not just an enabler of AI but a protector of AI ROI," Catanzano said. Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> The vendor's approach to developing agents and other cutting-edge applications eliminates the time, cost and risks of moving data and could be a competitive differentiator. https://cdn.ttgtmedia.com/rms/onlineimages/chatbot_g1150454068.jpg https://www.techtarget.com/searchdatamanagement/news/366643641/New-Starburst-platform-extends-AI-to-distributed-data Thu, 28 May 2026 08:00:00 GMT New Starburst platform extends AI to distributed data Enterprise systems have long been built around predictability, with leadership controlling outcomes through policy and access. AI breaks that model by learning and adapting in production, where a model can run normally while its outputs quietly shift, but with no red flags to warn governance that something has changed. It's precisely this kind of silent failure that sends traditional governance spiraling out of control. Most oversight models assume that problems will pop up as loud events, like breaches, failures or compliance issues, that teams can investigate after the fact, but AI problems don't always fail that way. AI can <a href="https://www.techtarget.com/searchenterpriseai/tip/How-to-identify-and-manage-AI-model-drift">fail silently through more subtle methods</a>, such as degraded judgment, inconsistent decisions and biased outcomes. By the time anyone notices, the system has often already caused economic damage. This forces a new approach to governance. Accepting that AI can fail quietly means companies need to rethink how they secure, monitor and control AI at scale. Visibility, validation and continuous control take precedence, and without them, leaders make decisions in the dark, based on systems they don't fully understand. <section class="section main-article-chapter" data-menu-title="AI changes how risk manifests in the enterprise"> <h2 class="section-title">AI changes how risk manifests in the enterprise</h2> In enterprise risk management, <a href="https://www.techtarget.com/searchcio/feature/8-top-enterprise-risk-management-trends">traditional methods</a> have long relied on shielding systems and enforcing policies, giving leaders confidence in the outcomes of their investments. Familiar controls, such as <a href="https://www.techtarget.com/searchsecurity/tip/Types-of-access-control">access control</a> and system stability, remain in place. Yet the problem lies with AI, which introduces a novel category of risk at the intersection of technology and decision-making. A technical system does not need to be compromised to be unreliable. It could follow every security protocol and still generate outcomes that trigger regulatory problems and damage a company's reputation. This scenario is the hardest for executives to catch, since the system doesn't crash; it just starts working differently, and small variations accumulate over time. Consider an AI system deployed to prioritize customer interactions. It passes security reviews, complies with internal policy and shows no breaches, outages or obvious warning signs. As customer behavior changes, the system adapts -- each decision looks reasonable, and performance metrics remain largely intact. Over time, specific customer segments experience longer delays, service commitments slip and complaints begin to rise, <a href="https://www.computerweekly.com/news/366618123/Lords-debate-government-approach-to-automated-decision-making">raising questions</a> around fairness, transparency and accountability in automated decision-making. Leadership is left searching for an explanation, but there is no single failure to point to. The system did not break or become compromised; it simply drifted from the organization's original intent. It alters how executives must think about supervision. In terms of governance, this is the hardest scenario to identify and trace -- AI risks cannot be assessed and revisited later. They evolve relentlessly, and as a result, governance must do the same. </section> <section class="section main-article-chapter" data-menu-title="The risks unique to AI systems"> <h2 class="section-title">The risks unique to AI systems</h2> It's the <a href="https://www.techtarget.com/searchdatamanagement/opinion/Data-quality-fuels-analytics-AI">quality of the data that matters</a> when building and training AI systems. With data flowing in from countless internal and external sources, the model can amplify even subtle flaws or biases and surface them in its decisions. Poorly managed feedback loops, automated data feeds and changes in the original data supply create the same effect -- AI systems that appear to function well but produce unreliable results, eroding confidence and complicating accountability. One of the biggest of these threats is model drift. As customer behavior and operating conditions change, AI systems can slowly fall out of sync with what they were initially designed to do. Often treated as a performance issue, it's really a governance issue. When <a href="https://www.techtarget.com/searchdatamanagement/opinion/Data-observability-for-AI-helps-curb-poor-model-performance">drift goes unnoticed</a>, organizations can end up with compliance and ethical problems that are hard to spot. Scaling AI systems compounds the risk. Training and update processes rely on shared tools, external data and prefabricated components. Without tight control over each of these, changes can wreak havoc across the entire system. For leaders, a single weak link can trigger a chain reaction that affects everyone. </section> <section class="section main-article-chapter" data-menu-title="Why lineage and provenance matter to leadership"> <h2 class="section-title">Why lineage and provenance matter to leadership</h2> Lineage and provenance give leaders the accountability they need when an AI system returns something it shouldn't. <a href="https://www.techtarget.com/searchdatamanagement/feature/How-data-lineage-became-a-boardroom-metric">Leaders need to know</a> what data the model was trained on, when it was introduced and who approved its use. <a href="https://www.techtarget.com/searchdatamanagement/opinion/Tracing-data-lineage-in-AI-systems">Lineage establishes the history</a> of the model itself, showing how teams built, updated and deployed it over time. Provenance traces the origin and handling of the data that shaped it. Without both, leaders are left guessing whether insufficient data, an unintended change or a breakdown in oversight caused a result. Without that clear view, dealing with regulators, auditors and customers also becomes extremely difficult. Simply stating that policies were followed is no longer sufficient. Organizations must clearly explain how a decision was made and what safeguards were in place. Provenance also plays a critical role in trust inside the organization. Employees are far more likely to rely on AI-driven insights when systems are transparent and well governed. When AI appears opaque or uncontrolled, adoption slows, skepticism grows and friction undermines broader AI initiatives. From a financial standpoint, the same visibility that builds trust also enables better investment decisions. When leaders can see what is working, what needs improvement and where controls are effective, they can allocate resources with confidence. Uncontrolled AI remains a black box, and black boxes pose financial risks that are difficult to quantify or justify. </section> <section class="section main-article-chapter" data-menu-title="AI expands what must be governed"> <h2 class="section-title">AI expands what must be governed</h2> AI isn’t the only asset that matters. All the supporting infrastructure that comes with it matters too. Training data, intermediate representations, user interfaces and feedback mechanisms all contribute to the picture, and many of those assets sit outside the bounds of traditional governance. Problems arise when leadership thinks they've got a handle on AI governance, but operational realities tell a different story. It's a blind spot that can be pricey to fill when exposed. Common indicators of this governance gap include the following: <ul type="disc" class="default-list"> <li>AI-related assets that are not captured in enterprise inventories.</li> <li>Limited visibility into how training data or feedback loops are changing over time.</li> <li>Inconsistent access controls across models, data and supporting systems.</li> <li>Reliance on one-time reviews instead of ongoing oversight.</li> </ul> Security teams can close this gap. The protocols they already manage -- identity, access, monitoring, and response – extend naturally to AI governance. </section> <section class="section main-article-chapter" data-menu-title="What security-first governance delivers"> <h2 class="section-title">What security-first governance delivers</h2> Security-first governance gives leadership confidence in AI-driven systems. Continuous oversight lets organizations assess how models behave as conditions change, rather than scrambling to explain outcomes after the fact. Teams can then detect issues early and course correct before real-world consequences. This approach also surfaces every modification rather than letting changes fly under the radar. It enables the organization to plan its budget so it can better respond to financial emergencies and regulatory pressure. In practice, leadership gains greater control over both risk and cost through the following: <ul type="disc" class="default-list"> <li>Clear visibility into changes made to models and data.</li> <li>Early identification of issues, leading to reduced remediation costs.</li> <li>More predictable investment in governance and oversight.</li> <li>Fewer surprise expenses driven by compliance or regulatory pressure.</li> </ul> Most importantly, this kind of governance enables organizations to add AI to any system with minimal risk. Defined boundaries in place let teams innovate and experiment within clear limits, and leadership can support AI initiatives with confidence. </section> <section class="section main-article-chapter" data-menu-title="Conclusion"> <h2 class="section-title">Conclusion</h2> Traditional oversight models no longer apply. Methods that work for fixed, predictable systems can't be counted on to control the rapid, adaptive nature of AI. The most practical alternative to tame the AI beast is by following a security-first approach grounded in visibility, verification and ongoing control, rather than a purely technical one. That approach rests on the following actions: <ul type="disc" class="default-list"> <li>Establish continuous visibility into how AI systems behave in real-world conditions.</li> <li>Validate data sources and model outcomes on an ongoing basis, not just at deployment.</li> <li>Enforce clear ownership and accountability for AI-driven decisions.</li> <li>Integrate AI oversight into existing security, risk and compliance processes.</li> <li>Measure success on reliability and trust over time, not only short-term gains.</li> <li>Detect and address unintended behavior early, before it escalates into business or regulatory impact.</li> </ul> AI governance is now a top priority, not because of the technology itself, but because controlling AI systems at scale requires deliberate, security-first oversight. Organizations that act on this will be better equipped to scale AI responsibly, apply restraint where needed and strengthen public trust as these systems take a more central role in their businesses. Liam Cleary is founder and owner of SharePlicity, a technology consulting company that helps organizations with internal and external collaboration, document and records management, business process automation, automation tool deployment, and security controls and protection. Cleary's areas of expertise include security on the Microsoft 365 and Azure platforms, PowerShell automation, and IT administration. Cleary is a Microsoft MVP and a Microsoft Certified Trainer. </section> AI systems fail quietly through drift, biased outputs and degraded judgment. A security-first governance approach gives leaders the visibility and continuous control to scale AI safely. https://cdn.ttgtmedia.com/rms/onlineimages/security_a135187239.jpg https://www.techtarget.com/searchdatamanagement/feature/Why-AI-forces-securityfirst-governance Wed, 27 May 2026 14:35:00 GMT Why AI forces security-first governance Without relevant, AI-ready data, agents are doomed to fail. <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/macmillan_andy.jpg" alt="Alteryx CEO Andy MacMillan">Andy MacMillan </div> In response, as agentic AI became the focal point of many enterprises' development initiatives over the past couple of years, longtime data preparation specialist Alteryx made its mission under CEO Andy MacMillan, who <a href="https://www.techtarget.com/searchbusinessanalytics/news/366616939/Alteryx-names-Andy-MacMillan-CEO-amid-ongoing-change">took over as the company's leader</a> in December 2024, to provide users with the canvas they need to make data AI-ready. Since then, however, Alteryx has expanded its ambitions under MacMillan. The emergence of AI over the past few years as a means of generating insights and automating business processes has forced many data management and analytics vendors to evolve. For example, data platform vendors Databricks and Snowflake now provide full-featured AI development environments. Database providers such as MongoDB and Couchbase similarly aim to become AI development platforms in addition to their historical focus. And analytics specialists including GoodData and Tableau are building on longtime semantic modeling capabilities to turn their platforms into context layers within agentic workflows. Alteryx, which <a href="https://www.techtarget.com/searchbusinessanalytics/news/252513750/Alteryx-makes-personnel-changes-to-navigate-cloud-journey">struggled to keep up with the pace of change</a> before being <a href="https://www.techtarget.com/searchbusinessanalytics/news/366563665/Alteryx-to-be-acquired-by-private-equity-firms-for-44-billion">sold to a private equity firm</a> in December 2023 and taken private to reorganize away from the public spotlight, is likewise evolving. However, beyond shifting from a data preparation platform for BI to a focus on AI-ready data, the vendor aims to become a critical part of agent workflows by enabling business users to add <a href="https://www.techtarget.com/whatis/definition/business-logic">business logic</a> -- rules, workflows and analysis -- to agents to help give them <a target="_blank" href="https://a16z.com/your-data-agents-need-context/" rel="noopener">the contextual awareness they require</a> to properly carry out their specific work. Unlike many AI development platforms that provide developers and engineers with capabilities for building agents, Alteryx's new Agent Studio is designed to empower Alteryx's user base of business analysts -- the experts in their domains with first-hand experience -- to connect data, logic and governance with AI. Its decentralized approach to development could be a differentiator. But, with only 11% of respondents to a recent Alteryx survey expecting responsibility for AI workflows to move to line-of-business domains over the next three years, this approach could be risky, according to consultant Donald Farmer, founder and principal of TreeHive Strategy, who noted that 11% is not a transformative number. In a recent interview before the start of Alteryx Inspire, the vendor's user conference in Orlando, Fla., MacMillan discussed the vendor's shift from a focus on data preparation for BI to AI-ready data. In addition, he spoke about Alteryx's different approach to developing agents, the struggles he sees from customers as they attempt to become <a href="https://www.techtarget.com/searchenterpriseai/feature/Businesses-gear-up-for-AI-agents-in-the-enterprise">agentic enterprises</a>, and what might be major themes in data and AI when Alteryx hosts its user conference in 2027. Editor's note: This Q&A has been edited for clarity & conciseness. When <a href="https://www.techtarget.com/searchbusinessanalytics/news/366617637/New-Alteryx-CEO-sees-platform-as-the-canvas-for-AI-prep">we last talked</a> shortly after you took over as CEO, you spoke about evolving Alteryx so it becomes the canvas for customers to prepare data for AI -- how is making data AI-ready different than making it ready for BI? Andy MacMillan: They're not entirely different. But what is different about it is the speed that agents are going to interact with data and <a href="https://www.techtarget.com/searchdatamanagement/feature/Data-and-AI-governance-must-team-up-for-AI-to-succeed">the governance around it</a>. With BI, you had snapshots. Everyone could look at the same dashboard for a week, and you could audit it. Now, if there is a <a href="https://www.techtarget.com/searchdatamanagement/feature/One-year-of-MCP-Support-a-must-for-data-management-vendors">Model Context Protocol</a> endpoint for an agent, and anyone can ask that agent a question, there's a different level of predictability that's needed. Now, instead of going to a Tableau or Qlik dashboard as I might have five years ago, I'm going to go to ChatGPT. But I want ChatGPT to give me answers with the same certainty and clarity that I would have gotten from those dashboards, and I also want ChatGPT's new capabilities, which are the analysis and probabilistic nature that can do the reasoning. Right now, I think people feel like they're trading those off. They feel like they have the consistency of their dashboard or the reasoning of the agent. Our goal is to make sure there isn't a tradeoff. When customers use Alteryx to prepare data, are there different things they need to do differently when getting it ready for AI than they did when preparing it for BI? MacMillan: I don't think there are things they have to do differently. I think there are just things overall that we all have to make sure that we're doing better, such as the governance around it to make sure <a href="https://www.techtarget.com/searchenterprisedesktop/feature/The-next-enterprise-AI-problem-is-visibility">there's visibility</a> and understandability and thinking through the permutations of how people are going to interact with the data can be different. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> Our ambitions have gotten bigger than simply data preparation. Now, [our ambitions] include providing some of the calculations and business logic that you could maybe argue is data prep, but is more than just cleaning data. </figure> <figcaption> Andy MacMillianCEO, Alteryx </figcaption> </div> </blockquote> What are different are the capabilities to build the workflow. Three years ago -- even two years ago -- Alteryx users could drag and drop tools onto a canvas and solve a problem, but you still had to know how to drag and drop all those tools. Now, you can type in what you're trying to do and watch it put the tools on the canvas that help you build the workflow. There's also the ability to use AI to interrogate the requirements. … At its core, the value proposition is empowering the person who knows the process the best, and knows the data the best, to be responsible for building [AI tools], and empowering them to keep it up to date. What does it require to be AI-ready? MacMillan: I think I have a different take than a lot of folks. There's been a lot of talk about <a href="https://www.techtarget.com/searchbusinessanalytics/feature/Agents-semantic-layers-among-top-data-analytics-trends">a semantic layer</a>. That's not bad -- putting labels on data and making sure it's clean is all reasonable. But I think to be AI-ready, data has to have gone through the operational understanding of what and how to use the data and how to pull it together for use cases, and then the agent has to know which use cases to use the dataset for. When I talk to folks about this, I point out that when you talk to the best people in your company about a topic, and you ask them a question, they usually ask you questions back. If I say, 'What was the revenue for our stores in California?', they respond, 'Do you want all the stores, or do you want the online sales that came through California?' Or maybe they point out that there were two stores in California that are no longer in business, but were in business during the first quarter and ask if that should be included. That's not a semantic layer problem. That's a business logic problem. Getting data AI-ready is not just having a semantic layer. It's also having business logic in a callable place, where that logic is in the calculation and the AI can call that and get the answer. That <a href="https://www.techtarget.com/searchenterpriseai/post/AI-agents-are-only-as-smart-as-the-data-that-feeds-them">business logic is the missing piece</a>. How does business logic get connected with AI to inform agents? MacMillan: There's a misconception that business logic is in all of an organization's old applications, and, having worked at some of these big <a href="https://www.techtarget.com/searchcloudcomputing/definition/Software-as-a-Service">SaaS</a> providers, I don't know that it always is. What we don't want to do is consume AI only through those applications. You want to be able to get to the data and pull it all together so you're not orchestrating AI on top of applications, but instead building logic that talks directly to data. Maybe the data came from those applications, but you're implementing that logic at the data layer in a visible, understandable, repeatable, auditable environment and connecting it with AI. Now, with business logic, there's a powerful platform for making AI work and understand the actual business. Where is Alteryx in its evolution toward becoming the canvas for getting data AI-ready? MacMillan: We're definitely there as we launch the capabilities that we're shipping. But I would say that our ambitions have gotten bigger than simply <a href="https://www.techtarget.com/searchbusinessanalytics/feature/Top-data-preparation-challenges-and-how-to-overcome-them">data preparation</a>. Now, [our ambitions] include providing some of the calculations and business logic that you could maybe argue is data prep, but is more than just <a href="https://www.techtarget.com/searchenterpriseai/tip/Clean-data-is-the-foundation-of-machine-learning">cleaning data</a>. We're helping the analysts and ops people of the world put their knowledge to work with the data in a bunch of different ways. Data prep is one of those ways, but building agents in Agent Studio is a lot more than data prep. That's taking prepared data and logic and activating it with AI. Beyond getting data AI-ready, what's a specific new role that Alteryx hopes to play? MacMillan: I think it's this business logic layer. What we're talking to customers about is making building AI agents simpler. Everyone today has access to AI, and most people that we talk to have access to a bunch of data. What we're trying to do is help them simply put that data to work for AI, and do that without trying to run through an application stack and without having some big orchestration project with 20 different platforms to get an agent to do the most basic sales and marketing things. All the data from sales and marketing applications is in a cloud data warehouse, and the sales and marketing operations team knows what that data means. We're letting them describe the logic, expose that logic to ChatGPT and create an agent. We're just trying to <a href="https://www.techtarget.com/searchenterpriseai/tip/Citizen-developers-are-redefining-enterprise-AI-development">provide our users a canvas to do that</a>. When you meet with customers, what are the biggest concerns you hear from them as they strive to become agentic enterprises? MacMillan: Data governance is a big one -- the tension between IT, <a href="https://www.techtarget.com/searchdatamanagement/definition/DataOps">DataOps</a> and the business. A lot of customers have a pristine data warehouse, but no one is allowed to use it, so that's not super helpful. Other customers have the opposite, where <a href="https://www.techtarget.com/searchdatamanagement/feature/Controlling-data-sprawl-requires-governance-discipline">data is everywhere</a> and they're struggling to manage it. The other big one that we're seeing is that the budget has shifted from AI being an IT-driven initiative to being a line-of-business initiative. With that shift has come responsibility, and with the responsibility has come a mandate to start using AI to solve actual problems, so customers are asking how to do that. They can't just use AI to write better emails. They have to use it to [improve operations]. That's the pressure I'm hearing at the moment from every one of our customers. That shift has happened quickly, and we're trying to be helpful as companies go through that shift. Things are evolving incredibly fast, so predicting the future is perhaps more difficult now than ever, but what do you think will be the major trends in data and AI a year from now? MacMillan: I think we're going to be talking about people modernizing their business logic away from their application portfolio and into an agent portfolio. That doesn't mean applications all go away, but I talk to so many people today that are constrained by their logic being kept in their enterprise resource planning and customer relationship management applications, and that constraint on their agentic growth is clearly going to be a problem. People are going to ask how to get to the data layer under that and go fast. People are going to realize they can build an agent when they get [constraints] out of the way and can just implement logic and go fast. We're going to move to an era when we agentify business logic, make it visible, understandable, repeatable, <a href="https://www.techtarget.com/searchdatamanagement/tip/Data-lineage-documentation-imperative-to-data-quality">auditable</a>, <a href="https://www.techtarget.com/searchbusinessanalytics/news/366618249/Trusted-data-at-the-core-of-successful-GenAI-adoption">trusted</a> and business-owned but running on the IT infrastructure environment. That's where we're headed. Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. As the vendor grows to meet changing customer needs, CEO Andy MacMillan says its goals have expanded beyond its data prep roots to include connecting agents with proper context. https://cdn.ttgtmedia.com/rms/onlineimages/ai_g1182183209.jpg https://www.techtarget.com/searchdatamanagement/feature/Evolving-Alteryx-focusing-on-AI-ready-data-logic-for-agents Wed, 20 May 2026 09:30:00 GMT Evolving Alteryx focusing on AI-ready data, logic for agents With many enterprises struggling to build AI tools that can be trusted in production, Informatica's latest product development initiatives aim to provide users with a foundation for building reliable agents. Unveiled on Wednesday during Informatica World, the vendor's user conference in Las Vegas, new features in <a href="https://www.techtarget.com/searchdatamanagement/news/366633382/Informatica-launches-agents-adds-new-AI-development-tools">the Informatica Intelligent Data Management Cloud</a> (IDMC) are highlighted by Agentic Multidomain MDM, which is a set of AI agents that continuously perform <a href="https://www.techtarget.com/searchdatamanagement/definition/master-data-management">master data management</a> tasks such as cleansing, enriching and stewarding data to maintain its quality, consistency and relevance. In addition, Informatica launched <a href="https://www.techtarget.com/searchapparchitecture/tip/An-overview-of-headless-architecture-design">headless</a> data management capabilities -- including native support for Model Context Protocol (MCP) to connect agents with data sources -- that enable users to oversee their data from a central repository while leaving the data where it lives. New integrations, including some with Salesforce, which <a href="https://www.techtarget.com/searchcustomerexperience/news/366624960/Salesforce-to-acquire-Informatica-in-8-billion-deal">acquired Informatica in 2025</a>, round out Informatica's IDMC additions. Given that the new features accelerate master data management to ensure that data can be trusted to inform AI tools, and that they provide a new architecture for both human and agentic workflows, they are valuable additions for Informatica users according to William McKnight, founder and president of McKnight Consulting. "This is an evolution of Informatica's IDMC from human-directed middleware into more of an autonomous data workforce that continuously cleans and governs data," he said. "It introduces headless data management … to provide front-end tools with real-time, context-rich data. By minimizing the traditional data trust bottleneck, this could enable AI to safely execute workflows independently." Plenty of other providers are also adding agentic features, McKnight continued. However, Informatica's <a href="https://www.techtarget.com/searchenterpriseai/feature/How-to-ensure-AI-transparency-explainability-and-trust">focus on trust</a> and native integrations with key Salesforce capabilities help distinguish the vendor. "Informatica differentiates through a focus on trusted data to drive live, real-time business transactions and automated actions," McKnight said. "Natively pairing with Salesforce allows it to stream trusted master data directly into front-end workflows. This headless framework could enable Informatica to embed complex data management into everyday business applications more fluidly than competitors." Based in Redwood City, Calif., Informatica enables customers to integrate, prepare and govern data to ready it for analytics and AI. Peers range from fellow specialists such as <a href="https://www.techtarget.com/searchdatamanagement/news/366580432/Collibra-launches-AI-Governance-unveils-GenAI-capabilities">Collibra</a> and <a href="https://www.techtarget.com/searchdatamanagement/news/366632699/Fivetran-DBT-Labs-merge-to-add-complementary-capabilities">Fivetran</a> to hyperscale cloud providers that offer master data management capabilities, including AWS, Google Cloud and Microsoft. <section class="section main-article-chapter" data-menu-title="A trust foundation"> <h2 class="section-title">A trust foundation</h2> As enterprises have <a href="https://www.gartner.com/en/newsroom/press-releases/2026-1-15-gartner-says-worldwide-ai-spending-will-total-2-point-5-trillion-dollars-in-2026">increasingly invested</a> in AI development over the past few years, but <a href="https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf">struggled to build</a> tools trustworthy enough to move into production, data management and analytics providers have recently made enabling users to discover and operationalize contextually appropriate data a priority. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> This is an evolution of Informatica's IDMC from human-directed middleware into more of an autonomous data workforce that continuously cleans and governs data. … By minimizing the traditional data trust bottleneck, this could enable AI to safely execute workflows independently. </figure> <figcaption> William McKnightPresident, McKnight Consulting </figcaption> </div> </blockquote> Without contextually relevant data -- such as supply chain data for a supply chain optimization agent -- AI tools will fail to deliver trustworthy outputs and projects will fail. Vendors, however, have taken different approaches to improving data discovery and retrieval. For example, Databricks <a href="https://www.techtarget.com/searchdatamanagement/news/366637142/New-Databricks-tool-aims-to-up-agentic-AI-response-accuracy">introduced Instructed Retriever</a> in January as an alternative to traditional retrieval-augmented generation, which has proven limited. MongoDB and Teradata have added capabilities that refine <a href="https://www.techtarget.com/searchdatamanagement/feature/Vector-search-now-a-critical-component-of-GenAI-development">vector indexing and search</a> to make contextually relevant data easier to discover. And GoodData and Tableau have built <a href="https://www.techtarget.com/searchenterpriseai/tip/Exploring-the-context-layer-for-AI-systems">context layers</a> that similarly aim to make it easier to feed agents contextually relevant data. Like GoodData and Tableau that built on existing capabilities to aid AI development -- in their case semantic layers -- Informatica is building on existing master data management capabilities to provide users with a trusted data foundation for AI. Agentic Multidomain MDM includes a Data Steward Agent to automate the time-consuming work of resolving quality issues and matching records to ensure accuracy. Informatica Agentic Integration automatically <a href="https://www.techtarget.com/searchdatamanagement/opinion/Why-agentic-AI-demands-both-structured-and-unstructured-data">joins structured and unstructured data</a> and feeds it into appropriate AI pipelines, and Data Quality and Metadata Enrichment Agents automate governance. Meanwhile, headless data management, automated by <a href="https://www.techtarget.com/searchdatamanagement/news/366537106/Informatica-unveils-plan-to-infuse-Claire-with-generative-AI">Informatica's Claire AI engine</a>, provides the architecture for organizations to automate their AI workflows. Driven by a desire to improve efficiency through AI, demand for high-quality data is increasing exponentially, according to Gaurav Pathak, senior vice president of product management for Informatica. Developing Agentic Multidomain MDM and headless data management was a response to the increasing demand for trusted data to fuel applications that improve business efficiency, he continued. "We have always focused on increasing data management productivity, and both of these [increase productivity]," Pathak said. "With headless data management, we are making that productivity available across every surface. That's the main driver." Customer feedback showed that many AI initiatives are held back by systems that <a href="https://www.techtarget.com/searchdatamanagement/feature/Reconsider-the-AI-readiness-gap-in-data-and-analytics">aren't ready for AI</a>, he added, noting that the new features aim to help users overcome systemic barriers. "The idea is to work with data quality tools to make sure that bad data does not reach agents, that context is available, make sure that no private, sensitive data is available to agents," Pathak said. </section> <section class="section main-article-chapter" data-menu-title="The user perspective"> <h2 class="section-title">The user perspective</h2> One of the customers using Informatica as part of its AI pipeline is Yum! Brands, a multinational fast food corporation based in Louisville, Ky., that operates such brands as KFC, Pizza Hut and Taco Bell. Yum! has been an Informatica user for about a decade, and over that time its use of Informatica has evolved from extract, transform and load workloads to <a href="https://www.techtarget.com/searchdatamanagement/definition/Extract-Load-Transform-ELT">get data in and out of systems</a> to master data management and governance. Now, the company is using Informatica as the context layer for AI, according to Kartik Pillai, Yum!'s director of data strategy, master data management, AI and data governance. "For us, Informatica plays a role in providing that golden record," he said. "It's about a golden record with context for agents -- the governance, the policies, the lineage, the quality that provides the holistic 360-degree record for an agent to act on. That becomes a feeder for some of our more custom agents." One such agent is for forecasting labor needs at individual restaurants, a process that requires highly specific data, Pillai continued. Another is for financial reporting such as same-store sales growth. As Yum! has experimented with AI, Pillai noted one of the barriers to moving pilots into production has been the ideal nature of proof-of-concept projects versus the messy reality of production environments. Governance, therefore, <a href="https://www.techtarget.com/searchdatamanagement/feature/Data-and-AI-governance-must-team-up-for-AI-to-succeed">has become critical</a> to ensure consistency so agents can be trusted. Agentic Multidomain MDM and headless data management, which Yum! has experimented with in recent weeks, could further enable the company to build production-ready agents, according to Pillai. "We have been conceptualizing with them, and I think definitely they will [help]," he said. "As far as getting things into production, it takes a bit of momentum to get it there." </section> <section class="section main-article-chapter" data-menu-title="Additional capabilities and next steps"> <h2 class="section-title">Additional capabilities and next steps</h2> Beyond Agentic Multidomain MDM and headless data management for AI, Informatica unveiled the following new capabilities: <ul type="disc" class="default-list"> <li>Data 360 Connector and Scanner, an integration with <a href="https://www.techtarget.com/searchcustomerexperience/news/366636353/Salesforce-adds-Informatica-to-Data-360-MuleSoft-fold">Salesforce Data 360</a> that enables the real-time, bi-directional flow of data between Salesforce Data 360 and any enterprise system.</li> <li>MDM Aware Data 360 to deliver trusted records to Salesforce Data 360 where users can activate data for real-time applications.</li> <li>Claire in Slack, providing agents for data quality, data discovery and governance -- among others -- in Slack conversations so users don't have to switch between environments to complete their work.</li> <li>Integrations with data platform and hyperscale cloud vendors simplify AI development, including availability of Informatica MCP servers in development environments such as <a href="https://www.theserverside.com/blog/Coffee-Talk-Java-News-Stories-and-Opinions/What-is-Amazon-Bedrock">Amazon Bedrock</a>, Databricks <a href="https://www.techtarget.com/searchbusinessanalytics/news/366625695/Latest-Databricks-tools-use-AI-to-simplify-AI-development">Agent Bricks</a>, Microsoft <a href="https://www.techtarget.com/searchenterpriseai/news/366616024/Microsoft-intros-Azure-AI-Foundry-for-building-AI-apps">Foundry</a> and Snowflake <a href="https://www.techtarget.com/searchdatamanagement/news/366625218/Snowflake-continues-to-add-AI-boost-Cortex-capabilities">Cortex AI</a>.</li> </ul> The integrations with Salesforce provide evidence that Informatica and Salesforce are a complementary fit, according to McKnight. However, accelerating master data management with Agentic Multidomain MDM is the most valuable new feature, he continued. "As a long-time proponent of master data management, I would have to say that the accelerating of master data management with agents stands out for its efficiency gains in this important area," McKnight said. Looking ahead, Informatica's product development plans focus on making both agents and humans more productive and accurate, according to Pathak. Toward that end, initiatives include using <a href="https://www.techtarget.com/searchdatamanagement/tip/Metadata-management-standards-examples-that-guide-success">metadata</a> to make relevant data available to agents, providing users reusable data products that can aid AI development, and expanding <a href="https://www.techtarget.com/searchapparchitecture/tip/AI-Agents-role-in-IT-infrastructure-is-expanding">AI-assisted data management and stewardship</a>. "Our vision for AI to be governed 90% by AI and humans providing 10% by supervising and [implementing] governance and guardrails," Pathak said. McKnight, meanwhile, suggested that as more organizations put AI tools into production, Informatica could address emerging issues such as <a href="https://www.techtarget.com/searchenterpriseai/tip/How-to-navigate-data-sovereignty-for-AI-compliance">data sovereignty</a> and multi-system governance. "As they are about infrastructure and not a database or cloud company, they could do things like expanding on their recent framework for Microsoft Fabric Open Mirroring by building zero-copy governance bridges that span multiple competing cloud ecosystems simultaneously," he said. "They could launch local data sovereignty agents explicitly designed for localized infrastructure." Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> Agents that continuously perform master data management tasks and a headless architecture for data management improve access to data that can be trusted to inform outputs. https://cdn.ttgtmedia.com/rms/onlineimages/ai_a252657224.jpg https://www.techtarget.com/searchdatamanagement/news/366643437/Informatica-update-aims-to-provide-trust-foundation-for-AI Wed, 20 May 2026 09:00:00 GMT Informatica update aims to provide trust foundation for AI Alteryx on Wednesday introduced new features that unite data with trusted business rules and workflows to aid customers building agents and other AI applications. Unveiled during the vendor's Inspire user conference in Orlando, Fla., Agent Studio and the Alteryx One MCP Server simplify converting data workflows into <a href="https://www.techtarget.com/searchenterpriseai/tip/How-to-build-your-first-agentic-AI-system">agentic AI systems</a>, enabling business analysts to use their expertise to build AI tools rather than rely on centralized IT teams. Launched <a href="https://www.techtarget.com/searchdatamanagement/news/366623973/Alteryx-One-launch-aims-to-unify-simplify-vendors-platform">in May 2025</a>, Alteryx One is Alteryx's platform for data management and insight generation, unifying previously disparate capabilities such as analytics automation and no-code data preparation. Agent Studio is a new feature within the Alteryx One platform that allows users to easily transform trusted datasets and <a href="https://www.techtarget.com/whatis/definition/business-logic">business logic</a> -- rules, workflows and analysis -- into autonomous agents that can be deployed in Alteryx or fed into the agent orchestration frameworks now provided by third-party vendors. MCP Server is Alteryx's version of a <a target="_blank" href="https://modelcontextprotocol.io/docs/getting-started/intro" rel="noopener">Model Context Protocol</a> server to extend agents beyond Alteryx One into applications such as Slack and Microsoft Teams and external AI models so the agents can securely access information beyond Alteryx's environment. In addition, Alteryx introduced new workflow deployment options, governance capabilities, and an Alteryx One desktop application to unify Alteryx tools such as Designer and AI Tooling for desktop users. With many enterprises making AI development a priority, and others opting for the security and cost-control of on-premises workflows over the cloud, the new features are significant because they address the varying needs of Alteryx customers, according to David Menninger, an analyst at ISG Software Research. "These new features provide an agentic AI framework for Alteryx's users, which is important given the focus on AI in today’s market," he said. "In addition, there is a revival of interest in on-premises capabilities both for governance reasons and to address cost concerns. Several of these features address those concerns." Based in Irvine, Calif., Alteryx is a longtime data management provider that enables customers to integrate and prepare data for analytics and AI initiatives. After a clumsy transition to the cloud and slow revenue growth, <a href="https://www.techtarget.com/searchbusinessanalytics/news/366563665/Alteryx-to-be-acquired-by-private-equity-firms-for-44-billion">the vendor was acquired</a> by a private equity firm and taken private so it could reorganize out of the spotlight of the public markets. Competitors include <a href="https://www.techtarget.com/searchdatamanagement/news/366633382/Informatica-launches-agents-adds-new-AI-development-tools">Informatica</a> and <a href="https://www.techtarget.com/searchbusinessanalytics/news/366641671/Latest-Qlik-tools-target-helping-users-achieve-AI-goals">Qlik</a>, among others. <section class="section main-article-chapter" data-menu-title="Fueling AI"> <h2 class="section-title">Fueling AI</h2> As enterprises <a target="_blank" href="https://kpmg.com/us/en/media/news/q1-ai-pulse2026.html" rel="noopener">increase their investments</a> in AI development but <a target="_blank" href="https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf" rel="noopener">frequently struggle</a> to move AI initiatives past experimentation and into production, feeding agents and other AI tools the high-quality, relevant data they require to properly perform has been a common hurdle. In response, data management and analytics vendors such as <a href="https://www.techtarget.com/searchdatamanagement/news/366637142/New-Databricks-tool-aims-to-up-agentic-AI-response-accuracy">Databricks</a>, <a href="https://www.techtarget.com/searchdatamanagement/news/366637414/MongoDB-launches-latest-Voyage-models-to-aid-AI-development">MongoDB</a> and <a href="https://www.techtarget.com/searchbusinessanalytics/news/366642778/Tableau-repositions-for-AI-unveils-new-knowledge-layer">Tableau</a> have prioritized providing tools that help customers discover and deliver contextually appropriate data to AI tools. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> These new features provide an agentic AI framework for Alteryx's users, which is important given the focus on AI in today’s market. In addition, there is a revival of interest in on-premises capabilities both for governance reasons and to address cost concerns. Several of these features address those concerns. </figure> <figcaption> David MenningerAnalyst, ISG Software Research </figcaption> </div> </blockquote> With Agent Studio and its MCP Server, Alteryx is similarly adding capabilities designed to help customers deliver trusted, relevant data to AI-powered systems in a move motivated by a combination of customer feedback and first-hand experience building agents, according to Alteryx CEO Andy MacMillan. "A lot of the AI capabilities, the idea that we want to have visible, trusted, auditable data in agents, has come from first-hand experience … being business analysts trying to bring data to AI," he said. However, by making Agent Studio and MCP Server part of Alteryx One -- a low-code/no-code platform for data management and insight generation -- Alteryx is taking a different approach to AI development than many other data management and analytics vendors. Instead of creating a development environment for centralized IT teams, it is <a href="https://www.techtarget.com/searchenterpriseai/tip/Citizen-developers-are-redefining-enterprise-AI-development">empowering business users</a> to build agentic AI tools. Beyond the first-hand experience MacMillan cited, an Alteryx survey of more than 1,400 business leaders showed that 11% of respondents expect responsibility for AI workflows to move to line-of-business domains over the next three years. "Agent Studio, MCP Server and a lot of the things we're talking about are designed around how to make AI trusted, and how to make it trusted is by empowering Alteryx users -- the business analyst -- to be the one to connect enterprise data, business logic and governance in a way that the business can depend on," McMillan said. Donald Farmer, founder and principal of TreeHive Strategy, noted that Alteryx is taking a novel approach by empowering business users to build <a href="https://www.techtarget.com/searchcustomerexperience/feature/The-front-office-is-being-rebuilt-around-AI-workflows">AI workflows</a>. However, while Alteryx is now providing the AI development capabilities it needs to remain viable, its approach is questionable, he continued.  "The work on an MCP server is necessary," he said. "Alteryx needs this capability to remain credible. Its historical differentiation has been business logic captured in the workflow. Exposing that through MCP is a coherent move. Whether enterprises want to route their [large language model] traffic through Alteryx workflows rather than building pipelines elsewhere is an open question." In addition, basing a strategy on 11% of organizations expecting to decentralize agent development management could prove dubious, according to Farmer. "That's not a transformative number," he said. "In fact, it is well within the margin of simple organizational drift." Menninger, however, countered that Alteryx has built a differentiated business over the years that the empowerment of business users builds upon. Alteryx's main focus has been data preparation. In addition, however, it provides analytics operations capabilities that ensure consistency and governance within analytics workflows. "These new features bring Alteryx's unique capabilities to the world of agentic AI," Menninger said. Specifically, they enable users to integrate <a href="https://www.techtarget.com/searchdatamanagement/definition/deterministic-probabilistic-data">deterministic</a> Alteryx workflows established over time with probabilistic agent-based processes, he continued. "By providing an agentic framework, Alteryx customers can more easily bring these two types of processes together," Menninger said. Beyond Agent Studio and the Alteryx One MCP Server, new Alteryx capabilities include the following: <ul type="disc" class="default-list"> <li>An Alteryx One desktop app for users that prefer a desktop environment to the web.</li> <li>New deployment options including Workspace Execution so users can run workflows in the cloud, Data Bridge to enable cloud-based workflows to securely connect with on-premises and private network data without moving it into the cloud, and Server Execution so analysts can view and manage server-based workflows from the cloud while running them on premises.</li> <li>Live Query and new connectors that allow users to work with data where it lives rather than moving it into Alteryx.</li> <li>Data Labels and asset certification that show where data comes from, who within an enterprise <a href="https://www.techtarget.com/searchdatamanagement/tip/The-data-ownership-blind-spots-putting-organizations-at-risk">is responsible for it,</a> and how it is being used to inform data and AI initiatives.</li> </ul> Data Bridge and Server Execution -- which is not yet generally available -- are valuable additions, according to Farmer. However, he noted that by launching capabilities that enable workflow orchestration from the cloud before making Server Execution GA, Alteryx, while putting in time-consuming product development work, appears to be prioritizing its cloud business over its historical base of <a href="https://www.computerweekly.com/feature/Why-run-AI-on-premise">on-premises users</a>. "Cloud-managed orchestration of on-premises workflows is exactly what large customers have been asking for, [but] shipping the cloud-native execution path first suggests the cloud business is being prioritized over the existing customer footprint," he said. </section> <section class="section main-article-chapter" data-menu-title="Looking ahead"> <h2 class="section-title">Looking ahead</h2> As Alteryx plans future product development, the shift from <a href="https://www.techtarget.com/searchenterpriseai/tip/Compare-top-AI-coding-tools">agentic coding</a> to agentic building and further removing AI development from centralized teams are focal points, according to MacMillan. Agentic AI tools such as Claude and ChatGPT <a href="https://www.techtarget.com/searchenterpriseai/feature/35-AI-content-generators-to-explore-in-2026">can write code</a> that helps Alteryx customers build agents. However, not all business users are experts in coding. The next step, therefore, is to enable large language models to not just write code, but build Alteryx workflows. "I think that is coming, agentic building for non-coders into environments that make sense for them that they trust," MacMillan said. "That's a really big one for us." Menninger, meanwhile, noted that enterprises struggle to integrate deterministic and <a target="_blank" href="https://www.bosch.com/research/bcai/probabilistic-modeling/" rel="noopener">probabilistic processes</a>. Therefore, adding more capabilities that enable customers to combine the two would benefit existing users and perhaps appeal to potential new ones. "Alteryx can help play a role in bringing these two worlds together by continuing to extend its agent-to-agent capabilities and supporting a mixture of those two types of activities," he said. Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> Capabilities include an MCP server and a tool that helps transform trusted data and logic into agents, with potential differentiation lying in the empowerment of business users. https://cdn.ttgtmedia.com/rms/onlineimages/ai_a279596285.jpg https://www.techtarget.com/searchdatamanagement/news/366643336/Latest-Alteryx-features-aim-to-boost-AI-powered-automation Wed, 20 May 2026 09:00:00 GMT Latest Alteryx features aim to boost AI-powered automation Streaming data specialist Confluent on Tuesday introduced new features for Confluent Cloud and Confluent Intelligence aimed at better enabling customers to build and secure AI applications fueled by real-time information. Revealed at the vendor's user conference in London, they include a fully managed <a href="https://www.techtarget.com/searchdatamanagement/feature/One-year-of-MCP-Support-a-must-for-data-management-vendors">Model Context Protocol (MCP) Server</a> that acts as a control center for developers and agents to build and manage streaming operations for AI using natural language, and built-in machine learning capabilities that detect and redact <a href="https://www.techtarget.com/searchsecurity/definition/personally-identifiable-information-PII">personally identifiable information</a> (PII) in Confluent's <a href="https://www.techtarget.com/searchdatamanagement/definition/Apache-Flink">Apache Flink</a> streaming engine. In addition, new Confluent capabilities include support for the open source <a target="_blank" href="https://agentskills.io/home" rel="noopener">Agent Skills</a> framework to add best practices for AI-powered operations, support for new large language models, and support for vector search on Amazon DynamoDB. Two of the biggest barriers enterprises face when trying to move AI projects into production are security related to sensitive data such as PII and the operational complexity of managing <a href="https://www.techtarget.com/searchdatamanagement/opinion/Real-time-data-streaming-for-AI-invest-where-it-matters">streaming data infrastructures</a>, according to Stephen Catanzano, an analyst at Omdia, a division of Informa TechTarget. Given that Confluent's new capabilities focus on those barriers to successful AI development, they are significant additions. "They directly address the two biggest barriers preventing AI projects from reaching production," Catanzano said. "By embedding automated PII redaction and private connectivity alongside natural language operations, Confluent is essentially removing the friction that causes eight in ten companies to struggle with scaling AI." Founded in 2014 to commercialize the open source <a href="https://kafka.apache.org/">Apache Kafka</a> streaming platform and based in Mountain View, Calif., Confluent was recently <a href="https://www.techtarget.com/searchdatamanagement/news/366636098/IBM-acquiring-Confluent-to-boost-AI-development-capabilities">acquired by tech giant IBM</a> to add streaming data capabilities to its AI development platform. <section class="section main-article-chapter" data-menu-title="Streamlining streaming for AI"> <h2 class="section-title">Streamlining streaming for AI</h2> Enterprises continue <a href="https://kpmg.com/us/en/media/news/q1-ai-pulse2026.html">to invest in AI development,</a> but also continue to <a href="https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf">struggle to build AI tools</a> that can be trusted enough to move into production. In response, many data management and analytics vendors -- including Databricks, MongoDB and Tableau, among others -- have recently introduced tools aimed at improving AI pipelines and the data they feed AI applications so that outputs are more accurate and the tools can be trusted in production. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> These capabilities are a nice step forward for developers as they build and govern agentic applications. While they don't differentiate Confluent in the market, they do help it stay competitive. </figure> <figcaption> Kevin PetrieAnalyst, BARC U.S. </figcaption> </div> </blockquote> Now Confluent is similarly adding new capabilities aimed at refining AI development with its strategy shaped by customer feedback, according to Sean Falconer, the vendor's head of AI. "A big part of it came directly from customers," he said, noting that the biggest challenges customers face when building AI applications no longer relate to AI models, but instead relate to the accessibility and relevancy of the data informing AI applications. "We saw growing demand from teams trying to operationalize AI, especially around making real-time data easier to work with and easier to secure." In particular, developing the <a href="https://www.computerweekly.com/feature/Gartner-How-AI-will-transform-managed-network-services">fully managed</a> MCP server and adding support for Agent Skills were motivated by interactions with users, Falconer continued. "We saw strong adoption of our open source MCP server," he said. "Customers were already using it to manage and troubleshoot streaming infrastructure through AI tools, so the next logical step was giving them a fully managed experience that's easier to use in production." Specific new capabilities in Confluent Intelligence and Confluent Cloud include the following: <ul type="disc" class="default-list"> <li>The fully managed MCP server and support for Agent Skills to manage streaming data for real-time AI tools.</li> <li>Automated PII detection and redaction in Flink SQL.</li> <li>Secure connectivity to Microsoft Azure-hosted services with support for Azure Private Link.</li> <li>An open source adapter that integrates Flink SQL on Confluent Cloud with <a href="https://www.techtarget.com/searchdatamanagement/news/366632699/Fivetran-DBT-Labs-merge-to-add-complementary-capabilities">DBT Labs</a> so data engineers can easily build and manage streaming data pipelines using a familiar framework.</li> <li>Support for new AI models from Anthropic and Fireworks AI to build real-time AI applications.</li> <li>Support for vector search on Amazon DynamoDB to expand Confluent's ecosystem.</li> </ul> Collectively, Confluent's new capabilities, while they don't substantially differentiate Confluent from competing vendors providing AI development capabilities, keep Confluent competitive, according to Kevin Petrie, an analyst at BARC U.S. "I do believe these capabilities are a nice step forward for developers as they build and govern agentic applications," he said. "While they don't differentiate Confluent in the market, they do help it stay competitive." Perhaps the most critical of the new capabilities is automated PII detection and redaction, Petrie continued, noting that his firm's research shows that AI adopters prioritize data privacy above all other aspects of a <a href="https://www.techtarget.com/searchdatamanagement/feature/How-executives-can-build-a-responsible-AI-framework">responsible AI framework</a>. "Confluent's automated redaction of PII in Flink helps enforce privacy policies and satisfy regulatory requirements such as GDPR or CCPA while maintaining the real-time service levels that AI often needs," he said. In addition, Petrie noted that support for Agent Skills -- which was originally developed by Anthropic and <a target="_blank" href="https://www.bishoylabib.com/posts/claude-skills-comprehensive-guide" rel="noopener">made open source in December 2025</a> -- could give Confluent a temporary advantage. "Confluent has some early-mover advantage with its support of Agent Skills, which are fast becoming a must-have open format for providing AI applications with the context they need to deliver value," he said. Like Petrie, Catanzano called out the value of automated PII detection and redaction. "It solves the fundamental blocker that security teams face when deciding whether to allow data into AI pipelines," he said. "This single capability can unlock entire use cases in regulated industries like healthcare and financial services that were previously off-limits." Collectively, the new features are logically constructed to help customers more effectively build and secure real-time AI tools, Catanzano continued. However, <a href="https://www.computerweekly.com/feature/Why-AI-is-forcing-enterprises-to-rethink-observability">model monitoring</a> and <a href="https://www.techtarget.com/searchenterpriseai/tip/What-are-the-benefits-of-an-MLOps-framework">MLOps capabilities</a> are not included and could help customers as they continue to invest in AI development. "They've focused heavily on the data layer and security controls, but they haven't addressed model monitoring, drift detection, or other MLOps concerns that also plague production AI systems," Catanzano said. "[However], that may be intentional given their focus on being the streaming foundation rather than a complete AI platform." </section> <section class="section main-article-chapter" data-menu-title="Looking ahead"> <h2 class="section-title">Looking ahead</h2> As Confluent plans future product development, continuing to add and enhance features that help enterprises move AI initiatives into production at scale remains a focus, according to Falconer. "You'll continue to see us invest in areas like MCP, Agent Skills, agents and real-time context delivery so developers can more easily build AI applications and agents that stay connected to what's happening in the business right now," he said. "A lot of the industry is realizing that AI is only as useful as the quality and freshness of the context behind it." Security and <a href="https://www.techtarget.com/searchdatamanagement/feature/Data-and-AI-governance-must-team-up-for-AI-to-succeed">governance</a> are also priorities, Falconer continued. "Enterprises want to move faster with AI, but they also need confidence that sensitive data is protected and that these systems operate within the right controls and policies, so a big part of our focus is making secure, governed real-time data access a built-in part of the platform," he said. Catanzano noted that from a competitive standpoint, Confluent provides a broader combination of streaming, governance and AI-native features than Kafka and some <a href="https://www.techtarget.com/searchdatamanagement/news/366640882/Redpanda-launches-streaming-engine-optimized-for-AI">other competing platforms</a>. To continue distinguishing itself from its peers, Catanzano suggested that Confluent add prebuilt capabilities such as industry-specific templates to further streamline real-time AI application development. "They could differentiate further by creating industry-specific templates and prebuilt streaming pipelines for common AI use cases -- fraud detection, personalization, predictive maintenance -- that combine their governance, connectivity and agent capabilities into turnkey solutions that reduce time-to-value for new customers in regulated industries," he said. Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> A fully managed MCP server and machine learning-powered data privacy capabilities aid customers attempting to move real-time AI applications into production. https://cdn.ttgtmedia.com/rms/onlineimages/cloud_g1223481405.jpg https://www.techtarget.com/searchdatamanagement/news/366643312/Latest-from-Confluent-streamlines-use-of-streaming-for-AI Tue, 19 May 2026 05:00:00 GMT Latest from Confluent streamlines use of streaming for AI When a production model starts misbehaving, the first question is rarely "what's wrong with the model?" It's "what changed upstream?" Data lineage enables enterprise teams to answer that question quickly, sometimes across dozens of pipelines, transformations and feature stores. For AI projects, lineage is the difference between a confident root-cause analysis and a week of detective work. <section class="section main-article-chapter" data-menu-title="What is data lineage?"> <h2 class="section-title">What is data lineage?</h2> <a href="https://www.techtarget.com/searchdatamanagement/tip/How-data-lineage-tools-boost-data-governance-policies">Data lineage</a> is the documented, queryable record of how data moves and changes through a system. At its simplest, lineage tells you where a piece of data came from, what was done to it along the way and where it ended up. A complete lineage graph captures sources, <a href="https://www.techtarget.com/searchdatamanagement/tip/The-difference-between-data-cleansing-and-data-transformation">transformations</a> and sinks: databases, event streams and third-party APIs feeding into joins, aggregations and feature engineering steps, which in turn feed <a href="https://www.techtarget.com/searchenterpriseai/tip/Explore-the-role-of-training-data-in-AI-and-machine-learning">training data sets</a>, feature stores and model inputs. The best implementations track this at the column level, letting teams trace not just a table but a single feature back to the raw event that produced it. </section> <section class="section main-article-chapter" data-menu-title="Why AI projects need lineage more than traditional analytics"> <h2 class="section-title">Why AI projects need lineage more than traditional analytics</h2> Traditional analytics workflows generally end at a dashboard, where a knowledgeable analyst can check the result. AI workflows extend much further, through training, evaluation, deployment and inference, often across many teams and time horizons. A model trained today may be served for months, and the features it relies on may undergo multiple transformations before reaching it. When something <a href="https://www.informationweek.com/data-management/11-irritating-data-quality-issues">drifts, breaks or produces a biased prediction</a>, lineage answers the critical questions: which upstream table fed this feature, when did its definition last change and which other models depend on the same source? Lineage is also <a href="https://www.techtarget.com/searchsecurity/tip/State-of-data-privacy-laws">increasingly a regulatory requirement</a>. Frameworks such as the EU AI Act and emerging financial services guidance require organizations to demonstrate the source of the data behind a decision. Reconstructing that after the fact is painful. Capturing it as the pipelines run is straightforward. </section> <section class="section main-article-chapter" data-menu-title="What good lineage captures"> <h2 class="section-title">What good lineage captures</h2> Effective lineage goes beyond simple table-to-table arrows. A useful lineage system records: <ul type="disc" class="default-list"> <li>Source-to-sink paths. Every hop a piece of data takes from origin to destination.</li> <li>Column-level dependencies. Field-by-field tracking, since feature drift usually traces to a specific column rather than a whole table.</li> <li>Transformation logic. The actual SQL or code that produced each derived value.</li> <li>Temporal context. A point-in-time view of the lineage as it looked when the model was trained.</li> <li>Model linkage. The connection between features and the specific models and versions that consume them.</li> <li>Owners and contacts. The team or individual responsible for each node in the graph.</li> </ul> </section> <section class="section main-article-chapter" data-menu-title="Getting started"> <h2 class="section-title">Getting started</h2> A perfect lineage graph on day one is not the goal. Teams should start by instrumenting the pipelines that feed their most important models. Most modern orchestrators, such as Airflow, Dagster and Prefect, along with transformation tools like dbt and Spark, emit lineage metadata natively or through open standards like OpenLineage. That <a href="https://www.techtarget.com/searchdatamanagement/tip/How-to-make-a-metadata-management-framework">metadata then flows into a catalog</a> where it can be queried and visualized, and coverage extends outward from there, prioritizing the paths where incidents tend to originate. Building lineage as a separate, manual artifact is a tempting mistake. Hand-maintained diagrams go stale within weeks. Lineage is only trustworthy when it is generated automatically from the systems that move the data. </section> <section class="section main-article-chapter" data-menu-title="The payoff"> <h2 class="section-title">The payoff</h2> With reliable lineage, debugging a model regression becomes a graph traversal rather than an archaeological dig. Impact analysis becomes a query: change this column and see which models are affected. Audits become a matter of pulling a record rather than reconstructing one. For AI teams that want to <a href="https://www.techtarget.com/searchdatamanagement/tip/Data-lineage-documentation-imperative-to-data-quality">ship faster and trust what they ship</a>, data lineage is not optional infrastructure. It is the map for a territory of modern data systems that has grown too large to navigate without one. Stephen Catanzano is a senior analyst at Omdia where he covers data management and analytics. Omdia is a division of Informa TechTarget. Its analysts have business relationships with technology vendors. </section> Data lineage records how data moves through AI pipelines, turning model debugging, impact analysis and audits into queries rather than manual investigations. https://cdn.ttgtmedia.com/visuals/digdeeper/5.jpg https://www.techtarget.com/searchdatamanagement/opinion/Tracing-data-lineage-in-AI-systems Mon, 18 May 2026 09:00:00 GMT Tracing data lineage in AI systems AI models are only as good as the data feeding them, yet most teams discover this the hard way when an upstream schema change quietly corrupts features or a subtle shift in event semantics drifts a model's accuracy off course. Data contracts are emerging as the discipline that prevents these failures by treating data the way software engineers treat APIs, providing producers and consumers with a deliberate, versioned interface. <section class="section main-article-chapter" data-menu-title="What is a data contract?"> <h2 class="section-title">What is a data contract?</h2> A data contract is a formal, enforceable <a href="https://www.techtarget.com/searchenterpriseai/feature/Business-vs-provider-AI-software-restrictions-to-know">agreement between a data producer and its consumers</a> that specifies what the data will look like and how it will behave. Unlike loose documentation or tribal knowledge, contracts are machine-readable and validated automatically. A typical contract defines the schema, including field names, types and nullability; the semantics, which specify what each field represents and how it's measured; quality expectations such as freshness, completeness and valid ranges; and operational guarantees covering SLAs, versioning policy and breaking-change procedures. Either side of the data exchange knows exactly what to expect, and either side can detect a violation the moment it occurs. </section> <section class="section main-article-chapter" data-menu-title="Why AI projects need them more than traditional analytics"> <h2 class="section-title">Why AI projects need them more than traditional analytics</h2> Dashboards tolerate imperfect data, since a missing row or a slightly off number is usually just a visual blip, but ML systems don't have the same tolerance. Feature pipelines, training data sets and online inference all assume stable distributions and consistent semantics, so when an upstream team adds a new enum value or starts populating a field that was previously null, <a href="https://www.techtarget.com/searchdatamanagement/opinion/Data-contracts-help-build-trustworthy-data-products-for-AI">models can degrade silently</a>. Training-serving skew often stems from undocumented producer behavior that no one thought to communicate. Data contracts catch these issues at the source, before a single upstream deployment undoes months of careful tuning. </section> <section class="section main-article-chapter" data-menu-title="Core components"> <h2 class="section-title">Core components</h2> An effective contract goes beyond informal documentation to capture the following: <ul type="disc" class="default-list"> <li>Schema and types. Structural definition of data in formats such as JSON Schema, Protobuf, or Avro.</li> <li>Semantic definitions. The meaning behind each field, including its units, time zones, and <a href="https://www.techtarget.com/searchdatamanagement/opinion/Why-data-semantics-matters-for-context-aware-systems">business meaning</a>.</li> <li>Quality rules. Measurable expectations for the data, such as row counts, null thresholds, and valid ranges.</li> <li>Ownership. Clear <a href="https://www.techtarget.com/searchdatamanagement/tip/The-data-ownership-blind-spots-putting-organizations-at-risk">accountability for the data set</a>, naming the producing team and an on-call point of contact.</li> <li>Versioning policy. A defined process for rolling out both backward-compatible and breaking updates without disrupting downstream consumers.</li> <li>SLAs. Operational commitments covering freshness, availability and incident response.</li> </ul> </section> <section class="section main-article-chapter" data-menu-title="Getting started"> <h2 class="section-title">Getting started</h2> Organizations should begin where the pain is sharpest, identifying the two or three data sets on which ML models depend most and codifying contracts for those first. Validation belongs in <a href="https://www.techtarget.com/searchsoftwarequality/definition/continuous-integration">continuous integration</a> for the producer and at ingestion for the consumer, so violations fail loudly rather than slipping through unnoticed. Use a schema registry to track versions and automate compatibility checks. Breaking changes warrant the same discipline as an API change: announce them in advance, version the contract, deprecate the old version on a clear timeline and then retire it. The cultural shift matters as much as the tooling. Producers must accept accountability for the data they emit, and consumers must articulate what they actually need from it. This conversation, made explicit and durable through a written contract, is the real value of the practice. Tooling enforces the agreement, but the agreement itself is what aligns teams. </section> <section class="section main-article-chapter" data-menu-title="The payoff"> <h2 class="section-title">The payoff</h2> Teams that adopt data contracts spend less time firefighting and more time improving models. Failures move from late-stage, hard-to-diagnose drift to early, actionable alerts that surface near the producer rather than near the model. For AI projects, where <a href="https://www.techtarget.com/searchdatamanagement/opinion/Data-quality-fuels-analytics-AI">data quality is destiny</a>, contracts are quickly becoming non-negotiable infrastructure. Stephen Catanzano is a senior analyst at Omdia where he covers data management and analytics. Omdia is a division of Informa TechTarget. Its analysts have business relationships with technology vendors. </section> AI models fail silently when upstream data shifts. Data contracts prevent this by making schema, semantics and quality a binding agreement between producers and consumers. https://cdn.ttgtmedia.com/visuals/digdeeper/3.jpg https://www.techtarget.com/searchdatamanagement/opinion/Understanding-data-contracts-for-AI-projects Wed, 13 May 2026 15:30:00 GMT Understanding data contracts for AI projects As MongoDB expands beyond its database roots to create a unified data platform for running AI tools in production, the vendor is adding new vector indexing capabilities and improving the performance of its core platform. Vector embeddings are numerical representations of data that make both structured and unstructured data easy to discover through various search methods, including similarity and keyword. Such searches feed relevant data into pipelines that provide agents and other AI applications <a href="https://www.techtarget.com/searchenterpriseai/tip/Exploring-the-context-layer-for-AI-systems">the proper contextual knowledge</a> they need to deliver accurate outputs. Unveiled in preview on May 7, Automated Voyage AI Embeddings in MongoDB Vector Search automates creating vector embeddings via MongoDB's Voyage models, reducing the time it takes to build a search infrastructure from weeks when performed by humans to minutes. In addition, the launch of MongoDB 8.3, made generally available on May 7, improves database performance to meet the higher demands that AI workloads place on systems than traditional data management and analytics workloads. The new version delivers higher <a href="https://www.techtarget.com/searchapparchitecture/tip/Read-and-write-considerations-when-designing-APIs">reads and writes</a>, higher <a href="https://www.techtarget.com/searchdatamanagement/definition/ACID">ACID</a> transactions without requiring any changes in code, and is capable of handling more complex operations than MongoDB 8.0, according to the vendor. Together, the new and improved capabilities represent MongoDB's advancement toward becoming a unified data platform for AI, according to Mike Leone, an analyst at Moor Insights & Strategy. "It's a step forward because the ingredients underneath are real," he said, noting that MongoDB's aspiration is grounded in its capabilities. "MongoDB owns a top-tier embedding model, the operational database, and now the wiring between them, and very few competitors can say all three are first-party and tightly integrated. That's makes the platform claim land for me instead of feeling like marketing." William McKnight, president of McKnight consulting, likewise noted that MongoDB's new capabilities are valuable for users and represent progress for the vendor. However, he also pointed out that MongoDB's competitors are <a href="https://www.techtarget.com/searchdatamanagement/news/366633117/Couchbase-ups-database-vector-search-indexing-capabilities">adding similar capabilities</a>. "These enhancements reduce manual plumbing and provide performance gains, allowing enterprises to deploy secure, high-speed AI agents with minimal operational complexity," McKnight said. "They could also be viewed as table stakes since all major platforms are similarly adding support for AI agents." <section class="section main-article-chapter" data-menu-title="Data discovery for AI"> <h2 class="section-title">Data discovery for AI</h2> Based in New York City, MongoDB is a longtime database vendor that has <a href="https://www.techtarget.com/searchenterpriseai/news/366627557/Database-vendor-MongoDB-embraces-GenAI">expanded beyond its roots</a> to create a data platform for AI workloads over the past few years in response to <a target="_blank" href="https://kpmg.com/us/en/media/news/q1-ai-pulse2026.html" rel="noopener">surging enterprise interest</a> in developing and deploying agents and other AI applications. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> It's a step forward because the ingredients underneath are real. MongoDB owns a top-tier embedding model, the operational database, and now the wiring between them, and very few competitors can say all three are first-party and tightly integrated. </figure> <figcaption> Mike LeoneAnalyst, Moor Insights & Strategy </figcaption> </div> </blockquote> Competing vendors including database specialists, data platform vendors and hyperscalers such as <a href="https://www.techtarget.com/searchdatamanagement/news/366640598/Oracle-AI-Database-update-aims-to-ease-developing-agents">Oracle</a> and <a href="https://www.techtarget.com/searchdatamanagement/news/366577632/Vector-search-and-storage-key-to-AWS-database-strategy">AWS</a> have also made it a priority to add features that enable customers to build and manage AI tools. MongoDB's new capabilities, however, keep the vendor current, and the simplicity of its platform provides some differentiation, according to McKnight. "While specialized rivals lead in raw vector latency, MongoDB offers operational simplicity and long-term memory management by eliminating the need to sync data between disparate systems," he said. "It also has high-end capabilities for JSON-styled data. Ultimately, it's a pragmatic choice that combines enterprise-grade reliability and high-performance JSON storage with integrated AI orchestration." Despite heightened enterprise interest in AI development and tools provided by vendors such as MongoDB and its competitors designed to simplify the complex process of building agents and other AI applications, most AI initiatives <a href="https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf">never make it into production</a>. The reasons for the high failure rate vary, but the inability to retrieve relevant data, without which AI tools can't be trusted to deliver accurate outputs, is among them. By automating the process of creating vector embeddings -- which follows <a href="https://www.techtarget.com/searchdatamanagement/news/366637414/MongoDB-launches-latest-Voyage-models-to-aid-AI-development">MongoDB's January release</a> of five Voyage AI embedding and reranking models -- MongoDB is addressing the data retrieval problems that plague many AI projects, according to Pete Johnson, the vendor's field chief technology officer. "Without consistent, high-accuracy retrieval, you can't trust the decisions that an agent makes, and without that trust, you can't put an agent into production," he said. "That's the sentiment we hear from customers." Despite the common sentiment that accuracy problems can be addressed by upgrading to a new large language model, inaccuracy based on irrelevant data is not an LLM problem, Johnson continued. "Bad AI often less an LLM problem and more of a retrieval problem," he said. "The LLM can only act on the information that it's given, if that information … is lacking the right context, then the output will inevitably be wrong." In addition to MongoDB, data management vendors unveiling new capabilities aimed at feeding agents and other AI tools with more appropriate context since the start of 2026 include <a href="https://www.techtarget.com/searchdatamanagement/news/366637142/New-Databricks-tool-aims-to-up-agentic-AI-response-accuracy">Databricks</a>, GoodData, Qlik and <a href="https://www.techtarget.com/searchbusinessanalytics/news/366642778/Tableau-repositions-for-AI-unveils-new-knowledge-layer">Tableau</a>, among others. Given the need to discover contextually relevant data through <a href="https://www.techtarget.com/searchenterpriseai/definition/retrieval-augmented-generation">retrieval-augmented generation</a> pipelines for AI, while platform performance improvements are valuable, the Automated Voyage AI Embeddings are the most significant of MongoDB's new capabilities, according to Leone. "It's the one because the embedding pipeline is where production RAG quietly dies," he said. "Teams ship something that demos beautifully, then six months later the data has drifted, the embeddings haven't, and the agent is confidently retrieving last quarter's reality. Closing that loop in the database keeps an agent trustworthy a year after it ships, and that's where the real customer value shows up." McKnight similarly noted the value of automating <a href="https://www.techtarget.com/searchdatamanagement/feature/Vector-search-now-a-critical-component-of-GenAI-development">vector embedding generation</a>. "Automated Voyage AI Embeddings have the potential to reduce deployment time by enabling semantic search quickly," he said. "By providing real-time data updates and top-tier retrieval accuracy, this feature ensures that AI agents operate with the most current and precise context available." Beyond automated vector embedding creation and added database performance with the launch of MongoDB 8.3, the vendor made a new integration with <a target="_blank" href="https://docs.langchain.com/oss/javascript/langgraph/overview" rel="noopener">LangGraph.js</a> generally available and added cross-region connectivity for AWS PrivateLink. Collectively, the new features are designed to advance MongoDB's goal of becoming a platform for AI, according to Ben Cefalo, the vendor's chief product officer for core products. "These updates advance automated retrieval and persistent agent memory as part of our mission to unify the agentic AI stack, strengthen the core database foundation for mission critical workloads and provide with the skills to deploy production AI," he said. <figure class="main-article-image full-col" data-img-fullsize="https://www.techtarget.com/rms/onlineimages/how_a_vector_database_works-f.png"> <img data-src="https://www.techtarget.com/rms/onlineimages/how_a_vector_database_works-f_mobile.png" class="lazy" data-srcset="https://www.techtarget.com/rms/onlineimages/how_a_vector_database_works-f_mobile.png 960w,https://www.techtarget.com/rms/onlineimages/how_a_vector_database_works-f.png 1280w" alt="A graphic displays how a vector database works" data-credit="Informa TechTarget" height="196" width="560"> <div class="main-article-image-enlarge"> </div> </figure> </section> <section class="section main-article-chapter" data-menu-title="Next steps"> <h2 class="section-title">Next steps</h2> With so many <a href="https://www.techtarget.com/searchenterpriseai/feature/Businesses-gear-up-for-AI-agents-in-the-enterprise">enterprises building agents</a> and so many data and analytics providers trying to appeal to those enterprises by simplifying AI development, the vendors that best serve the needs of existing users and potentially capture new ones will be those that help customers see and fix problems quickly, according to Leone. "The next year is going to expose a lot of agents that looked great in a demo and quietly fail in production, and the vendors who win will be the ones who help customers catch that early," he said. Consequently, he suggested that MongoDB add <a href="https://www.techtarget.com/searchitoperations/podcast/AI-observability-Why-old-monitoring-fails-in-the-GenAI-era">agent observability</a> capabilities so developers and engineers can address potential issues with AI tools before they cause problems in production or get scrapped before they ever make it that far. "If I were MongoDB, I'd lean hard into agent observability and evaluation as a first-party capability, since that's the credibility layer behind every 'trust an agent at scale' claim they're already making," Leone said. "Owning that gives AI-native teams one less thing to stitch together from outside the platform." McKnight, meanwhile, suggested that MongoDB broaden its support for complex <a href="https://www.techtarget.com/searchdatamanagement/definition/data-structure">data structures</a>. He noted that the vendor excels at operational simplicity, but support for data structures such as tensors and matrices would enable it to better handle high-dimensionality data. "Furthermore, incorporating built-in search enhancements such as native spellcheck and real-time recommendations would bridge the gap between its current document-store roots and the specialized capabilities of pure-play search engines," McKnight said. Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> With enterprise data workloads feeding AI pipelines, the longtime database vendor is evolving -- along with competitors -- by building capabilities for cutting-edge development. https://cdn.ttgtmedia.com/rms/onlineimages/code_g1019737194.jpg https://www.techtarget.com/searchdatamanagement/news/366642768/MongoDB-adds-new-vector-performance-capabilities-to-aid-AI Fri, 08 May 2026 14:27:00 GMT MongoDB adds new vector, performance capabilities to aid AI With the introduction of the Autonomous Knowledge Platform, Teradata is planning to provide a new infrastructure for AI. Unveiled on Thursday, Teradata's new capabilities are designed to integrate AI development and management with analytics and data in a single system that can be deployed across cloud, on-premises and <a href="https://www.techtarget.com/searchcloudcomputing/post/Why-hybrid-cloud-architecture-is-becoming-the-default-for-AI">hybrid environments</a>. Capabilities of the Autonomous Knowledge Platform, among others, include Teradata AI Studio, which is a suite for developing and operating AI tools, a natural language interface for executing agentic workflows, and prebuilt agents that perform tasks such as infrastructure management and <a href="https://www.techtarget.com/searchenterpriseai/feature/FinOps-can-manage-AI-computing-costs-experts-say">cost optimization</a>. Because the new platform empowers agents and unifies previously disparate AI, analytics and data management capabilities, it is a significant addition for Teradata users, according to Stephen Catanzano, an analyst at Omdia, a division of Informa TechTarget. "The Autonomous Knowledge Platform represents a strong addition because it shifts from reactive to proactive infrastructure … that provides the business context and governance necessary for agents to sense, decide, and act reliably across enterprise environments, which wasn't previously possible in an integrated way," he said. Kevin Petrie, an analyst at BARC U.S., called the new platform, "important," noting that it will help Teradata compete in <a href="https://www.techtarget.com/searchbusinessanalytics/tip/Top-11-business-intelligence-challenges-and-how-to-overcome-them">an evolving market</a> for data management and analytics vendors as traditional business intelligence is replaced by AI-powered insight generation and process automation. "This is an important addition," he said. "The Autonomous Knowledge Platform makes Teradata more competitive in this space and enables its customers to layer agentic AI capabilities onto their existing data environments." Based in San Diego, Teradata is a data management and analytics provider that has prioritized enabling users to build and deploy AI tools with recent product development initiatives. In January, the vendor <a href="https://www.techtarget.com/searchdatamanagement/news/366637641/Teradatas-AgentStack-aims-to-simplify-building-managing-AI">unveiled Enterprise AgentStack</a>, a suite scheduled for general availability by midyear, designed to simplify developing and governing agents. In March, Teradata added <a href="https://www.techtarget.com/searchdatamanagement/news/366639802/Teradata-updates-vector-indexing-suite-to-aid-AI-development">new vector indexing capabilities</a> to better enable users to discover and retrieve the relevant data agents require to perform properly. <section class="section main-article-chapter" data-menu-title="Infrastructure for AI"> <h2 class="section-title">Infrastructure for AI</h2> Throughout 2026, customer feedback has led data management and analytics vendors to add capabilities that enable enterprises to develop agents that can be trusted to deliver accurate outputs so they can be put into production. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> The Autonomous Knowledge Platform represents a strong addition because it shifts from reactive to proactive infrastructure … that provides the business context and governance necessary for agents to sense, decide, and act reliably across enterprise environments. </figure> <figcaption> Stephen CatanzanoAnalyst, Omdia </figcaption> </div> </blockquote> Many organizations experimented with agents dating back to 2024, but few were able to build agents trustworthy enough to deliver any return on their investments. One of the reasons many AI initiatives <a target="_blank" href="https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf" rel="noopener">never made it past the pilot stage</a> was that the data retrieval processes used to feed AI pipelines couldn't discover and deliver enough high-quality, relevant data for agents to perform as intended. In response, vendors such as Databricks, Domo, GoodData, MongoDB, Qlik, Snowflake, Tableau and ThoughtSpot have all introduced new capabilities aimed at better enabling customers to <a href="https://www.techtarget.com/searchdatamanagement/feature/Data-quality-fast-failures-and-quick-wins-key-to-AI-success">successfully build agents</a> <a href="https://www.techtarget.com/searchdatamanagement/feature/Data-quality-fast-failures-and-quick-wins-key-to-AI-success"></a>rather than merely experiment with agentic AI development. Driven by customer feedback, Teradata is similarly aiming to improve the success rate of AI development initiatives, first with capabilities introduced earlier this year and now with the Autonomous Knowledge Platform, according to Sumeet Arora, the vendor's chief product officer. "Customer feedback was central," he said. "It came from hundreds of direct conversations with enterprises about how their relationship with data is changing -- who uses the platform, how they use it, and in what ways they need it to work differently as AI agents become part of daily operations. Those signals shaped every major element of the platform. Specific elements of the platform include the following: <ul type="disc" class="default-list"> <li>AI Studio to provide a single place for organizations to build, deploy and <a href="https://www.techtarget.com/searchdatamanagement/feature/Data-and-AI-governance-must-team-up-for-AI-to-succeed">govern AI tools</a>, including an agent for hybrid data retrieval, end-to-end AI and machine learning pipelines, and model lifecycle management tools.</li> <li>Tera, an AI-powered workspace featuring a natural language interface where users can execute agentic workflows.</li> <li>Tera agents, which are prebuilt tools for specific tasks.</li> <li>Teradata Cloud, the Autonomous Knowledge Platform's first available deployment option featuring elastic compute and active compute capabilities to address the cost and performance of AI workloads and integrations with data sources to reduce data duplication.</li> <li>Teradata Factory to enable on-premises deployments for customers with <a href="https://www.techtarget.com/whatis/definition/data-sovereignty">data sovereignty</a> and <a href="https://www.techtarget.com/searchenterpriseai/tip/Global-AI-legislation-and-regulation-tracker">regulatory concerns</a>.</li> </ul> "Underneath all of it is … AI moving closer to the data, not data moving to AI," Arora said. "That's an architectural principle that has shaped the platform from the ground up." From a competitive standpoint, even as a spate of other data management and analytics providers introduce capabilities aimed at improving the AI development process, Teradata's new suite is distinguished from those of competing vendors in some ways, according to Catanzano. In particular, he noted that capabilities which attempt to eliminate the need to choose either <a href="https://www.techtarget.com/searchdatacenter/tip/AI-capacity-planning-Balancing-flexibility-performance-and-risk">high performance</a> or low cost, and either cloud or on premises, are potential differentiators. In addition, the concept of autonomous knowledge -- the delivery of business context to agents -- is significant. "Autonomous knowledge that embeds business context, semantics and lineage directly into the platform gives agents trusted, governed understanding rather than just data access, setting it apart from vendors offering basic AI infrastructure," Catanzano said. "It seems to be a new, unique approach." Regarding the configuration of the Autonomous Knowledge Platform, he added that it seems logically built. However, Catanzano suggested that more features that create <a href="https://www.techtarget.com/searchbusinessanalytics/news/252515720/Gartner-Augmented-analytics-ecosystem-for-BI-now-key">a data and AI ecosystem</a> through integrations would add further effectiveness. "More clarity on real-time integration capabilities with existing enterprise systems and third-party tools would strengthen confidence in its ability to operate seamlessly across complex, heterogeneous environments," he said. Like Catanzano, Petrie called out the value of giving users the option to deploy their AI systems on premises, noting that BARC's research shows enterprises are expressing greater <a href="https://www.techtarget.com/searchenterpriseai/tip/How-to-navigate-data-sovereignty-for-AI-compliance">concern about data sovereignty</a> driven by regulatory mandates and US political developments. "While not unique in the industry, the Factory option for on-prem deployments is critical," he said. "Data platform vendors must meet [data] sovereignty requirements to compete in the global arena." In addition, Petrie noted that Teradata's new cost control and model lifecycle management capabilities help the Autonomous Knowledge Platform stand apart from competing AI development and management suites. "Many AI adopters struggle to anticipate and measure their consumption of AI tokens, which -- as with cloud compute -- can lead to budget-breaking bills," he said. "I also like Teradata's model lifecycle management capabilities. .... The more Teradata can help data and AI teams optimize how they build, train, and iterate ML models, the better they reduce complexity and speed AI projects." </section> <section class="section main-article-chapter" data-menu-title="Looking ahead"> <h2 class="section-title">Looking ahead</h2> After introducing the Autonomous Knowledge Platform, one of Teradata's next initiatives is to deepen the platform's capabilities to improve its ability to handle agentic <a href="https://www.techtarget.com/searchhrsoftware/feature/Beyond-Containment-Structuring-IT-for-enterprise-AI-at-scale">AI workloads at enterprise scale</a>, according to Arora. In addition, adding industry-specific context for AI similar to what ThoughtSpot is doing with its <a href="https://www.techtarget.com/searchbusinessanalytics/news/366640328/ThoughtSpot-domain-specific-Spotter-agents-target-AI-success">domain-specific Spotter agents</a> is part of Teradata's product development roadmap, Arora continued. "The frame for the next six months [is] serving enterprises with agents, and agents themselves [with agents]," he said. "Both are customers of this platform." Focusing on industry-specific agentic capabilities is wise, according to Catanzano.  "Teradata could expand its ecosystem by developing industry-specific agent templates and prebuilt autonomous workflows tailored to verticals like healthcare, finance, and manufacturing," he said. A marketplace for agents developed by third parties and integrations with <a href="https://www.techtarget.com/searchenterpriseai/feature/AI-agent-frameworks-A-guide-to-evaluating-agentic-platforms">agentic platforms</a> would also serve the needs of Teradata's users and perhaps attract new customers, Catanzano added. "Creating a marketplace for third-party agents and integrations would attract new users seeking rapid deployment while giving existing customers more flexibility to customize autonomous intelligence for their unique business processes," he said. Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> As many enterprises prepare to move past experimenting with agents, the vendor's new platform is purpose-built to help users move pilots into production. https://cdn.ttgtmedia.com/rms/onlineimages/iot_g1199144987.jpg https://www.techtarget.com/searchdatamanagement/news/366642649/Teradatas-latest-targets-putting-agentic-AI-into-production Thu, 07 May 2026 08:30:00 GMT Teradata's latest targets putting agentic AI into production Following closely on the heels of its March acquisition of master data management specialist Reltio, SAP on Monday revealed that it plans to acquire data lakehouse vendor Dremio and AI model developer Prior Labs to better enable AI development and management. As enterprise data management workloads evolve to become enablers of AI-powered insight generation and process automation, vendors such as SAP -- which <a href="https://www.techtarget.com/searchsap/news/366619376/SAP-data-cloud-Databricks-integration-aims-to-unify-AI-data">provides data platform capabilities</a> as part of its array of offerings beyond Enterprise Resource Planning (ERP) -- are attempting to simplify AI development and management for their customers. SAP's acquisition of <a href="https://www.techtarget.com/searchdatamanagement/news/366619541/Reltio-adds-real-time-data-delivery-to-fuel-fast-decisions">Reltio</a>, which SAP disclosed on March 27, adds master data management capabilities that enable users to unify their data to make it more easily accessible for AI and analytics workloads. The acquisition of Dremio, which provides a data lakehouse platform optimized for open source <a target="_blank" href="https://iceberg.apache.org/" rel="noopener">Apache Iceberg</a> tables that make data interoperable between platforms that support the format, will similarly simplify access to data for enterprise workloads. Purchasing Prior Labs, meanwhile, adds tabular foundation model capabilities that enable organizations to use data stored in tables to fuel predictive AI initiatives. "These acquisitions make sense strategically," Kevin Petrie, an analyst at BARC U.S., said. "SAP specializes in structured data, which remains the leading input for AI initiatives, and Prior Labs enriches the predictive AI capabilities that SAP users can apply to that data." Financial terms of the purchases were not disclosed, and each remains subject to regulatory approval and other customary closing requirements. SAP's <a href="https://www.techtarget.com/searchdatamanagement/news/366634167/Dremio-Cloud-An-autonomous-lakehouse-powered-by-AI-agents">Dremio</a> acquisition is expected to close during the third quarter of this year, while its purchase of Prior Labs is expected to close during second or third quarter of 2026. Founded in 2015 and based in Santa Clara, Calif., Dremio had raised $410 million in venture capital funding before its acquisition. Prior Labs, founded in 2024 and based in Berlin, had raised nine million Euros in funding. <section class="section main-article-chapter" data-menu-title="Additive capabilities"> <h2 class="section-title">Additive capabilities</h2> Although enterprises <a target="_blank" href="https://www.gartner.com/en/newsroom/press-releases/2026-1-15-gartner-says-worldwide-ai-spending-will-total-2-point-5-trillion-dollars-in-2026" rel="noopener">continue to invest heavily</a> in developing agents and other AI applications aimed at making employees better informed business processes more efficient, many are <a target="_blank" href="https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf" rel="noopener">struggling to build AI tools</a> that can be trusted to properly perform in production. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> These acquisitions make sense strategically. SAP specializes in structured data, which remains the leading input for AI initiatives, and Prior Labs enriches the predictive AI capabilities that SAP users can apply to that data. </figure> <figcaption> Kevin PetrieAnalyst, BARC U.S. </figcaption> </div> </blockquote> Feeding such applications with high-quality, relevant data is one of the main barriers that halt AI initiatives. Through integrations developed since the start of 2025, SAP now enables users of its Business Data Cloud to access data in Snowflake and Databricks. The acquisition of Dremio's <a href="https://www.techtarget.com/searchdatamanagement/news/366545117/Lakehouse-architecture-the-best-fit-for-modern-data-needs">lakehouse capabilities</a> will extend the reach of SAP's Business Data Cloud to a new array of external data sources -- including on-premises databases -- without forcing users to move data into SAP. "With Dremio, we're able to add the modern lakehouse architecture. … We're super excited about this opportunity," Irfan Khan, president and chief product officer of SAP data and analytics, said during a virtual press conference. David Menninger, an analyst at ISG Software Research, noted that the significance of the acquisition is that it demonstrates SAP's ongoing evolution toward enabling access to more than just <a href="https://www.techtarget.com/whatis/video/An-explanation-of-SAP-ERP">SAP's ERP data</a> for its users to build effective AI and analytics tools. "Dremio furthers SAP's recognition that enterprise customers have lots of data that is not in SAP and represents a commitment to provide support for those data sources," Menninger said. Petrie, meanwhile, noted that Dremio's lakehouse platform is particularly valuable to SAP customers because of its data federation capabilities. Data federation is a data management strategy that creates virtualized views of data so it doesn't have to be moved or copied, which eliminates the cost and risk of migrating data between platforms as well as any <a href="https://www.techtarget.com/whatis/definition/data-sovereignty">data sovereignty</a> concerns. "The Dremio acquisition … is critical because migration complexity and rising sovereignty concerns prevent organizations from moving all their analytics and AI inputs into SAP," Petrie said, adding that Dremio rates highly in BARC's user evaluation surveys. The acquisition of Prior Labs is aimed at better enabling SAP users to build models that fuel <a href="https://www.techtarget.com/searchenterpriseai/tip/Generative-AI-vs-predictive-AI-Understanding-the-differences">predictive AI initiatives</a>, according to Philipp Herzig, SAP's chief technology officer. "Similar to large language models for unstructured data, we want to democratize the access for predictive AI, a market which is as large as the generative AI market," Herzig said during the press conference. "That is exactly where Prior Labs comes in." Menninger noted that enterprises have struggled to use LLMs as part of their AI pipelines given that the non-deterministic nature of LLMs can lead to different outputs based on the same input. Tabular foundation models produce more reliable outputs. "It is hard to govern processes where the output may change each time the process is run," Menninger said. "Prior Labs is focused on delivering foundation models for structured, tabular data. That way, users can have the flexibility of LLMs applied in a deterministic way." </section> <section class="section main-article-chapter" data-menu-title="Consolidation continues"> <h2 class="section-title">Consolidation continues</h2> Beyond better enabling SAP users to develop agents and other AI tools, SAP's acquisitions of Dremio and Prior Labs are part of <a href="https://www.techtarget.com/searchdatamanagement/feature/4-trends-that-will-shape-data-management-and-AI-in-2026">a growing consolidation trend</a>. Consolidation tends to come and go in cycles. For example, there was significant consolidation among data and analytics vendors in 2007 when IBM acquired Cognos, Hyperion was bought by Oracle and SAP purchased BusinessObjects. More followed in 2019 when Tableau was acquired by Salesforce and Google bought Looker within days of each other. Beginning with <a href="https://www.techtarget.com/searchcustomerexperience/news/366624960/Salesforce-to-acquire-Informatica-in-8-billion-deal">Salesforce's acquisition of Informatica</a> in a deal that ultimately closed in November 2025, another consolidation wave seems to be rising. As enterprises attempt to simplify complex AI pipelines and lower the high cost of AI development by reducing the number of tools and vendors needed to create AI workflows, independent vendors are finding it more difficult to compete. Last November, data integration vendor Fivetran and data transformation specialist DBT Labs <a href="https://www.techtarget.com/searchdatamanagement/news/366632699/Fivetran-DBT-Labs-merge-to-add-complementary-capabilities">agreed to merge</a>. In December, <a href="https://www.techtarget.com/searchdatamanagement/news/366636098/IBM-acquiring-Confluent-to-boost-AI-development-capabilities">IBM acquired streaming data vendor Confluent</a>. Now, first with Reltio and then Monday's moves to buy Dremio and Prior Labs, SAP is boosting its AI development capabilities through acquisitions of previously independent companies. "The acquisitions are part of an inherent cycle in the software industry," Menninger said, while noting that there is usually room for a few independent vendors in each market segment. "Large companies acquire smaller companies as a way to supplement their R&D efforts. As markets become more mature, such as the data management market, it's natural for many of these companies to get acquired." Petrie similarly noted that acquisitions reflect platform vendors such as Salesforce, IBM and SAP taking advantage of the market to add capabilities that enable them to capture more of their <a href="https://www.computerweekly.com/microscope/news/366596056/Gartner-AI-is-driving-customer-spending">customers' spending</a>. In addition, such vendors are adding capabilities that provide competitive advantages by enabling access to competitors' platforms. "They recognize that migrations are tough and they need to help customers manage heterogeneous data environments," he said. SAP, without being specific, continues to look for further acquisition opportunities, according to Herzig. Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> The tech giant's latest purchases add a data lakehouse that enables users to access data across systems and tabular foundation models that fuel predictive AI. https://cdn.ttgtmedia.com/rms/onlineimages/money_g1050046190.jpg https://www.techtarget.com/searchdatamanagement/news/366642794/SAP-acquisitions-of-Dremio-Prior-Labs-target-AI-development Mon, 04 May 2026 10:24:00 GMT SAP acquisitions of Dremio, Prior Labs target AI development Mike Capone is stepping down from his role as CEO of Qlik. <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineImages/capone_mike.jpg" alt="Qlik CEO Mike Capone">Mike Capone </div> Mike Lipps, Qlik's board chair, has been named interim CEO while the vendor's board conducts a search for a permanent successor, according to a Qlik spokesperson. Capone's move comes just over two weeks after Qlik held Connect, its annual user conference in Kissimmee, Fla., where it <a href="https://www.techtarget.com/searchbusinessanalytics/news/366641671/Latest-Qlik-tools-target-helping-users-achieve-AI-goals">unveiled new features</a> collectively aimed at helping customers deploy the vendor's AI tools to generate insights as well as develop, deploy and manage agents and other AI applications of their own. "After more than eight years as CEO of Qlik, I've made the difficult decision that now is the right time for me to step down from the role," Capone posted on LinkedIn. "Leading Qlik has been a true privilege. I've been fortunate to work alongside friends and colleagues who care deeply, customers who pushed us to be better, and partners who have helped take Qlik further than we could have alone." Based in King of Prussia, Penn., Qlik is a longtime analytics vendor that under Capone's leadership evolved into a more full-featured data platform provider <a href="https://www.techtarget.com/searchbusinessanalytics/news/252526899/Qlik-launches-new-cloud-based-data-integration-platform">featuring data integration</a> and <a href="https://www.techtarget.com/searchbusinessanalytics/news/366638938/Qlik-launches-agentic-experience-to-fuel-AI-powered-analysis">AI capabilities</a>. <section class="section main-article-chapter" data-menu-title="Leadership in changing times"> <h2 class="section-title">Leadership in changing times</h2> Capone was appointed Qlik's CEO in January 2018 after serving three years as chief operating officer at Medidata Solutions. His resignation comes at a time when the data management and analytics industries are evolving. Capone joined Qlik during an era when self-service analytics fueled by robust data visualizations represented the cutting edge of business intelligence (BI). Qlik was viewed as one of the leading BI vendors, but was acquired by private equity firm Thoma Bravo in 2016 for $3 billion and taken private so the vendor could transform for the cloud away from the scrutiny of the public market. Qlik launched Qlik Sense Cloud Business in 2016, which was replaced in 2018 by <a href="https://www.techtarget.com/searchbusinessanalytics/news/252471573/Qlik-Sense-Business-improves-Qliks-cloud-AI-capabilities">Qlik Sense Business</a>, a fully managed SaaS version of the vendor's enterprise analytics platform deployed on Qlik's own cloud. By January 2022, after a 5-year process to reorganize and expand by adding data integration capabilities through a series of acquisitions, <a href="https://www.techtarget.com/searchbusinessanalytics/news/252511695/Qlik-planning-an-IPO-files-application-with-the-SEC">Qlik filed paperwork</a> with the Securities and Exchange Commission for an initial public stock offering and a return to the public markets. However, economic uncertainty delayed Qlik's plans, and then OpenAI's November 2022 launch of ChatGPT sparked a seismic change for all data management and analytics vendors. Suddenly, with customers <a target="_blank" href="https://www.gartner.com/en/newsroom/press-releases/2026-1-15-gartner-says-worldwide-ai-spending-will-total-2-point-5-trillion-dollars-in-2026" rel="noopener">turning their attention to AI development</a> rather than traditional BI tools such as reports and dashboards, data management and analytics providers had to become enablers of AI rather than analytics. Qlik evolved to meet the needs of its customers, making AI its focal point over the past few years. Going forward, as <a href="https://www.techtarget.com/searchbusinessanalytics/news/366610199/Qlik-AutoML-update-targets-trust-with-visibility-simplicity">Qlik continues to progress</a>, it will do so under the leadership of a new CEO. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> [Capone] inherited a well-regarded BI vendor and walked out with a full-stack data, integration and AI platform that the majority of the world's largest enterprises actually depend on. The piece that impressed me most was the agentic pivot over the last 18 months. </figure> <figcaption> Mike LeoneAnalyst, Moor Insights & Strategy </figcaption> </div> </blockquote> Perhaps Qlik's most significant growth under Capone came with respect to agentic AI, according to Mike Leone, an analyst at Moor Insights & Strategy who followed Qlik throughout Capone's tenure. "[Capone] inherited a well-regarded BI vendor and walked out with a full-stack data, integration and AI platform that the majority of the world's largest enterprises actually depend on," Leone said. "The piece that impressed me most was the agentic pivot over the last 18 months, which landed faster and with more coherence than what I've seen from most of his peers in the legacy analytics space." In addition, revenue growth, successfully integrating an array of acquired companies and building a strong leadership team are among Capone's noteworthy accomplishments, he continued. Donald Farmer, founder and principal of consulting firm TreeHive Strategy and Qlik's vice president of innovation and design from 2011-16, similarly noted that Capone took over as the vendor's CEO at a difficult time and led the vendor's expansion into data integration and AI. "Mike Capone led Qlik through an unenviable period," he said. "Leveraged buyouts by private equity are a notoriously uncomfortable experience for staff, customers and partners. And with changes in market conditions, this has been a particularly complex time. … Capone has sustained Qlik as a largely cohesive company and technology stack, which must have been a great challenge." In addition to Qlik, established data and analytics vendors such as Alteryx, <a href="https://www.techtarget.com/searchbusinessanalytics/news/365535741/Sisenses-Orad-stepping-down-Katz-named-new-CEO">Sisense</a>, <a href="https://www.techtarget.com/searchbusinessanalytics/news/366571855/Snowflake-CEO-Slootman-steps-down-Ramaswamy-takes-over">Snowflake</a>, Tableau and <a href="https://www.techtarget.com/searchbusinessanalytics/news/366612092/ThoughtSpot-taps-Salesforce-exec-Karkhanis-to-be-new-CEO">ThoughtSpot</a> have all changed their CEOs over the past few years, as the market has transitioned from a focus on traditional analytics to AI as a means of managing, exploring and analyzing data, and vendors have had to shift their strategic focus. Meanwhile, others including Confluent and Informatica were acquired -- and Fivetran and DBT Labs <a href="https://www.techtarget.com/searchdatamanagement/news/366632699/Fivetran-DBT-Labs-merge-to-add-complementary-capabilities">elected to merge</a> -- as the high cost of AI development makes it more difficult for independent specialists to compete with full-featured platform vendors. "As AI and agentic technologies reshape the world, the organizations that seize this opportunity will be those that can turn trusted data into meaningful action," Capone wrote on LinkedIn. "I am confident that Qlik is strongly positioned to help customers do exactly that." One of Capone's most significant accomplishments as Qlik's CEO was to help the vendor to do just what he suggested other organizations do, which is to <a href="https://www.techtarget.com/searchbusinessanalytics/news/366623963/Qlik-evolves-to-keep-up-with-latest-AI-analytics-trends">evolve with the times</a>, according to David Menninger, an analyst at ISG Software Research. "Capone has led a two-pronged evolution of Qlik," he said. "First, he oversaw the transition from on-premises software delivery and perpetual licensing to cloud-based software-as-a-service-model subscription licensing. Second, he oversaw the extension of Qlik from an analytics-only focus to a data and analytics focus and eventually into AI as well." </section> <section class="section main-article-chapter" data-menu-title="The next CEO"> <h2 class="section-title">The next CEO</h2> While Qlik declined to specify what traits and experience it will be looking for in its next CEO, it's likely that someone committed to Qlik's role as the connective layer for AI will be the vendor's next leader, according to Leone. When Snowflake CEO Frank Slootman departed in February 2024, Sridhar Ramaswamy, who had been Snowflake's senior vice president of AI for nine months after joining Snowflake in 2023 when <a href="https://www.techtarget.com/searchdatamanagement/news/366538520/Snowflake-acquisition-of-Neeva-to-add-generative-AI">the vendor acquired Neeva</a>, was named its new leader. Since then, Snowflake, which was slower to embrace AI than some of its competitors, has <a href="https://www.techtarget.com/searchdatamanagement/news/366638535/Snowflake-launches-new-AI-tools-unveils-OpenAI-partnership">aggressively added AI capabilities</a> and built an environment for customers to develop AI tools. Similarly, among others, <a href="https://www.techtarget.com/searchbusinessanalytics/news/366640328/ThoughtSpot-domain-specific-Spotter-agents-target-AI-success">ThoughtSpot</a> and <a href="https://www.techtarget.com/searchbusinessanalytics/news/366622614/Tableau-enters-the-agentic-AI-era-with-the-launch-of-Next">Tableau</a> have focused on AI and kept up with the latest trends in AI development as their leadership has changed. "I'd want someone fluent in the full data and AI stack as a system, with depth that reaches well beyond analytics," Leone said. "Qlik's competitive ground now is the connective tissue between data integration, governance, and agentic execution, and the next leader has to be able to hold that whole picture and prioritize across it." Priorities for the new CEO should include continuing Qlik's momentum with agentic AI, upgrading governance features that engender trust in SI outputs, and improving messaging related to its data engineering and lakehouse capabilities, he continued. Menninger likewise suggested that Qlik's next CEO will need to focus on properly positioning the vendor with respect to AI given that competition includes not only other data and analytics vendors but also <a href="https://www.techtarget.com/searchenterpriseai/tip/How-to-choose-the-right-LLM-for-your-needs">large language model developers</a>. "Qlik and other analytics providers face competition from multiple fronts," he said. "In addition to guiding Qlik through the competitive landscape, the new leader will also have to navigate the financial markets, creating a path for investors to recoup their investments." Qlik, however, is not alone among analytics providers with respect to needing to find its role as <a href="https://www.techtarget.com/searchenterpriseai/feature/Businesses-gear-up-for-AI-agents-in-the-enterprise">AI continues to gain prominence</a>, Menninger continued. "The key to Qlik's future, and many other analytics vendors, is how they tackle the AI world," he said. "In much the same way that Qlik had to manage the transition from on premises to cloud, they need to successfully manage the transition from BI to AI." Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> With AI the dominant trend in data and analytics, the vendor's leader leaves after guiding it through its cloud transition and additions of data integration and AI capabilities. https://cdn.ttgtmedia.com/rms/onlineimages/code_g1304896250.jpg https://www.techtarget.com/searchbusinessanalytics/news/366642652/Qliks-Capone-departs-after-eight-years-as-CEO Thu, 30 Apr 2026 15:02:00 GMT Qlik's Capone departs after eight years as CEO With AI workloads requiring higher performance, fresher data and more complete transparency than traditional analytics workloads, vector database specialist Qdrant launched new features in Qdrant Cloud to address the demands of AI development. Accelerated indexing via <a href="https://www.techtarget.com/searchdatacenter/tip/How-do-CPU-GPU-and-DPU-differ-from-one-another">graphics processing units</a> (GPU) addresses performance by building the vector indexes that enable AI tools to retrieve relevant data substantially faster than was previously possible with Qdrant Cloud. Meanwhile, multi-AZ clusters guarantee that data is always available by replicating data across three availability zones and <a href="https://www.techtarget.com/searchsecurity/tip/What-CISOs-need-to-know-about-AI-audit-logs">audit logging</a> captures all operations performed through the Qdrant API <a href="https://www.techtarget.com/searchenterpriseai/feature/How-to-ensure-AI-transparency-explainability-and-trust">to provide transparency</a>. Given that the new features address practical issues that directly result in whether an AI tool can perform well enough to move into production, they are significant additions for Qdrant Cloud customers, according to Devin Pratt, an analyst at IDC. "This release is about making Qdrant Cloud more production-ready," he said. "It should help customers move faster, reduce operational risk and put stronger controls around AI retrieval." Vectors are numerical representations of data, including unstructured data such as text and audio, that make data searchable by agents and other automated systems so it can be discovered and used to inform AI and analytics applications. Qdrant, based in Berlin and New York City, is a vector database vendor that competes with fellow specialists such as <a href="https://www.techtarget.com/searchdatamanagement/news/366631366/Vector-database-vendor-Pinecone-eyes-future-under-new-CEO">Pinecone</a> and Weaviate as well as broad-based data management providers that offer vector database capabilities including <a href="https://www.techtarget.com/searchdatamanagement/news/366577632/Vector-search-and-storage-key-to-AWS-database-strategy">AWS</a>, Databricks and <a href="https://www.techtarget.com/searchdatamanagement/news/366583139/Oracle-adds-vector-search-capabilities-to-database-platform">Oracle</a>. <section class="section main-article-chapter" data-menu-title="Performance for production"> <h2 class="section-title">Performance for production</h2> Vector databases <a target="_blank" href="https://www.linkedin.com/pulse/why-vector-databases-now-hot-topic-abhishek-soni-fvacc/" rel="noopener">were introduced</a> in the early 2000s but remained a niche feature until OpenAI's November 2022 launch of ChatGPT marked significant improvement in generative AI (GenAI) technology and sparked <a target="_blank" href="https://www.gartner.com/en/newsroom/press-releases/2026-1-15-gartner-says-worldwide-ai-spending-will-total-2-point-5-trillion-dollars-in-2026" rel="noopener">surging interest in AI development</a>. <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> This release is about making Qdrant Cloud more production-ready. It should help customers move faster, reduce operational risk and put stronger controls around AI retrieval. </figure> <figcaption> Devin PrattAnalyst, IDC </figcaption> </div> </blockquote> AI tools such as chatbots and agents require far more relevant data to be accurate than traditional data products, including reports and dashboards. In addition, they benefit from real-time data, so the outputs they deliver include input from the most current available information. With unstructured data representing most of all data, vector databases help provide the data volume AI tools demand. In addition, they can process data at high speed to guarantee the freshness of the data being fed into AI pipelines. As a result, throughout 2023 and 2024, <a href="https://www.techtarget.com/searchdatamanagement/feature/Vector-search-now-a-critical-component-of-GenAI-development">the popularity of vector databases exploded</a>. However, most data management tools, including vector databases, were not designed for AI. They were usable when enterprises were experimenting with AI, developing pilot initiatives to learn and refine their plans for AI before putting tools into production. But vector indexing alone did not deliver high enough accuracy for most projects to move past experiments, nor did vector databases have enough power to maintain performance under the scale of AI workloads. Now, to address the different demands of AI development, numerous vendors are replacing their capabilities with those designed to better enable enterprises to move AI projects into production. For example, Databricks <a href="https://www.techtarget.com/searchdatamanagement/news/366637142/New-Databricks-tool-aims-to-up-agentic-AI-response-accuracy">launched Instructed Retriever</a> and MongoDB introduced <a href="https://www.techtarget.com/searchdatamanagement/news/366637414/MongoDB-launches-latest-Voyage-models-to-aid-AI-development">new embedding and reranking models</a> to improve the data retrieval process, GoodData and InsightSoftware -- among others -- added and improved semantic modeling and other tools that address <a href="https://www.techtarget.com/searchdatamanagement/opinion/Why-data-semantics-matters-for-context-aware-systems">the context fed to AI</a>, and vendors including Actian and Teradata have added vector databases to address AI workloads. Now, Qdrant is similarly adding capabilities designed to improve AI development with the additions of GPU-accelerated indexing, multi-AZ clusters and audit logging in Qdrant Cloud. Like Pratt, Kevin Petrie, an analyst at BARC U.S., similarly noted that the new features address the needs of AI developers and are therefore valuable additions. "These features strengthen Qdrant's position as a vector search specialist that helps AI developers build sophisticated agentic applications," he said. "Qdrant seems to be thriving in this niche." Better performance and increased transparency are especially valuable for <a href="https://www.techtarget.com/searchcloudcomputing/tip/Is-your-compute-strategy-ready-for-AI-workloads-in-the-cloud">AI workloads</a>, Petrie continued. "Faster indexing helps operationalize applications in less time, which is critical as enterprises move into full-scale production with agentic AI," he said. "Audit logging is critical because AI adopters are finally starting to take governance seriously. They need transparent, explainable workflows to comply with internal policies and external regulatory requirements." GPUs are chips that provide the compute power that systems require to carry out workloads. Traditionally, many systems were built with central processing units, but GPUs provide substantially more power and are therefore better suited for <a href="https://www.computerweekly.com/microscope/news/366634677/AI-driving-GPU-demand">the demands of AI</a>. Multi-AZ clusters assure a system's reliability by replicating data across different availability zones within a region so that if availability in one zone goes down, the system still operates in the others with no delay and no need for users to act. And audit logging provides a trail that users can follow to address <a href="https://www.techtarget.com/searchenterpriseai/feature/AI-regulation-What-businesses-need-to-know">AI's unique compliance</a> and security requirements. All were added to address the different demands AI workloads place on vector databases, according to Bastian Hofmann, head of product at Qdrant. "Vector search is running in production at scale for our enterprise customers," he said. "Multi-AZ and audit logging came directly from customer requirements -- higher uptime … and compliance visibility are essential when vector search sits on the critical path of your application." GPU-accelerated indexing was made available in Qdrant's open source database in 2025. Now, with CPUs not providing enough performance to power enterprise AI workloads at scale, Qdrant is adding power to its fully <a href="https://www.techtarget.com/searchnetworking/definition/managed-network-services">managed service</a>. "As production datasets and write volumes have grown, CPU-only indexing is no longer sufficient for certain workloads," Hofmann said. "Bringing GPU indexing to Qdrant Cloud means customers can run these heavier workloads in production without managing GPU infrastructure themselves." Beyond aiding existing Qdrant Cloud customers, the new features could help Qdrant distinguish its vector database capabilities from those of <a href="https://www.techtarget.com/searchdatamanagement/tip/Top-vector-database-options-for-similarity-searches">competing platforms</a>, according to Pratt. In particular, he noted that with high availability and audit logs becoming commonplace, the performance enabled by GPU-powered indexing -- speeding up how quickly Qdrant's database can prepare large or changing datasets for search -- could prove to be a competitive advantage. "The most differentiated capability in this release is faster indexing," Pratt said. "The availability and audit features matter, but they are quickly becoming enterprise expectations." Petrie similarly noted that the new features help Qdrant Cloud stand apart from other vector search offerings. However, vector search alone has proven insufficient for feeding AI and <a href="https://www.techtarget.com/searchenterpriseai/definition/retrieval-augmented-generation">retrieval-augmented generation</a> (RAG) pipelines. Adding more retrieval methods could therefore further differentiate Qdrant from competitors, Petrie continued. "AI and RAG workflows need broader retrieval capabilities," he said. "They need to search text via keyword matching, find table values via SQL queries, identify relationships via knowledge graphs, and so on. So …. I would recommend that Qdrant broaden its retrieval methods and source data types to remain competitive in an increasingly multimodal world." <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/how_a_vector_database_works-f.png" alt="A graphic shows how a vector database works.">Informa TechTarget </div> </section> <section class="section main-article-chapter" data-menu-title="Looking ahead"> <h2 class="section-title">Looking ahead</h2> Just as Qdrant's latest features are aimed at fueling AI workloads, the vendor's product development roadmap is focused on further improving scalability, performance and <a href="https://www.techtarget.com/searchbusinessanalytics/feature/Talend-CEO-discusses-importance-of-mining-relevant-data">search relevance</a>, according to Hofmann. In addition, Qdrant plans to add more transparency and <a href="https://www.techtarget.com/searchnetworking/tip/End-to-end-network-observability-for-AI-workloads">observability capabilities</a>, he continued. "On Qdrant Cloud, we're focused on operational simplicity -- easier cluster management, fewer manual steps -- and deeper integrations into enterprise systems so teams can plug Qdrant into their existing infrastructure without friction," Hofmann said. Pratt, meanwhile, suggested that Qdrant address how easy it is to use specialized vector search as part of a broad data ecosystem. It doesn't need to become a full-featured data platform that provides all the capabilities itself, he noted. But deeper integrations with <a href="https://www.techtarget.com/searchenterpriseai/feature/AI-agent-frameworks-A-guide-to-evaluating-agentic-platforms">AI development frameworks</a>, data warehouses, lakehouses, cloud platforms, <a href="https://www.techtarget.com/searchdatamanagement/feature/Why-enterprise-AI-initiatives-fail-without-governance">AI and data governance tools</a> and other capabilities that make up a data and AI stack would be beneficial. "One of Qdrant’s opportunities is to make specialized vector search easier to use inside the enterprise data platforms customers already rely on," Pratt said. Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management. </section> As customers look to move past experimentation and put pilots into production, the vendor's new features better prepare its platform for modern enterprise workloads. https://cdn.ttgtmedia.com/rms/onlineimages/code_g1133705410.jpg https://www.techtarget.com/searchdatamanagement/news/366642580/Qdrant-boosts-performance-reliability-to-meet-AI-needs Wed, 29 Apr 2026 14:18:00 GMT Qdrant boosts performance, reliability to meet AI needs Search Data Management Resources and Information from TechTarget 60 webmaster@techtarget.com