Agent Washing

What does agent washing mean?

A vendor used to call its product a customer-support bot. Now the same product is described as an autonomous agent that can investigate, decide and act.

The software may not have changed much. The label has.

Agent washing is the criticism that a company is calling something “agentic” without adding the autonomy the word suggests. Gartner has applied the term to vendors marketing AI assistants, robotic process automation and chatbots as agents despite limited agent-like capability.

What makes something agent washing?

Every vendor at that conference called their chatbot an “agent.” Total agent washing.

Example written for this dictionary.

The accusation starts with a practical question: what can the system do on its own? Can it observe what is happening, decide what to do next, take actions and adjust its plan, or does a person still approve every step?

There is no universally accepted autonomy test. Still, simply renaming a fixed workflow or chat interface is the clearest case. Gartner estimated in June 2025 that only about 130 of the thousands of vendors describing themselves as agentic met its standard for meaningful agentic capability.

Why the label matters

“Agent” has become a powerful word in enterprise AI. It can affect budgets, buying decisions and expectations about how much work a system can actually handle. That makes the difference between automation and autonomy more than a vocabulary argument.

Gartner also forecast that more than 40 percent of agentic AI projects would be canceled by the end of 2027, citing escalating costs, unclear business value and inadequate risk controls. That is a prediction, not a measured failure rate. Agent washing adds another source of confusion: a system can sound autonomous long before it actually is.

Like AI wrapper, the phrase is usually dismissive. The difference is the complaint: “wrapper” questions how much value a product adds, while “agent washing” questions whether its claimed autonomy is real.

Where the phrase came from

The phrase follows older technology labels such as cloud washing and AI washing, both used when companies attach a fashionable technical term to something that does not fully deserve it.

Gartner helped push agent washing into wider business and technology discussion as agentic AI became one of the industry's biggest marketing themes in 2024 and 2025.

Source & further reading

Gartner: Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027