Give every customer their own AI budget
Set a spending limit for every customer, or team, or agent. Stop overspend the moment it happens, not when the invoice lands. And see exactly what each one cost, so you can bill it back.
Selling AI is easy. Controlling it isn't.
The provider sends one bill. Your customers are many. Everything you can't see or stop lives in that gap.
You can't tell who costs what
The provider bill is a single number. You've no way to know what any one customer, team, or feature actually cost you.
One heavy user wipes the margin
A single customer on a flat plan can quietly burn the profit you made on ten others, and you find out when the invoice arrives.
Your plan limits aren't real
Your pricing says "500 AI actions a month." Nothing actually stops someone using 5,000 and handing you the bill.
Budget. Enforce. Bill.
No rebuild. Point your existing AI calls through us, and set limits on the customers you already have.
Give everyone a limit
Map a budget to each customer, team, or agent. Free gets £5, Pro gets £50, Enterprise whatever you choose — tied to the plans you already sell.
Stop overspend in real time
We check every call against the budget and block it the instant the limit's reached — before the cost lands, not after the invoice.
Charge for what was used
Get the exact cost of every customer, ready to drop into an invoice. The number your provider's bill never breaks down for you.
Works with your existing OpenAI or Anthropic setup — one line of config, or our API. See the docs →
Soft caps, or hard caps. Your call.
Most teams are well served by soft caps. Turn on hard caps for the budgets where going over isn't an option.
Simple and instant
Spend is tracked as it happens, and calls are blocked the moment a budget is hit. Works with every model and workload.
A limit that holds under load
For customer-billed AI, regulated spend, or anything high-stakes. We reserve each call's worst-case cost before it runs — like a hotel holding a deposit on your card at check-in — then settle to the real amount after. Concurrent calls can't collectively bust the limit.
What's in the box
Set up in an afternoon. Works with your existing providers.
Per-customer budgets
A budget for every customer, team, or agent — enforced automatically, in real time.
Real-time enforcement
Calls are blocked the moment a limit is hit. Decisions land in milliseconds, not minutes.
Per-customer usage core
See exactly what each customer used. Export it, bill on it, prove it.
Soft or hard caps
Block at the limit, or reserve cost up front for a guarantee that holds under load — per budget.
200+ models
OpenAI, Anthropic, Gemini, Mistral, Cohere, DeepSeek — one integration, pricing kept current.
Instant alerts
Slack, webhook, or email when a budget is hit or crosses 80%.
Loop protection
Velocity limits catch runaway loops before a bug burns a whole budget.
Full audit trail
Every allowed and blocked call, durably recorded — the record behind what you bill.
Prompts never stored
We record usage and cost, never the contents of your calls.
Every customer's cost, itemised
A real billing report — daily and monthly breakdowns, cost per call, CSV export. Not a mockup.
Every row is durably recorded with timestamps. The same data you bill on is the data you defend a chargeback with — one record your finance team can audit on demand.
Start free. Pay as you grow.
Simple monthly tiers. Hard caps available as an add-on whenever you need them.
The short version
The things people ask before trying it. The full detail lives in the docs.
Do I have to change my code?
What's the difference between soft and hard caps?
Which models and providers do you support?
Is my data private?
What does it cost?
Can I bill my own customers from this?
Stop guessing what each customer costs you
Set a budget for every customer, enforce it in real time, and bill on exactly what was used.