Infor builds industry-specific AI agents to cut hallucinations

Infor's approach starts from the detail of what its customers actually do. Rick Rider, senior vice president of AI innovation, said the company initially focused its agents on common bottlenecks and repetitive tasks, then moved to a model of common cores shared across similar industries with lightweight configuration for micro-verticals such as chemicals or HVAC. Suresh Jayaraman's dairy example shows how far that detail runs: a liquid receipt means the agent has to understand the fat content and thickness of the milk, while a protein or red meat receipt means bringing an entire animal into the receiving process and knowing the country of origin. That kind of context is what the company argues generic agents lack.

The hiring side is part of the same story. Rider said Infor is hiring domain experts to work alongside technologists, and Jayaraman noted that finding tech people and programmers is easier than finding industry-focused people with deep industry knowledge. His reason is about who is in the room: when you go in front of the customer, you are not talking to IT people but to business people.

The bet, then, is that reliability in enterprise agents comes less from the model and more from how tightly the agent is wrapped around a specific industry's processes. How far that holds is likely to depend on whether the guardrails and scoping in the 2026.10 release can cut hallucinations in practice, and on whether Infor can keep recruiting the industry specialists its configuration work depends on.