Cloudflare detailed the internal AI engineering infrastructure it built for its own workforce of 3,683 employees, using its own products as the foundation. The system routes requests through AI Gateway for centralized LLM management, uses Workers AI for cost-efficient open-source model inference, and exposes tools to employees through MCP servers behind a unified portal. The architecture has three layers: a platform layer handling authentication and routing via Cloudflare Access and AI Gateway, a knowledge layer that uses Backstage service catalogs and AGENTS.md files to supply codebase context, and an enforcement layer that runs AI-powered code review in CI/CD pipelines and checks compliance against internal engineering standards.
