When Block open-sourced its internal coding agent Goose, it seemed like a straightforward move. But as the tool gained traction, a critical question emerged: who really controls it? The answer led Block to hand Goose—along with trademarks and governance—to the Linux Foundation's new Agentic AI Foundation. Here are the six pivotal reasons behind that decision.
- 1. From Internal Tool to Open Source
- 2. Early Adoption Revealed Governance Gaps
- 3. The Trademark Roadblock
- 4. Why a Foundation Was the Right Home
- 5. The Agentic AI Foundation: A Strategic Alliance
- 6. Expediency and Future Growth
1. From Internal Tool to Open Source
Goose began its life as an internal productivity enhancer at Block, the fintech company behind Square and Cash App. Originally built to help developers write and debug code faster, it quickly proved its worth across the organization. Block rolled it out to all 12,000 employees, and the tool's success prompted a pivotal decision: release it as open source. That leap mirrored the trajectory of Amazon's internal infrastructure becoming AWS—what starts inside can become a powerful external service. By open-sourcing Goose under a permissive license, Block invited the broader tech community to adopt and improve the tool, turning a narrow internal asset into a potential industry standard.
2. Early Adoption Revealed Governance Gaps
After the public release, Goose saw rapid adoption—developers flocked to experiment with the coding agent. But Manik Surtani, former head of open-source at Block and now CTO of the Agentic AI Foundation, noted that success came with headwinds. The biggest issue was a lack of transparency in project governance. Enterprise users, in particular, were hesitant to bet on a tool whose decision-making processes were opaque. Without clear community leadership or a neutral steward, trust eroded. The very qualities that make open source attractive—collaboration, shared ownership, clear rules—were missing. Block realized that to sustain momentum, Goose needed more than a license; it needed a trusted governance structure.
3. The Trademark Roadblock
Even after releasing the code, Block retained control over the Goose trademark. This seemingly minor detail had major consequences. Enterprises, especially those with compliance-heavy procurement processes, view trademark ambiguity as a red flag. They asked: Could Block change the terms? Could they assert control later? Surtani described how this hampered enterprise adoption. The tool wasn't truly free and open if one company still held a key asset. For Goose to "fully unfurl its wings and soar," the project had to be completely vendor-neutral. This meant transferring not only the code but also the brand and community assets to a neutral party.
4. Why a Foundation Was the Right Home
The solution to governance and trademark issues was to place Goose under a foundation. Foundations provide a legal and organizational framework that ensures neutral stewardship, open governance, and long-term stability. They can hold trademarks, manage contributions, and provide a conflict-resolution mechanism. For Block, this was the only path to build the trust needed for widespread enterprise adoption. But which foundation? The team evaluated options and ultimately landed on the Linux Foundation—a proven home for critical open-source projects like Kubernetes, Node.js, and the Cloud Native Computing Foundation. Its track record and community of practice made it an obvious choice.
5. The Agentic AI Foundation: A Strategic Alliance
Rather than drop Goose into an existing Linux Foundation project, Block helped establish a new entity: the Agentic AI Foundation (AAIF). This foundation launched with not just Goose but also two other projects: Model Context Protocol (MCP) from Anthropic and Agents.MD. By bundling complementary tools, AAIF aims to create a home for agentic AI development—tools that let AI act on behalf of users. Surtani explained that the decision to start with a trio was partly driven by a desire to create immediate critical mass and show that the foundation was about more than one company's project. It signaled a collaborative ecosystem from day one.
6. Expediency and Future Growth
Why not wait for a more comprehensive set of projects? Surtani admitted that launching AAIF with just three tools was partly a matter of expediency. The team wanted to get the foundation operational quickly to maintain momentum around Goose. Delaying would risk losing community interest and trust. The immediate priority was to establish governance, transfer trademarks, and provide a clear home for contributions. With the foundation up and running, AAIF can now expand its umbrella to include more projects. Surtani hinted that additional tools could join soon, making AAIF the go-to foundation for agentic AI. The quick launch was a strategic move to ensure that Goose and its siblings could fly while still building the nest.
Block's decision to hand Goose to the Linux Foundation wasn't an abdication—it was a maturation. By recognizing that an open-source project needs more than code—it needs a trusted governance home—Block ensured that Goose could achieve its full potential. The Agentic AI Foundation now has a strong starting lineup and the organizational support to grow. For enterprises looking to adopt AI agents, the message is clear: this is a project built to last.