EXPLORINEER
en

Archive — US — techcrunch

MCP AI agent security startup Runlayer launches with 8 unicorns, $11M fromKhosla’s Keith Rabois and Felicis

Runlayer, a new AI security startup, launched with $11M in funding to secure the Model Context Protocol (MCP), which has become a standard for AI agents. Founded by Andrew Berman, the company boasts clients including eight unicorns and aims to address significant security vulnerabilities associated with MCP implementations. Notably, Runlayer plans to offer a comprehensive security solution integrating various protective features.

Key points

  • Runlayer launched with $11M in seed funding to address security vulnerabilities in the Model Context Protocol (MCP) used by AI agents.
  • Andrew Berman, the founder, previously created Nanit and an AI video conferencing tool, Vowel, which sold to Zapier in 2024.
  • Runlayer's client base includes eight unicorns and public companies such as Gusto, Instacart, and Ramp.
  • The MCP protocol allows AI agents to interact with data autonomously, but lacks inherent security, making implementations vulnerable to attacks.
  • Security issues have prompted the development of numerous MCP security products, including gateways and access controls.
  • Runlayer aims to distinguish itself by offering an all-in-one security solution with threat detection, observability, and customizable permissions.

Why it matters

  • The emergence of Runlayer coincides with a broader trend of increasing focus on AI security, particularly as AI technologies become more widespread.
  • MCP vulnerabilities have been documented in major applications, highlighting critical gaps in current security practices and the need for enhanced solutions.
  • The trend towards integrating AI capabilities into business processes underscores the necessity of robust security frameworks to protect sensitive data.

Runlayer’s entry into the market emphasizes the critical need for security in AI implementations. By targeting MCP vulnerabilities with a comprehensive solution, it addresses a growing demand, potentially impacting the future landscape of AI security in enterprises.