Block Open-Sources Goose, an AI Agent Scaled to 60% Weekly Usage Across Company

Summary

Block, the financial services and payments company, has open-sourced Goose, an AI agent that achieved significant internal adoption with 60% of the company's workforce using it on a weekly basis. The release marks a notable milestone in making internal AI infrastructure publicly available, allowing developers outside the company to leverage the same agent technology that demonstrated strong enterprise adoption.

A key highlight of the Goose open-source release is the 'recipe file' system, which appears to be an innovative approach to configuring and orchestrating AI agent behavior. This recipe-based architecture likely allows developers to compose complex agent workflows and behaviors through declarative configuration rather than code, making it more accessible for non-technical users and enabling rapid customization of agent capabilities.

This open-source release positions Block as a contributor to the broader AI agent ecosystem, following the trend of major technology companies sharing internal tools and models with the developer community. The decision to open-source Goose suggests Block's confidence in the agent's architecture and its potential value to the wider industry.

• The release signals growing maturity and standardization of enterprise AI agent tooling

Editorial Opinion

Block's decision to open-source Goose is significant because it validates AI agents as a core internal productivity tool—when 60% of a company uses something weekly, the bet is real. The recipe file approach suggests Block has solved a genuine UX problem around agent composition, and open-sourcing it could become a widely adopted pattern in the agent engineering space. This is the kind of practical, battle-tested infrastructure that tends to drive real adoption in the open-source community.