How Block Uses AI Agents: Inside a Payments Giant's Agent Strategy

Block reportedly deployed its open-source Goose AI agent across its 12,000-person workforce in eight weeks by automatically installing it on company laptops, offering approved MCP tools, multiple model-provider options, and internal DevRel support designed for non-technical users. Its governance model centered on a curated MCP-server allowlist, OAuth and identity-provider integration, and separate engineering and security reviews before tools could be used, while dynamic tool selection and conversation summarization sought to limit context-window overload. Employees shared cross-functional use cases ranging from sales-record management and procurement guidance to natural-language fraud analysis, and a Hack Week produced more than 60 internal MCP servers, with the catalog later expanding further. Block reported a 40% increase in production code shipped per engineer since September 2025, though this metric does not represent company-wide productivity or ROI. The account also describes Block’s early Buzz workspace platform, which uses cryptographic agent identities and signed actions, and argues that large-scale agent adoption depends on identity, access controls, credential management, monitoring, provider choice, and user enablement; it contrasts Block’s internally built controls with MintMCP’s managed gateway offerings.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.