The Linux Foundation has launched a new initiative aimed at creating shared standards for AI agents, as major technology companies move beyond chatbots toward systems that can take autonomous actions. The effort, called the Agentic AI Foundation (AAIF), is designed to prevent the AI agent ecosystem from fragmenting into incompatible and tightly controlled platforms.
Founding Contributions and Initial Projects
At launch, the AAIF is anchored by open source donations from Anthropic, Block, and OpenAI. Anthropic is contributing its Model Context Protocol (MCP), which provides a standardized way to connect AI models and agents to tools and data. Block is donating Goose, its open source agent framework. OpenAI is contributing AGENTS.md, a lightweight instruction file developers can add to code repositories to define how AI coding tools should behave.
These tools are intended to serve as core infrastructure for the emerging AI agent ecosystem.
Other members joining the foundation include AWS, Bloomberg, Cloudflare, and Google, signaling broad industry participation in shaping shared technical foundations for agent interoperability and safety.
Goals Around Interoperability and Governance
OpenAI engineer Nick Cooper said protocols function as a shared language that allows different agents and systems to communicate without requiring custom integrations for each deployment.
“We need multiple [protocols] to negotiate, communicate, and work together to deliver value for people,” Cooper told TechCrunch, adding that openness makes it unlikely that a single company will control the agent ecosystem.
Linux Foundation executive director Jim Zemlin said the goal is to avoid an environment dominated by proprietary, closed systems. He said the AAIF will coordinate interoperability, safety patterns, and best practices for AI agents under a neutral governance structure.
The AAIF will operate through a “directed fund,” where companies contribute money through membership dues. The Linux Foundation said project roadmaps will be overseen by technical steering committees, and that no single member will have unilateral control.
Block’s Goose Framework and Open Collaboration
Block, best known for Square and Cash App, is positioning Goose as proof that open source agent frameworks can scale. The company said thousands of its engineers use Goose each week for tasks including coding, data analysis, and documentation.
Block AI tech lead Brad Axen said open sourcing Goose allows outside contributors to improve the framework while also benefiting Block’s internal development. Donating Goose to the Linux Foundation also allows the framework to be tested across a broader community and integrated with other shared standards such as MCP and AGENTS.md.
Anthropic’s Protocol Donation and Adoption Strategy
Anthropic is contributing MCP to place the protocol under neutral governance and encourage broad adoption. MCP co-creator David Soria Parra said the aim is to establish a single integration layer developers can use across different AI clients.
“The main goal is to have enough adoption in the world that it’s the de facto standard,” Soria Parra told TechCrunch, adding that developers benefit when they can build once and deploy across multiple platforms.
Oversight, Adoption, and Industry Impact
The Linux Foundation said the AAIF is focused specifically on standards and orchestration for AI agents, alongside shared safety practices. Zemlin said early success will be measured by the degree to which vendor-built agents adopt shared standards in real deployments.
OpenAI’s Cooper said success would also depend on the continued evolution of the standards, rather than allowing them to stagnate within the foundation.
Zemlin acknowledged that even in open ecosystems, dominant implementations can emerge through usage and developer adoption. He pointed to Kubernetes as an example where market leadership developed through technical merit rather than vendor control.
Implications for Developers and Enterprises
For developers and enterprise users, the immediate benefit of standardization is reduced need to build custom connectors, more predictable agent behavior across systems, and simpler deployment in regulated environments.
Longer term, the foundation’s backers are aiming for an interoperable agent ecosystem built on shared infrastructure rather than closed platforms, mirroring how open standards shaped the modern web.
Featured image credits: Freepik
For more stories like it, click the +Follow button at the top of this page to follow us.