Runlayer Raises $11M Seed to Secure AI Agents and Model Context Protocol Workflows
Runlayer, a startup focused on addressing security vulnerabilities in the rapidly adopted Model Context Protocol (MCP) used by AI agents, has emerged from stealth with a $11 million seed funding round aimed at scaling its platform and bringing robust AI security tools to enterprise customers. The company’s launch and funding milestone mark a significant step in building a trusted infrastructure layer for autonomous AI operations in business environments.
Founded in 2024 by serial entrepreneur Andrew Berman — whose previous ventures include baby‑monitor maker Nanit and the AI video conferencing tool Vowel, acquired by Zapier in 2024 — Runlayer is designed to secure AI agents as they access and execute tasks across enterprise systems without human oversight. The funding round underscores investor confidence in Runlayer’s mission to tackle the complex security gaps left by MCP’s rapid adoption in the AI ecosystem.
The $11 million seed round was led by Khosla Ventures’ Keith Rabois and Felicis, both well‑known investors in early‑stage technology companies, particularly in AI and cybersecurity. Alongside them, a network of angel advisors and backers — including security leaders and founders from adjacent technology domains — are supporting Runlayer’s growth as it brings its AI security platform to market.
Runlayer’s platform focuses on the Model Context Protocol, an open‑source standard originally released by a team at Anthropic that has since become widely adopted by major AI model makers, including OpenAI, Microsoft, AWS, and Google. MCP enables AI agents to autonomously access and manipulate data across systems, which promises to unlock powerful automation workflows for enterprises. However, the protocol was not built with comprehensive security mechanisms, leaving many implementations open to vulnerabilities such as data exfiltration or unauthorized operations.
Runlayer’s solution aims to provide an all‑in‑one security framework that combines secure gateways, real‑time threat detection, observability across agent activity, granular permission models mapped to existing identity systems like Okta and Entra, and tools that allow IT and security teams to build custom enterprise integrations. By offering a unified console that tracks every MCP request and aligns AI agent permissions with human identity policies, Runlayer intends to help organisations mitigate risk while enabling the productivity benefits of autonomous AI workflows.
Since operating in stealth for four months prior to its public launch, Runlayer has already signed dozens of enterprise customers, including eight unicorn or publicly traded companies such as Gusto, dbt Labs, Instacart, and Opendoor, reflecting strong early market demand for security tools tailored to MCP environments. Additionally, the company has brought on board David Soria Parra, the lead creator of the MCP specification, as an angel investor and advisor, underscoring the strategic alignment between the protocol’s evolution and Runlayer’s security vision.
Andrew Berman, Runlayer’s CEO, emphasised that while AI adoption has surged across industries, the infrastructure supporting autonomous agents often lacks the visibility and safeguards required for enterprise deployment. Drawing on his experience building early MCP servers and leading AI efforts at Zapier, Berman said Runlayer was founded specifically to address “blind spots” in protocol security and help companies confidently embrace AI automation without exposing sensitive systems or data.
The seed funding will be used to continue product development, scale the engineering team, and accelerate Go‑To‑Market efforts as Runlayer rolls out its general availability platform. The company is positioning itself as a critical component in the secure adoption of AI at scale, offering organisations the tools they need to manage risk without compromising on the productivity gains promised by autonomous AI agents.
As enterprises increasingly integrate AI into core workflows — from data retrieval and analysis to automated business process execution — the need for robust security layers that can keep pace with agentic activity is becoming more apparent. Runlayer’s emergence from stealth with significant backing from well‑known investors signals strong belief in its potential to shape the secure future of AI infrastructure, addressing real pain points in how organisations deploy and govern intelligent systems.
