Infor Deploys Micro-Vertical AI Agents to Eliminate Enterprise Workflow Hallucinations
Infor is building industry-tailored AI agents for manufacturing, aerospace, and food sectors to reduce hallucinations and automate complex enterprise workflows with human oversight.
Enterprise software provider Infor (US) LLC is taking a domain-focused approach to artificial intelligence by developing specialized AI agents tailored to specific sectors. Speaking during an interview on theCUBE at Infor Velocity Week with hosts Christophe Bertrand and Alison Kosik, company executives explained how embedding deep operational context directly into enterprise workflows helps reduce AI hallucinations and provides reliable, deterministic outputs for complex business processes. The initiative targets several key verticals, including industrial manufacturing, aerospace and defense, automotive, and food and beverage.
Addressing hallucinations through micro-vertical context
Generic AI models often struggle to deliver consistent, deterministic outcomes in specialized business environments. According to Suresh Jayaraman, senior vice president of product management and development at Infor, relying solely on broad AI systems frequently results in hallucinations and unpredictable answers. To overcome this limitation, Infor is building its AI agents directly on top of its existing suite of enterprise applications.
Rather than creating completely distinct software architectures for every single niche, Infor uses a shared core engine across related sectors, combined with light configuration layers for specific micro-verticals such as chemicals or heating, ventilation, and air conditioning (HVAC). Rick Rider, senior vice president of AI innovation at Infor, noted that off-the-shelf AI fails to meet enterprise requirements because every customer operates differently. Consequently, Infor initially directed its agents at resolving repetitive operational tasks and common workflow bottlenecks. The upcoming 2026.10 software release further reinforces these capabilities by introducing safety guardrails through explicit scopes and security controls.
The practical impact of this micro-vertical customization becomes clear in sectors like food and beverage processing. Jayaraman highlighted that receiving raw materials requires entirely different data inputs depending on the goods involved. For instance, receiving dairy products demands tracking specific metrics like fat content and liquid viscosity. Conversely, processing livestock or red meat requires recording details such as country of origin and specialized handling rules. By training agents on these specific micro-vertical nuances, systems can manage domain-specific rules accurately without generating erroneous data.
Balancing autonomous execution with human oversight
While Infor's industry-tailored agents are capable of handling multi-step tasks independently, human intervention remains a crucial component of the architecture. Jayaraman explained that agents can autonomously construct purchase orders and evaluate demand trends across systems. However, sensitive operational decisions—such as reallocating an existing order from one customer to another—still require explicit human approval to ensure accountability.
Building these systems also requires a shift in workforce talent. Rider and Jayaraman noted that Infor is actively hiring industry domain experts to work alongside software engineers. According to Jayaraman, finding skilled programmers is relatively straightforward compared to sourcing experts with deep, practical knowledge of niche industrial processes. Because sales and implementation discussions occur with business leaders rather than IT departments, domain expertise is essential for designing effective agent workflows.
What it means for developers
For software engineers and AI developers, Infor's strategy highlights the growing importance of combining base AI capabilities with deep domain-specific context and strict guardrails. While general-purpose LLMs excel at processing natural language, enterprise-grade deployment demands tight integration with underlying business logic, deterministic constraints, and domain-tailored data structures.
Developers building software across diverse industries often need to experiment with multiple foundational architectures to determine which models handle specific structured tasks best. Through services like https://apixoai.online, developers can access top AI models cheaply through one API key, making it easier to evaluate different model behaviors and test guardrails across complex workflow scenarios.
Ultimately, Infor's shift toward micro-vertical AI agents demonstrates that reducing hallucinations in enterprise automation depends heavily on structured scope definition, targeted security parameters, and human-in-the-loop fallback mechanisms. As automated agents take over routine inventory, demand forecasting, and order creation tasks, developers must focus on setting precise boundary conditions and ensuring domain experts contribute directly to prompt and workflow engineering.
Source: Infor uses industry-specific AI to address agent hallucinations — SiliconANGLE AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
