**Runlayer Emerges from Stealth to Tackle AI Security with $11 Million in Funding**
On Monday, a new security startup named Runlayer emerged from stealth mode, announcing an $11 million seed funding round led by Keith Rabois of Khosla Ventures and Felicis. The company, founded by serial entrepreneur Andrew Berman, aims to address growing security concerns around the Model Context Protocol (MCP), a rapidly adopted open-source standard that allows artificial intelligence (AI) agents to independently access, modify, and execute business data and processes across organizations.
**Background: The Rise of MCP and Its Security Gaps**
The Model Context Protocol was introduced in November 2024 by David Soria Parra and his team at Anthropic. MCP quickly became the backbone for enabling AI agents to interact with enterprise systems, giving them the ability to retrieve, manipulate, and act on data without direct human supervision. This technology has become widely supported by major AI players, including OpenAI, Microsoft, AWS, and Google, and is implemented in thousands of enterprises — from tech companies like Atlassian, Asana, Stripe, and Block to organizations in banking and consumer goods.
As Andrew Berman, Runlayer’s CEO, told TechCrunch, “Everyone talks about AI, but AI is really only as useful as the tools and the resources it has access to.” MCP serves as the critical connective tissue between powerful AI models and the data or applications they act upon. However, in its rush to widespread adoption, MCP has revealed significant security vulnerabilities. The protocol itself provides minimal inherent security features, leaving organizations that deploy MCP servers exposed to a variety of attacks.
**Security Incidents Highlight Urgent Need**
Recent incidents have underscored the severity of these risks. In May, Invariant Labs researchers uncovered a prompt injection vulnerability in MCP servers that allowed unauthorized access to private GitHub repositories. Similarly, in June, Asana found and patched a flaw in its MCP server that could have exposed customer data. These are not isolated cases; a growing list of vulnerabilities has been discovered in various MCP server setups, highlighting a systemic problem that needs urgent attention.
In response, a wave of MCP security products has emerged, developed by industry heavyweights like Cloudflare, Docker, and Wiz, as well as numerous startups. The most common approach has been to create “gateway” products — layers that authenticate and regulate what AI agents can access in an organization’s digital environment.
**Runlayer’s Approach: Comprehensive Security for MCP**
Runlayer seeks to differentiate itself in this crowded field by offering an all-in-one security platform for MCP environments. Rather than focusing solely on access control, Runlayer combines a secure gateway with advanced features such as:
- **Threat Detection:** Analyzes every MCP request in real time to identify suspicious activity or potential breaches. - **Observability:** Monitors all agent-driven activity across MCP servers, providing IT teams with comprehensive oversight and audit trails. - **Enterprise Development Tools:** Allows organizations