In use internally
AI Gateway
One endpoint for every AI model.
A single, OpenAI-compatible endpoint in front of every model provider. Our products and agents call one API — the gateway picks the model, keeps keys safe and keeps costs in check.
Why we built it
Every product we ship talks to AI models — and every provider has its own API, limits and outages. We wanted one place to control which model answers, what it costs and what happens when a provider goes down.
What it does
Built for real work, every day.
01
One API for every model and provider
02
Keys, budgets and rate limits per team, product and agent
03
Automatic fallbacks, caching and full request logs
How it works
Three steps.
- 01Your app or agent sends a standard request to one endpoint
- 02The gateway checks the key and budget, then routes to the best model for the job
- 03If a provider fails it falls back automatically; every call is logged with cost and latency
Use cases
Where it helps.
- Switch models without touching product code01
- Give every team and agent its own key and spending limit02
- Keep AI features online when a provider has an outage03
UDEX Labs
All 8 projects