Block, the fintech company behind Square and Cash App, recently open-sourced its internal coding agent Goose and then transferred it to the Linux Foundation by establishing the Agentic AI Foundation (AAIF). This decision came after facing governance and trademark issues that hindered enterprise adoption. To understand the reasoning behind this move, we've compiled key questions answered by Manik Surtani, former head of open-source at Block and now CTO of the AAIF.
1. What was Goose, and why did Block initially open-source it?
Goose began as an internal tool at Block—an AI-powered coding agent designed to assist developers with repetitive tasks, code generation, and debugging. It proved so effective that Block decided to share it with the broader tech community. The open-source release under a permissive license allowed external developers to use, modify, and contribute to Goose. According to Manik Surtani, the move mirrored successful examples like Amazon’s internal cloud tools evolving into AWS. Block hoped that community adoption would accelerate development and validate Goose’s utility beyond their own walls. Within weeks of the public release, Goose saw rapid adoption, suggesting strong demand for such agentic tools.
2. What problems emerged after Goose was open-sourced?
Despite the initial enthusiasm, Block soon encountered “headwinds,” as Surtani called them. The most critical issue was lack of governance transparency. Because Block retained ownership of Goose’s trademarks, the project wasn’t truly “free and open” from an enterprise perspective. Companies considering adopting Goose were hesitant: they feared future licensing changes or legal complications tied to Block’s control. This trademark ownership hampered enterprise adoption, even though the code was permissively licensed. Without a neutral steward, potential corporate users worried about vendor lock-in or sudden shifts in project direction. The team realized that for Goose to reach its full potential, a structural change was necessary.
3. Why did Block decide to transfer Goose to a foundation?
Transferring Goose to a foundation was the most effective way to address governance and trademark issues. Foundations provide a neutral, transparent legal structure that reassures enterprise adopters. By placing Goose under a foundation, Block could separate the project’s ownership from any single company, ensuring long-term stability and community-driven decision-making. Surtani explained that this model also facilitates collaborative development across competing organizations, which is essential for a tool like Goose that works best when broadly integrated. The foundation route offered a clear path to building trust, encouraging contributions, and scaling adoption without the stigma of corporate control.
4. How did the Agentic AI Foundation (AAIF) come into existence?
The Agentic AI Foundation was created as a new arm of the Linux Foundation, purpose-built for agentic AI tools. Block collaborated with the Model Context Protocol (MCP) team at Anthropic and other early developers. According to Surtani, launching the AAIF with Goose, MCP, and Agents.MD was partly a matter of expediency—they wanted a working foundation quickly. By anchoring the foundation with these three mature projects, they provided immediate momentum. The Linux Foundation’s established governance frameworks and neutral reputation made it the ideal home. The AAIF is now open to new projects that align with its mission to advance agentic AI in a transparent, community-governed manner.
5. What role did Manik Surtani play in this transition?
Manik Surtani served as head of open-source at Block during Goose’s initial release and later became the CTO and co-founder of the Agentic AI Foundation. He was deeply involved in the strategic decision to move Goose to a foundation, working with internal teams, the MCP community, and Linux Foundation representatives. Surtani explained that his experience with open-source governance at Block helped identify the trademark and transparency issues early. Post-transition, he continues to guide the AAIF’s technical direction and community engagement, ensuring that Goose and its sibling projects thrive under neutral stewardship.
6. What benefits does placing Goose under the Linux Foundation bring?
Placing Goose under the Linux Foundation provides several concrete advantages. First, the foundation offers a proven governance model that includes clear trademark policies, intellectual property management, and conflict resolution mechanisms. This boosts enterprise confidence because companies can adopt Goose without fearing unilateral changes by a single vendor. Second, the Linux Foundation’s neutrality attracts diverse contributors, including competitors like AWS, which can collaborate on MCP within the same structure. Third, the foundation provides infrastructure for funding and community management, reducing Block’s ongoing burden. Finally, it ensures long-term survival—even if Block shifts priorities, Goose remains independently maintained.
7. What other tools joined Goose in the AAIF, and what’s next?
The Agentic AI Foundation launched with three initial projects: Goose (the coding agent), Model Context Protocol (MCP) from Anthropic, and Agents.MD (a documentation and tooling project). Surtani noted that this trio was chosen for expediency to get the foundation operational. However, the AAIF is designed to welcome additional agentic AI tools that meet its governance criteria. Future plans include expanding the foundation’s scope to cover interoperability standards, security guidelines, and shared tooling for the agentic AI ecosystem. Developers and organizations interested in contributing or hosting their projects under the AAIF can apply through the Linux Foundation.