The Linux Foundation unveiled a new initiative Monday aimed at standardizing the rapidly evolving field of autonomous artificial intelligence (AI) systems, bringing together major tech companies and three key open-source projects under one umbrella.
The Agentic AI Foundation arrives as the industry shifts from conversational AI chatbots toward more autonomous systems capable of making decisions and coordinating tasks independently. The foundation will house Anthropic’s Model Context Protocol (MCP), Block’s goose framework, and OpenAI’s AGENTS.md standard.
“We are seeing AI enter a new phase, as conversational systems shift to autonomous agents that can work together,” said Jim Zemlin, the Linux Foundation’s executive director. The foundation aims to provide neutral governance as these tools become critical infrastructure.
The centerpiece is MCP, which Anthropic released a year ago. The protocol has rapidly gained traction as a standard way to connect AI systems with external tools and data sources. More than 10,000 servers now use MCP, and it has been integrated into popular platforms including Claude, ChatGPT, Microsoft Corp.’s Copilot, and Google’s Gemini.
Mike Krieger, Anthropic’s chief product officer, said the company initially developed MCP to solve internal problems before open-sourcing it last November. Donating it to the Linux Foundation ensures it “stays open, neutral, and community-driven,” he said.
Block contributed goose, an open-source framework released earlier this year that combines language models with standardized integrations for building AI agent workflows. Manik Surtani, Block’s head of open source, framed the foundation as taking “a stand for openness” in AI development.
OpenAI donated AGENTS.md, a markdown-based standard launched in August that helps AI coding assistants understand project-specific requirements across different repositories. The standard has already been adopted by more than 60,000 open-source projects.
The foundation counts eight platinum members: Amazon Web Services, Anthropic, Block, Bloomberg, Cloudflare Inc., Google, Microsoft, and OpenAI. Nearly two dozen companies have joined at the gold level, including IBM Corp., Cisco Systems Inc., Salesforce Inc., and Snowflake Inc., while another two dozen signed on as silver members.
The formation reflects growing industry recognition that autonomous AI systems will require shared standards and governance structures as they become more capable and widespread across enterprise applications.
Ryan Galgon, a product director at Pinterest, said companies increasingly are hewing to the open-source model and choosing to adopt multiple smaller models that work with one another — rather than massive computing models that require significant computing at a steep cost and riskier safety.
“The AI industry has reached the point where scalable agents require common protocols, predictable behavior models, and a governance home that evolves at the pace of real adoption,” says Mitch Ashley, vice president/practice lead, DevOps and AppDev, at the Futurum Group. “Moving MCP, AGENTS.md, and other emerging standards into the Agentic AI Foundation’s neutral stewardship creates the stability vendors and practitioners have been asking for.”
“This announcement shows that open, interoperable agentic infrastructure will define the next phase of AI development,” he said. “Enterprise teams want agents they can trust across environments, and vendors need a place to align on the rules that make that possible.”
