Agentic industrial AI and the indispensable connected worker
In this podcast, Innovapptive CEO Sundeep Ravande says an emerging era of execution-driven industrial AI will require the judgment and data-gathering skills of frontline workers.
Industrial AI has generally been focused on providing predictive tools and analytics for production managers, plant supervisors and field workers to run on their fixed workstations and mobile tablets. It hasn't done much to directly automate industrial processes themselves, though that is rapidly changing as agentic AI gives robots more autonomy and begins to make decisions on behalf of humans.
In this episode of Enterprise Apps Unpacked, Sundeep Ravande, co-founder and CEO of Houston-based Innovapptive, says a major shift is underway from standalone AI copilots and analytics to execution-driven industrial AI that embeds intelligence in critical workflows and coordinates workers' activities.
Quick teamwork is essential
According to Ravande, the trend toward autonomous industrial AI has special relevance to the frontline workforce: the machine operators, maintenance technicians, safety inspectors and warehouse workers who operate, maintain or repair heavy machinery in asset-intensive industries like oil & gas, mining and manufacturing.
Besides providing more accurate predictions of equipment failure and recommending effective fixes, embedded agentic AI also closes a perceived gap between back-end business systems and the software used by frontline workers. Innovapptive says its connected worker platform extends back-end ERP enterprise asset management and supply chain functions with mobile apps for work orders, instructions, scheduling, inventory management and other tasks commonly performed by frontline workers.
"We truly believe where value gets created is to convert the insight to action on the frontline," Ravande said. "Insights by themselves don't really create value."
As an example, he explained why it's important to integrate mobile and operational technology with back-office systems in a massive chemical plant with 1,000 frontline workers. When an anomaly is detected, four personas must spring into action and coordinate their activities. There are operators responsible for keeping machines running, maintenance personnel, workers in stores and warehouses who manage and ship parts, and safety officers who ensure repairs are done safely.
The human in the loop is essential in this scheme. For one thing, AI could recommend an action that doesn't sufficiently account for the safety risks. "You don't want something to be predicted, and the human just blindly went and acted on it," Ravande said. "In the industrial environment, it's critical that real-world experience is combined with intelligence, because sometimes intelligence can be off."
The human element also helps minimize operational risk. For example, the knowledge graphs agents use to understand context might be inaccurate because information was entered after the fact or is paper-based. "You need to build frontline execution excellence where the data is recorded in your system of record as work gets done, because without that, you risk having the wrong context," he said.
AI's agentic era provides a tremendous opportunity for humans and agents to work together on the frontline to convert insights into action, Ravande said. He predicted today's semi-autonomous plants could evolve into fully autonomous ones in five to 10 years.
Other topics discussed in the podcast include the following:
- How the Innovapptive platform integrates with back-end systems, such as SAP Enterprise Asset Management.
- The connectivity standards used to bring real-time field data into the system.
- Case studies from the chemical industry.
- The importance of multi-agent orchestration.
David Essex is an industry editor who creates in-depth content on enterprise applications, emerging technology and market trends for several Informa TechTarget websites.
