Glossary / Strategy & Leadership

Agent Washing

What vendors mean: they'd never admit to it. What it actually means: slapping the word agent on a chatbot, a workflow, or last year's RPA to ride the hype — the 2026 version of 'AI-powered.'

Strategy & Leadership

The term

Agent washing is rebranding existing, non-autonomous software as an “AI agent” because agents are what buyers are funding this year. The label gets applied to scripted chatbots, rules-based automations, and RPA that shipped long before anyone called them agentic. Industry analysts have flagged that a large share of self-described “agentic AI” vendors are doing exactly this.

Why it matters to you

You are the target. A procurement cycle driven by the word “agent” rather than by a defined outcome is how you end up paying agent prices for chatbot capability. The word tells you nothing about what the thing can actually decide or do on its own.

What that means in practice: Ignore the label and ask three questions. What decision does it make without a human? What is it allowed to do — and touch — when it decides? And how do you see and stop it when it’s wrong? If a vendor can’t answer those crisply, you’re looking at agent washing. Buy the capability, never the noun.

Get the next Brief

One operator. Every other Wednesday.

Plus the AI Glossary and the Failure Museum.
Real names. Real numbers. Honest analysis.