What is agent washing?

Agent washing is the practice of marketing existing products such as chatbots, AI assistants, or RPA tools as "agentic" or as "AI agents" even though they lack essential agentic capabilities. Typically they cannot pursue a goal across multiple steps on their own, choose their tools themselves, or respond to unforeseen situations. The term echoes "greenwashing" and was popularized mainly by the analyst firm Gartner.

Why does this matter?

In 2025 Gartner estimated that of the thousands of vendors describing themselves as "agentic", only about 130 offer real agentic capabilities. Anyone selecting solutions should therefore check whether a product plans and acts on its own or merely answers requests – otherwise they invest in expectations the product cannot meet.

What does this look like in practice?

A vendor advertises an "invoice agent". Testing shows that the product extracts invoices via text recognition and then follows a fixed rule set; every deviation lands in a queue. Three questions expose it: can it pursue a goal over several steps? Does it choose its own tools? What happens with a case nobody anticipated?

Not to be confused with

Not a verdict against rule-based solutions: RPA or chatbots may be exactly the right choice — the problem is only the wrong label. Nor the same as the copilot-versus-agent distinction, which describes two legitimate designs, whereas agent washing describes a misleading label.