I spent years building identity and security products at Okta and Auth0. I watched the industry obsess over enterprise readiness while AI agents started doing real work underneath. Browsing the web, calling APIs, handling support, managing infrastructure.

The question that kept bugging me: how do you even know when one of these agents gets compromised?

Prompt injection isn't theoretical. It's happening today. Agents are being manipulated into leaking data, executing unauthorized actions, and trusting inputs they shouldn't. The tools that exist to stop this were built for human operators, not autonomous systems making decisions in milliseconds.

So I built Decoy. No fear tactics. No hidden pricing. Just honest, reliable security for a new kind of software.

I treat customers the way I want to be treated. Public pricing always. No hidden agendas. Simple and reliable.