Infor uses industry-specific AI to address agent hallucinations
SiliconANGLE Sloane Kali Faye ● Covered by 2 sources
Infor is building AI agents for specific industries, not generic ones. The pitch: fewer hallucinations, because the software knows the business rules.
Based on reporting by SiliconANGLE, Sloane Kali Faye — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Infor is betting that the way to make AI agents more reliable is to make them less general. The company is building industry-specific agents for industrial manufacturing, aerospace and defense, automotive, and food and beverage, and it’s doing that on top of its existing industry applications. Suresh Jayaraman, the company’s senior vice president of product management and development, said generic agents can work, but they don’t always give the same answer twice. That inconsistency, he said, leads to hallucinations.
The company’s answer is to lean on industry context and narrow the problem. Rick Rider, senior vice president of AI innovation, said Infor started with common bottlenecks and repetitive tasks, then added guardrails in its 2026.10 release through security and scopes. He said the company does not see “out-of-the-box AI” in its world. Instead, it works with customers to figure out what they actually want, then configures around that.
That configuration gets very specific very fast. Rider said Infor uses shared cores across similar industries, with lightweight setup for micro-verticals such as chemicals or heating, ventilation and air conditioning. Jayaraman gave food and beverage as the clearest example: a dairy receiving process needs details like fat content and milk thickness, while protein or red meat receiving needs country of origin and other extra settings. The same broad workflow, yes. The same rules, absolutely not.
Infor is also drawing a line between what agents can do alone and where humans still need to step in. Agents can create an order and check demand on their own, Jayaraman said, but moving one customer’s order to another still needs human approval. And beyond the software, the company is hiring domain experts to work with technologists, because the people on the other side of the table are business users, not IT staff. That part may matter as much as the model itself.
My take — AI-written commentary, not fact-checked reporting
This is the right instinct. Enterprise AI keeps getting tripped up by pretending every workflow is a chat prompt with a tie on; most businesses want software that knows the rules and stops making stuff up. The real moat now is domain knowledge, which is a less glamorous answer than “general intelligence” but a much more useful one.
Read more about this at: SiliconANGLE