One key, many hands.
Everyone shares one key. Nobody can say who spent what, or why.
The budget layer for AI work
A budget and a purpose for every piece of AI work. Your people move fast. You see what the spend produced.
Works with the AI provider you already pay for. Prompts are never stored.
Checkout retry fix
Will produce Retry logic for failed payments, with tests.
Requested by Priya · Approved by Dana
Session active · VS Code
Copilot through Brigaidier · 2 approved models
Before this call reaches your provider
The problem
Everyone shares one key. Nobody can say who spent what, or why.
The invoice says how much. Not which project, or what came of it.
So good people wait for permission, or quietly work around it.
How it works
Plug in your own AI provider. Choose the models and their prices. The key never leaves the server.
Someone requests a budget and says what it will produce. You approve, adjust or decline.
They start a session from VS Code or the browser. No keys to copy.
Every call is tied to a person and a project. Spend sits next to progress.
Works with the provider you already use.
Who it’s for
If you fund the work
If you do the work
Safety and control
It never leaves the server. People get short-lived keys instead.
Every call is checked for person, project, model, limits and budget before it reaches your provider.
End a session mid-call, freeze a budget, or stop all AI access at once.
Every admin action is logged and exportable. Prompts and responses are not.
Why we built this
Leaders want to know AI is worth it. Engineers want to get on with the job. Most tools pick a side. We built one both can trust.
Every dollar of AI should have an owner and a reason.
Pilots
Tell us about your team. We’ll open your workspace once it’s approved.