Infor Targets Workflow Hallucinations With Industry-Specific AI Agents

Enterprise software provider Infor (US) LLC is developing industry-specific artificial intelligence agents to automate operational workflows while curbing model hallucinations, executives said in an interview broadcast by SiliconANGLE's theCUBE during Infor Velocity Week.

Rather than relying on generic out-of-the-box models, Infor is embedding automated agents directly into its existing vertical applications across manufacturing, aerospace and defense, automotive, and food and beverage sectors. Infor Senior Vice President of Product Management and Development Suresh Jayaraman noted that general-purpose agents struggle with determinism in corporate environments, leading to higher rates of hallucination when handling nuanced tasks.

To maintain control, Infor's 2026.10 release introduced guardrails based on security rules and scoped permissions, according to Rick Rider, senior vice president of AI innovation. The company's architectural model uses shared functional cores across broad industries, supplemented by lightweight micro-vertical configurations for subsectors such as chemicals or heating, ventilation and air conditioning.

Process requirements can diverge significantly even within a single industry, Jayaraman pointed out. In food and beverage operations, for example, receiving dairy involves testing liquid metrics such as fat content and thickness, whereas meat processing intake requires tracking country of origin and physical livestock attributes.

While the agents can autonomously generate purchase orders and evaluate demand, Infor retains human-in-the-loop controls for sensitive decisions, such as shifting order allocations between customers. To support development, the company is pairing software engineers with industry domain specialists who understand operational business requirements.