What it is
Agent washing is renaming old products (chatbots, RPA, assistants) as "AI agents" without any real agency. By analogy with greenwashing: a surface-level rebrand instead of a change in substance.
Where it came from
Gartner coined the term in 2025 against the backdrop of the "year of AI agents" hype. Once "agentic" became the buzzword, marketing slapped it on everything in sight — from autocomplete to simple scripted bots.
Why it took off (as a topic)
Out of thousands of vendors making "agentic" promises, Gartner counted only about 130 as the real thing.
Gartner projects that more than 40% of agentic AI projects will be shut down by the end of 2027 — because of cost, unclear value and weak risk control. This is the sobering-up after a year of inflated expectations — which is exactly why the topic is hot: budgets are being burned on pseudo-agents.
How to use it right now
- The real-agent check: does it take actions (not just answer)? Does it plan the steps itself? Does it adapt to something new?
- Ask for a demo on your data and your edge cases, not on a rehearsed scenario.
- Ask the vendor: what exactly is autonomous, where is a human required, how is ROI calculated?
- Start with a pilot that has a measurable metric — not with a "transformation".
What to watch out for
Most "agentic" projects in 2025 are early PoCs riding the hype, often applied where they don't belong. That doesn't mean "AI agents are a scam" — it means you need skepticism toward labels and a sober check on value before signing.