One Invoice, Many Owners
Vendor bills arrive as totals, with no department or user attached, so nobody owns the spend.
Netra meters tokens from Claude Code, Claude API, OpenAI API, Codex, and more, attributes each to a user and department, and caps spend before the invoice.
Get a DemoTokens by Department
Product views on this page use illustrative data.
The Challenge
The invoice says how much. It doesn’t say who, or for what.
Vendor bills arrive as totals, with no department or user attached, so nobody owns the spend.
One autonomous run can outspend a week of chat; seat counts no longer predict the bill.
Without a cap, an overrun appears with the invoice, after the tokens are spent.
What You Get
Measured at the endpoint and Gateway. No SDK, no code changes.
Every metered token shows its USD cost by user, department, model, and channel.
Users and departments from your identity provider, purpose from use-case scenarios, price from your Cost Table.
A quota sets the ceiling per channel and model; Circuit Break blocks the channel once it’s spent.
Cap and Circuit Break
Token Quota turns a budget into a limit. Circuit Break enforces it.
Quotas
Scope, reset period, usage against limit, and the action On Exceeded
In this example, Engineering is at 82% of its Claude Code quota; at the limit, Circuit Break blocks Claude Code for Engineering only.
AI Chargeback by Department
Budget owners rarely accept a charge they can’t check. Start with showback; charge back once the numbers hold up.
Next Steps
FAQ
Each metered token is attributed to a user and department from your identity provider and priced at your own rates.
Yes, most teams do. Share department totals as showback; budget owners can check the user, tool, and use case behind each figure.
Yes. Export Report in Token Insights gives finance the token and cost breakdown without using the console.
Tokens and cost by user, department, model, and use case, which channels drive spend, and where a quota makes sense.