The AI ​​engineering stack we developed internally – on the platform we deliver

The AI ​​engineering stack we developed internally – on the platform we deliver

By The Cloudflare Blog
Publication Date: 2026-04-20 13:00:00

In the last 30 days, 93% of Cloudflare’s R&D organization was using AI coding tools based on infrastructure we built on our own platform.

Eleven months ago we started a big project: really integrating AI into our engineering stack. We had to build the internal MCP servers, access layer, and AI tools needed to make agents useful at Cloudflare. We brought engineers from across the company together to form a tiger team called iMARS (Internal MCP Agent/Server Rollout Squad). The sustained work ended up with the Dev Productivity team, which also owns much of our internal tools, including CI/CD, build systems, and automation.

Here are some numbers that capture our own Agent AI usage over the last 30 days:

  • 3,683 internal users A total of around 6,100 employees actively use AI coding tools (60% company-wide, 93% in research and development).

  • 47.95 million AI requests

  • 295 teams currently use agent AI tools and coding assistants.

  • 20.18 million AI Gateway requests per month

  • 241.37…