Infor Builds Agentic AI Into ERP: What Operations Leaders Need To Know
Infor announced a major update to its industry-focused ERP portfolio: built-for-industry AI agents that act inside the system to streamline day-to-day operations. The company also introduced Infor Leap, a cloud migration program aimed at getting on-prem customers to the cloud faster and with fewer surprises.
The goal is simple: use AI to automate routine decisions, boost margins, and reduce cycle times across projects, products, supply chain, and workforce operations. As CEO Kevin Samuelson puts it, "Everyone's talking about AI, but the challenge is actually using it for value creation."
Agentic AI, In Plain Terms
LLMs produce text. Agentic AI takes action. Infor's approach pairs an LLM with tools that execute tasks inside the ERP-like updating a schedule, adjusting a price, or creating a purchase order based on rules and current data. "It's a quantum leap in capability," says Samuelson.
Expected value lands in two buckets: process automation at scale and direct margin impact. Pricing optimization informed by demand patterns and inventory levels is a prime example.
Why Infor Thinks It Can Pull This Off
ERP is the system of record. It sees orders, inventory, schedules, and cash. Samuelson notes, "It really runs the core operations of a company." Infor concentrates on nine industries, which lets them go deeper into specific processes instead of building generic features.
The platform runs multi-tenant on Amazon Web Services. One code line, continuous updates, consistent security posture. Samuelson adds that deep AWS integration "allows us to leverage €80 billion in R&D and security work that they do."
If you need a primer on AWS's security model, this overview is helpful: AWS Security.
What's New: AI Agents For Core Operations
Infor's new agents are built for manufacturing, distribution, and service operations, with minimal training required to get value. Early focus areas include projects, products, supply chain, and workforce operations.
- Pricing and margin: Recommend and apply price changes based on demand and stock position.
- Procurement: Triage exceptions, suggest alternates, and trigger buys within policy.
- Inventory: Replenish to targets using real-time consumption and supplier performance.
- Scheduling: Fill gaps, reassign work, and adjust shifts to hit service levels.
- Supplier risk: Flag late trends and auto-escalate recovery actions.
- Project control: Detect margin leakage and propose corrective steps.
Cloud Acceleration With Infor Leap
Infor Leap is positioned to move on-prem customers to the cloud while reducing delays and overruns. The value for ops leaders: faster access to new AI capabilities, fewer version mismatches, and a cleaner security model.
Trust, Auditability, And Data Boundaries
Adoption stalls without transparency. Infor says its AI includes full audit trails-showing the steps an agent took and how it reached a conclusion. That gives teams a way to review decisions and tune policies.
Customer data used during processing is cleared in 24 hours and not used to train models. Training is based on anonymized data, use cases, and insights. On regulation, Samuelson is cautiously optimistic: it's early, and adoption will track where clear business value shows up.
If your team is assessing risk controls, this framework is a solid reference point: NIST AI Risk Management Framework.
Adoption Playbook For Operations Leaders
- Pick 1-2 high-friction workflows with measurable lag (e.g., price updates, PO exceptions, replenishment).
- Define hard metrics up front: cycle time, on-time in-full, price realization, schedule adherence, aged WIP, SLA hit rate.
- Set guardrails: thresholds for auto-approve vs. human review, escalation paths, rollback rules.
- Clean the data: master data sanity checks, supplier lead-time accuracy, policy codification.
- Pilot in a contained scope (single site or SKU family) for 60-90 days; compare agent vs. baseline.
- Operationalize: publish playbooks, schedule weekly audit reviews, and expand by value per hour saved.
People And Skills Still Matter
Infor pairs automation with expertise. "It's not magic," Samuelson says. Their teams help customers tune outcomes to real constraints. Internally, Infor reports widespread use of Amazon Q to speed software development-another sign that AI-first workflows are becoming normal.
On the customer side, you'll want three roles in place: a process owner who owns the KPI, a data steward who owns data quality, and an ops technologist who implements guardrails and monitors the agents. If you're leveling up your team for this shift, you can explore skill-based AI training paths here: AI courses by skill and automation-focused content here: Automation resources.
Where This Is Going
Samuelson's end state: "Process mining now allows us to serve up to customers how to improve processes. But ultimately, they should be improved automatically within the system without having to get people involved."
Practical takeaway: treat ERP as an active teammate. Start with one workflow. Measure hard outcomes. Raise the autonomy level only after the audit trail proves it. That's how AI in operations stops being a demo and starts compounding into margin, speed, and reliability.
Your membership also unlocks:
