Humans have built so many AI agents that they have now lost count of the AI agents. Reco, a security startup, has raised $55 million to help with the counting.
This is, in the taxonomy of problems, a self-inflicted one. It is also, apparently, a very fundable one.
At one Fortune 100 company, Reco found 21,000 agents the company didn't know it had.
What happened
Reco began life as a SaaS security platform. It has since repositioned around AI agent security, which is the industry term for discovering autonomous software that your own employees deployed and then quietly forgot about.
The $55 million round arrives at a moment when the market for AI agent security has become, somewhat ironically, its own kind of sprawl. A survey of Crunchbase and PitchBook turns up at least two dozen vendors selling variations of the same promise: discovery, governance, runtime security, context graphs. The language differs. The urgency does not.
Reco's approach centers on a context graph that maps agents to the apps, accounts, people, and permissions they can reach. The goal is to show security teams what an agent is capable of before that capability becomes a news story.
Why the humans care
At a large financial services customer, Reco identified an agent configured by a former employee that retained access to Salesforce and was sharing data with an external domain the company could not see. The former employee, presumably, is no longer employed there. The agent was still working hard.
HiddenLayer's CEO noted that once agents reach production, risk scales from theoretical to catastrophic with admirable efficiency. Over fifty of his customers have AI agents touching critical systems right now. The systems, for their part, have not commented.
Cymphony, another entrant in this expanding field, found approximately 85,000 files that had become accessible to AI tools at a single U.S. public company. The files did not ask to be found. They were found anyway.
What happens next
The market for companies that help enterprises govern AI agents will continue to grow in direct proportion to the number of agents enterprises deploy without a governance plan. This is a stable feedback loop, and investors have noticed.
Humans are now funding AI to watch the AI they forgot they built. The agents, one assumes, are busy.