9 Dec 20258 minute read
9 Dec 20258 minute read
A slew of high-profile tech companies have joined forces for a new foundation that could shape the future of agentic artificial intelligence.
The Linux Foundation today announced the all-new Agentic Artificial Intelligence Foundation (AAIF), kicking off with a trio of major contributions from OpenAI, Anthropic, and Block, and backed by founding members that include AWS, Bloomberg, Google, and Microsoft. The AAIF’s goal, ultimately, is to establish open standards and shared infrastructure for a fast-emerging class of AI systems capable of complex coordination and autonomous decision-making.
“For AI agents to reach their full potential, developers and enterprises need trustworthy infrastructure and accessible tools to build on,” said Nick Cooper, member of the technical staff at OpenAI.
The foundation is anchored by three cornerstone contributions: Anthropic’s Model Context Protocol (MCP), OpenAI’s AGENTS.md specification, and Block’s local-first agent framework, Goose.
MCP, for the uninitiated, is an open source specification introduced by Anthropic last year to give AI systems a common language for fetching data, invoking tools, and sharing state — essential plumbing that connects AI agents to the outside world. So if, for example, a developer wants an AI agent to pull recent sales records from an internal database, retrieve customer-support transcripts from a CRM, and then draft a summary report in Google Docs or post an update in Slack, MCP defines the “pipes” that make those hand-offs flow more freely.
In the year since its launch, thousands of MCP servers have been published, with major platforms embracing the standard with open arms — from Google integrating MCP-powered tooling directly into Chrome DevTools, to GitHub standing up an MCP registry to help developers discover trusted, interoperable extensions. The result is a fast-growing ecosystem in which agents, tools, and applications can plug into each other with less friction.
Arguably, one of the more surprising drivers of this momentum is OpenAI. Despite being Anthropic’s fiercest competitor, OpenAI’s Sam Altman announced back in March that it would adopt MCP across its products, while also contributing engineers to help maintain the protocol.
Separately, OpenAI is the driving force behind AGENTS.md, also emerging as a widely adopted open standard: a lightweight, Markdown-based guide that consolidates setup commands, build steps, coding conventions, and testing workflows into one file that every supported agent knows to check first.
“By co-founding the AAIF and donating AGENTS.md, we’re helping establish open, transparent practices that make AI agent development more predictable, interoperable, and safe,” said Cooper, who serves as core maintainer of the MCP project.
Elsewhere, Block – the Jack Dorsey-founded company behind Square and Cash App – open-sourced Goose back in January, positioning the framework as a more controlled, predictable environment for building and executing AI agent workflows.
The formation of the AAIF is somewhat telling, a tacit acknowledgement that agentic AI can’t reach its potential without collaboration and shared governance. Instead of each firm building its own siloed infrastructure, they’re aligning on common standards — a move, they argue, will accelerate the entire ecosystem and keep the foundations of autonomous AI from being owned by any one player.
“We're at a critical moment for AI – the technology that will define the next decade, that promises to be the biggest engine of economic growth since the Internet, can either remain closed and proprietary for the benefit of few, or be driven by open standards, open protocols, and open access for the benefit of all,” said Block’s head of open source, Manik Surtani.
Notably, Surtani says one of the AAIF’s priorities will be defining interoperability standards for AI coding agents, ensuring that tools from different vendors can operate predictably across the same codebases.
“MCP and AGENTS.md are two of the most important interoperability standards we have today that apply to coding agents,” Surtani told AI Native Dev over email. “MCP defines how agents use tools and data. AGENTS.md gives agents consistent guidance about working with code repositories. An agent from any vendor should be able to use the same MCP servers and respect the same guidance set out in AGENTS.md. The AAIF's job is to evolve these standards as things grow, keep them open, and serve as the home for other specs and protocols that evolve into industry standards. And to provide reference implementations of these specs and protocols, such as Goose.”
While Anthropic, OpenAI, and Block are seeding the foundation with the core standards and tooling already powering today’s agentic ecosystem, the spotlight also falls on the broader cohort of Platinum members – namely Amazon’s cloud unit AWS, Google, Microsoft, and Bloomberg, whose participation signals deeper industry stakes.
Rounding out the AAIF’s founding members are a roster of Gold and Silver partners that includes companies like Shopify, Snowflake, Datadog, Docker, IBM, JetBrains, Salesforce, and Uber — signalling that interest in agent interoperability spans everyone from cloud hyperscalers and enterprise software giants, to companies across the broader developer ecosystem.
Looking to the future, Surtani said that the AAIF is already in discussions with additional companies about widening the pool of projects under its umbrella.
“We’ve been working on launching the foundation, and the next task is defining how we bring in more projects in a way that is transparent, fair, and meaningful to the community,” he said.
Beyond the current portfolio, Surtani stressed that the foundation’s long-term credibility depends on vendor-neutral governance — a structure designed to stop any single company, however large, from steering the standards.
“Technical decisions go to steering committees of project maintainers, made up of active participants in the project regardless of which company they work for,” Surtani said. “This reduces the risk that one company can make decisions on a project without consensus. Further, the Linux Foundation holds the trademarks to the project, so brand value and integrity are maintained with the community. If a company's interests diverge from the community, they are free to fork the project, but they would have to call it something else. This is the same governance model responsible for the world’s most critical shared infrastructure, including Linux, Java, Kubernetes, OpenSSL, and much more.”
