Joining the Agentic AI Foundation as an ambassador

I’ve been accepted as an ambassador for the Agentic AI Foundation (AAIF). The news on its own isn’t very interesting; what the foundation is and why it exists is the part worth writing down.

AAIF is a Linux Foundation project. Its remit is the plumbing that agents get built on: open, vendor-neutral standards rather than any single company’s stack. The detail that made me pay attention is the membership. The platinum members are AWS, Anthropic, Google, Microsoft and OpenAI, the firms that compete hardest on the models themselves, agreeing to develop the connective tissue between agents and the world in the open. The projects the foundation hosts are ones I already build with and write about here: the Model Context Protocol, Goose, AGENTS.md, and agentgateway.

That last part is why I applied rather than the other way around. A lot of what ends up on this blog already sits directly on top of these standards: the MCP server and client walkthroughs, the agent harness build, the writing on agent experience. I didn’t pick MCP because a foundation told me to. I picked it because it was the thing that made tools portable across models, and standardising that in a neutral place is what stops it fragmenting back into five incompatible vendor protocols.

The programme is not a badge to put in a bio. The commitment is one public, project-based contribution a month: a tutorial, a talk, a walkthrough, a post, something a developer can use to actually get one of these projects working rather than just hear that it exists. The foundation files this under awareness, activation and contribution. I read activation as “stop people bouncing off the docs”, which is most of what I try to do here anyway, so it’s a commitment I can keep without pretending.

The foundation describes itself as “the neutral and open foundation built on transparency, collaboration, and standardization to advance the public interest in agentic AI innovation.” I’m wary of mission statements, but the structure behind this one is real: a Linux Foundation governance model, a technical committee, and competitors funding shared infrastructure they don’t individually control. That is a better place for the standards underneath agents to live than inside any one company’s product roadmap.

In practice this changes very little about what you’ll read here, just where it points. Expect more hands-on, code-first writing on MCP, AGENTS.md and the projects around them. If you’re building on any of this and hit a sharp edge, that’s exactly the kind of thing worth a post, so tell me what broke.