Bairo GonzalezLeandro · Martinez
Start your guided journey The story

In the world · Artificial intelligence · 9 December 2025

MCP and the Agentic AI Foundation: the open agent standard

In December 2025, Anthropic, OpenAI and Block brought MCP, AGENTS.md and goose to the Linux Foundation. Why an open agent standard changes the game.

bairogonzalez.com team, drawing on Bairo's story · Published

On December 9, 2025, the Linux Foundation created the Agentic AI Foundation (AAIF), a neutral home for the open standards of AI agents. Anthropic donated the Model Context Protocol (MCP), OpenAI donated AGENTS.md and Block donated goose. Google, Microsoft, Amazon Web Services, Bloomberg and Cloudflare joined as members. In practice, competitors agreed to speak the same language to connect agents to tools and data.

What happened

According to the Linux Foundation announcement, the foundation was born with three projects:

  • MCP, the protocol that connects AI models and agents to tools, data and applications. Anthropic opened it up in November 2024. When the foundation was created, more than 10,000 MCP servers had already been published, used in products such as Claude, ChatGPT, Gemini, Microsoft Copilot, Cursor and VS Code.
  • AGENTS.md, a simple file, created by OpenAI, that tells a coding agent how to work inside a project. More than 60,000 open-source projects had already adopted it.
  • goose, from Block, a local, open agent framework built on MCP.

OpenAI presented itself as a co-founder of the foundation. To TechCrunch, OpenAI engineer Nick Cooper described the protocols as a common language that lets different agents and systems work together without each developer reinventing the integrations.

Why it matters

Before a standard, every link between an AI and a system was custom-built: one bridge to email, another to the spreadsheet, another to the database. Each company repeated the work, and every change of AI provider meant redoing everything.

An open standard, with neutral governance, changes three things:

  • Cost. The integration is built once and serves several models.
  • Freedom. A small business is not locked into a single AI provider.
  • Security. With a common standard, good practices for security and permissions can be discussed and improved by everyone, instead of each party inventing its own.

It is the same movement that made the internet work: open protocols that no one owns alone.

In Bairo's view

For Bairo Leandro Gonzalez Martinez, a common agent standard is the condition for AI to truly be an extension of the human within reach of anyone, and not a privilege of those who can pay for expensive integrations. He says he ran into that wall early: AI consumes tokens, tokens cost money, and those who need it most are usually those who can least afford it. In his reading, an open standard lowers that wall.

The idea speaks to the design of CLAIN, the agent orchestrator he founded. CLAIN works as a team: specialist agents, each in a niche, and an orchestrator that brings the result together and hands the decision back to a person. For many specialists to talk to one another and to outside systems, they need a common language. The goal of 142 niche AIs in 142 days, announced by Bairo on October 1, 2026, is a goal, and an open standard is what makes that kind of multiplication viable.

Bairo likes to recall that the rule running through every cell is just one: the machine organizes, the person decides. An open standard does not change that rule. It only makes it apply to more people.

Where this meets the ecosystem

  • CLAIN is under development and in internal operation, coordinating cells of agents that serve the ecosystem's ventures.
  • BTZ Chain proposes to be, in the founder's words, an orchestrator that talks to other networks: the same interoperability logic, applied to records.
  • qrqbits works on file protection with post-quantum cryptography, a safeguard that grows in importance as agents start moving data between systems.

Each venture is a distinct entity, with its own registration. What links them is the choice to build on standards that others can use too.

Sources

Read also

More on this topic

In the world: Artificial intelligence