The AI engineering stack we built internally — on the platform we ship
Quick Answer
Cloudflare's internal AI engineering stack, utilized by 93% of R&D, processed 51.83 billion tokens via Workers AI, achieving a 77% cost reduction compared to proprietary models.
Quick Take
The integration of AI tools has significantly increased developer velocity, with merge requests surging nearly double the baseline.
Key Points
- 3,683 internal users actively used AI coding tools in the last month.
- AI Gateway handled 688.46k requests daily, routing to four providers.
- Workers AI processed 51.47 billion input tokens last month.
- Merge requests surged from ~5,600 to over 10,900 per week.
- Kimi K2.5 on Workers AI is 77% cheaper than proprietary models.
Source Excerpt
We built our internal AI engineering stack on the same products we ship. That means 20 million requests routed through AI Gateway, 241 billion tokens processed, and inference running on Workers AI, serving more than 3,683 internal users. Here's how we did it.
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from Cloudflare AI
See more →
Temporary Cloudflare Accounts for AI agents
Cloudflare introduces Temporary Accounts on Cloudflare Workers, enabling AI agents to deploy live Workers instantly using 'wrangler deploy --temporary'. This feature removes barriers for AI agents, allowing them to operate more efficiently in real-time environments.
