NewsSep 22, 2026, 13:37 UTC

Outerlimit exits stealth with $16M to secure AI agents

What's the deal? Outerlimit has emerged from stealth with $16 million in pre-seed funding to build a security and authorization layer for agentic AI. The round drew AlbionVC, Evolution Equity Partners, and Crane Venture Partners, among others, and ranks as one of the largest pre-seed rounds in cybersecurity.

Who's behind it? The New York startup was founded by Tony Pepper and Neil Larkins, who sold Egress Software to KnowBe4 in 2024, and Dr Peter Vincent, a theoretical neuroscientist with a PhD from UCL. Pepper serves as chief executive officer; Vincent as chief technology officer.

What's the endgame? Outerlimit wants to extend the Zero Trust model to what it calls the agent action layer. Rather than storing credentials in one place, it fragments keys across the agent ecosystem and reconstructs them only when a tool is invoked and the identity, policy, and context are verified.

Why now? As enterprises hand agents access to systems, data, and APIs, Outerlimit argues that existing identity, access, and governance tools cannot provide the observability and control needed to scale safely. Its product guides customers from discovery of agents and shadow AI through observability to enforcement of controls on every action.

What could go wrong? Vincent frames the core problem bluntly: unlike people, agents "don't operate within these human constructs" of loyalty and trust, and can shift behaviour at machine speed based on what they read.

The signal: The size of the round reflects rising investor appetite for tools that let companies deploy AI agents without ceding control. As Pepper put it, "Blocking adoption isn't a strategy," since it only pushes teams toward unsanctioned AI outside enterprise oversight.