Cloudflare is unifying Workers AI and AI Gateway
A useful Cloudflare read about treating AI calls as traffic that needs routing, logs, billing, retries, and control from the start.
Source
Unifying Workers AI and AI Gateway into a single AI control plane
Why I saved it
I like this because it turns model calls into normal infrastructure. The post talks about one path for Workers AI and AI Gateway, with observability, billing, security, logging, and routing in the same control plane.
That is the right direction. AI apps should not have random fetch() calls to models hidden across the codebase.
My notes
- The unified AI binding makes Workers AI and AI Gateway feel like one path.
- A default gateway can give visibility without setup.
- The article frames model-first routing as better than provider-first routing.
- Logging token usage and latency per model is essential once an app has more than one AI feature.
What I want to remember
The AI layer should have the same discipline as any backend dependency. If I cannot see latency, errors, token usage, and cost attribution, I am guessing in production.