newsMay 20, 2026 2 min read

Gartner Warns of 'Agent Washing' in Supply Chain Planning Software

Gartner used its Supply Chain Symposium press cycle to call out the gap between vendor agent-AI claims and what actually changes a decision. The recommendation to SCP leaders is unglamorous: build the operational discipline, data, and governance to make autonomy land — and stop buying the badge.

Glowing orange warning triangle centered on a dark field of hollow agent-card outlines
CrateOS monitoring note: agent washing is a clean term for a real problem. A chatbot wrapper over a forecast is not an agent. A workflow that needs a human to approve every step is not autonomy. The buying decision worth taking time over is which capabilities materially change a decision and which just rebrand last quarter's UI.

On May 20 at the Supply Chain Symposium/Xpo in Barcelona, Gartner published a warning that the supply chain planning software market is increasingly subject to "agent washing" — vendor claims of agentic AI that, on inspection, do not materially change how decisions get made. Jan Snoeckx, Senior Director Analyst in Gartner's Supply Chain practice, framed the operator job clearly: "SCP leaders should prepare for an agentic AI future, but they need to separate meaningful capability from market noise. The priority today is not full autonomy, but building the operational discipline, architectural flexibility and decision frameworks that allow agentic AI to scale." Gartner forecasts that by 2030, cross-functional multi-agent systems will execute 35% of enterprise workflows with minimal human approvals — up from 3% in 2025 — but most of what's shipping today still improves user experience (query interpretation, recommendations, conversational support) rather than decision quality.

For operators, this is permission to slow down a buying cycle that the vendor calendar wants to accelerate. The right diagnostic on any "agentic" pitch is concrete: which decision is the agent allowed to make end-to-end, what is the rollback path, where does the data come from, and how is the outcome measured. If a demo cannot answer those four questions on a real workflow — receiving, slotting, exception triage, replenishment — it is a chat layer, not an agent. The work that pays off in this cycle is the unglamorous prerequisite Gartner names: a clean operational data graph, role boundaries that allow autonomy at the right grain, and a decision framework that says explicitly what an agent owns and what stays human. Vendors will catch up. Operators that skip the prerequisite will be running washed agents on top of broken processes.

Your exceptions are costing you.Let's talk.