Runlayer stepping out of stealth with an $11M seed round feels like watching three engineers walk out of a burning building carrying the blueprint for a safer city while everyone else is still arguing about smoke alarms. The round was co-led by Keith Rabois at Khosla Ventures and Felicis, and the timing lands like a well-placed bass drop. MCP has become the connective tissue of AI, adopted by OpenAI, Anthropic, Microsoft, AWS, Google, Atlassian, Asana, Stripe, Block, and just about every enterprise sprinting toward agentic automation. The problem is that MCP shipped wide but not fortified, leaving security teams trying to defend a protocol that moves faster than their playbooks. Runlayer saw the gap early and built the guardrails before the crashes started piling up.
Andrew Berman did not stumble into this. After building Nanit, then creating Vowel and selling it to Zapier, he became Director of AI and helped develop one of the first MCP servers while working directly with OpenAI and Anthropic. Tal Peretz brought his Israeli Air Force discipline, building Zapier MCP in 2 days and watching it become their fastest-growing product. Vitor Balocco spent years wrestling real LLM attack vectors in production. This founding trio did not study the problem; they lived it at scale, which explains why dozens of customers signed during just 4 months of stealth, including Gusto, Rippling, dbt Labs, Instacart, Opendoor, and Ramp. Enterprises like these do not gamble on vibes. They buy clarity, control, and the confidence that their AI agents will not turn into uninvited guests wandering through private systems.
The advisor roster adds even more gravity. David Soria Parra, co-creator of MCP at Anthropic, brings protocol-level insight you cannot replicate. Travis McPeak at Cursor brings cloud and appsec depth. Nikita Shamgunov, who built Neon and SingleStore, knows what it takes to scale high-performance infrastructure into something unstoppable. When people like that attach their names, it signals the market that this is not another security startup claiming it can outsmart every attacker on the planet. This is targeted, protocol-native defense.
Runlayer built a platform that forces AI agents to behave like responsible adults. Gateway protection acts as the checkpoint for every MCP request. Threat detection catches tool poisoning, shadowing, injections, and fake MCPs before they land. Observability exposes agent behavior that enterprises could never see before. Permissions stay synced to human users through Okta or Entra, locking down access without slowing teams down. Developers slide it into existing tools like Cursor, VS Code, Claude Desktop, GitHub Copilot, ChatGPT, and 300+ MCP clients with zero disruption.
With this $11M fueling hiring, deeper integrations, and expanded threat detection, Runlayer is positioning itself as the safety layer for the entire AI ecosystem. Watching this team move, it is clear the story is just starting to get interesting.
Startups Startup Funding Venture Capital Series Technology Innovation Tech Ecosystem Startup Ecosystem Hiring Tech Hiring
