Agentic AI Foundation

What the foundation is

The Agentic AI Foundation (AAIF) is a project of the Linux Foundation. It launched in December 2025 with three projects under its care ^[1]. Its stated role is to provide neutral, vendor-neutral governance for agent interoperability, security and accountability ^[1].

That framing addresses a practical problem. AI agents — software that takes actions on a user's behalf, calling tools and other services — are increasingly built by different companies and expected to work together. Shared specifications make that possible, but only if the specifications are not controlled by a single vendor. A foundation structure puts the specifications under rules that any participant can rely on, rather than under one company's roadmap.

What it hosts

The foundation's portfolio includes the Agent2Agent protocol, Agent Router, agentgateway, AGENTS.md, goose and the Model Context Protocol ^[1]. These projects cover different parts of building and connecting agents: how agents talk to one another, how requests are routed to them, and conventions and tooling that developers use around them. The point of gathering them in one place is that they are maintained as shared infrastructure rather than as separate releases from separate companies, consistent with the foundation's stated focus on interoperability, security and accountability ^[1].

The portfolio has grown quickly. By 22 October 2026 the foundation hosted 10 projects, up from the three launch projects ^[1].

Membership

Membership is organised into Gold, Silver and Associate tiers ^[2]. The foundation reached 305 member organisations in under a year ^[2], and 60 new member organisations joined in the quarter ending October 2026 ^[2].

Among the new Gold members were Ant Group, the Gates Foundation, MongoDB, Nutanix, Royal Bank of Canada, Vercel and Vultr ^[2]. The spread of that list — a payments and technology group, a philanthropic foundation, a database company, an infrastructure vendor, a bank, and cloud and hosting providers — indicates that interest in agent standards is not confined to AI model developers.

People and events

Mazin Gilbert is the foundation's Executive Director ^[2]. Manik Surtani is its CTO and co-founder ^[1].

The foundation makes its announcements at the AGNTCon + MCPCon North America events it runs ^[1], which places its news cycle alongside a conference programme rather than only in written releases.

Where it sits

Agentic AI began as a series of individual product releases from individual companies. The Agentic AI Foundation represents a different arrangement: the connective layer — the protocols and conventions that let agents from different vendors interoperate — is placed under neutral governance, with a tiered membership structure and a staffed leadership, and the project count and membership numbers are reported as measures of how much of the field has signed on ^[1][2]. Whether that arrangement holds as the technology changes is the open question; what the foundation has established is the structure itself, and a membership base that grew from three launch projects and an initial cohort to ten projects and more than 300 organisations within roughly a year ^[1][2].