OpenAI compatible, with a finance department attached.
Point your existing client at the endpoint, change the base URL and the key, and keep the rest of your code.
- Drop in compatible
- Budget per key
- Cost per call recorded
-
Your names for your models
A deployment maps the name you use to a provider and an upstream model. Swapping what sits behind the name is an administrative change, not a release.
-
A key that cannot run away with the bill
Each key carries its own budget. When the ceiling is reached the request is refused in the standard error shape, with the reason in it.
-
Money reserved before the request leaves
The platform holds the worst case cost of a call before it goes upstream and settles the real amount afterwards, so parallel traffic cannot outspend the balance.
-
A record you can reconcile
Every call is written to the trail with its tokens, its cost, the key that made it and the person or agent behind it.
Operational detail
- Standard error shapes, so existing clients handle refusals
- Streaming responses report usage as well
- Keys are stored as hashes and shown once
- Per organization and per key spend limits, set in the console
What your team gets
Five parts, one account, one set of permissions.
-
Chat and Fusion
Ask across models, grounded in your own documents, with citations you can open.
-
Agents
Automations that do the work, ask before anything irreversible, and leave a trail.
-
Knowledge
Your documents, searchable by meaning, with access granted per base and per group.
-
Code and Data
Functions, a document store and hosted sites, without leaving the platform.
Change two lines and try it
The base URL and the key. Everything else in your client stays where it is.