Photo: Kindel Media / Pexels
Outerlimit raises $16M to combat rogue AI agents before they cause harm
The New York startup emerged from stealth with one of the largest pre-seed rounds in cybersecurity history, betting that AI agents need guardrails at the point of action, not after the damage is done.
Outerlimit, a New York-based startup that just stepped out of stealth mode, announced a $16 million pre-seed funding round on September 22, one of the largest early-stage raises on record for a cybersecurity firm. The round was led by AlbionVC, Evolution Equity Partners, and Crane Venture Partners, with angel investors from Mistral AI and Goldman Sachs also participating.
Outerlimit’s founders, CEO Tony Pepper, Neil Larkins, and CTO Dr. Peter Vincent, argue that consequence-based controls are fundamentally inadequate for this new reality. Their pitch is straightforward: stop focusing on what happened and start preventing what’s about to happen.
The company’s platform is built around three core functions. Discover finds all AI agents and tools operating within an enterprise environment. Observe monitors their behaviors in real time. Enforce applies deterministic policies at the exact moment an agent tries to use a tool or take an action.
Outerlimit uses a distributed architecture with cryptographic enforcement. Rather than storing sensitive access credentials in one place, the system fragments those credentials and secrets throughout the ecosystem. Those fragments are only reassembled when three conditions are simultaneously verified: the identity of the agent requesting access, the policy governing what that agent is allowed to do, and the specific execution context of the action being attempted.
AI, tech, and the markets they move—in one daily briefing.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
The founding team brings credibility to the bet. Tony Pepper and Neil Larkins have established track records in the tech industry, while Dr. Peter Vincent brings expertise in theoretical neuroscience. The investor roster includes AlbionVC and Evolution Equity Partners, with angel investors connected to Mistral AI and Goldman Sachs.
Traditional identity and access management systems were built for a world where users are humans who log in, perform tasks, and log out. AI agents don’t follow that pattern. They can spawn sub-agents, chain together tool calls, and adapt their behavior based on outputs they receive, all without a human in the loop.