Token-Cost Audit
We go into your LLM stack and cut your inference bill — wrong-tier routing, missing caching, reasoning models where you don't need them.
£500flat fee
Built for teams spending £5k+/month on inference.
Most teams shipping AI features overpay for tokens — but the routers and gateways that fix the mechanics (routing, caching, batching) are free, and you may already run them. What you can't get for free is the neutral verdict: which vendor and tier are mispriced for your actual workload, where your contract and terms expose you, and which models are heading for deprecation. We measure your stack against each provider's current published pricing, read by hand at the time of the audit — and hand you that verdict, with no %-of-savings incentive to complicate it.
Audit
We map your stack + real usage against current model prices and the open/closed spread.
Verdict
A ranked call on which vendor and tier are mispriced for your workload, the terms and deprecation risks, and the switch plan with its projected monthly saving.
Ship
We implement the harness changes — or hand your team the exact diff.
Cut your bill or you don't pay. If the teardown doesn't surface at least its own £500 fee in annualised savings, we waive the fee — you keep the findings either way. Tell us your stack and we'll come back within a day.
Prices keep moving after the audit ends. The audit is a one-off — it is priced and delivered on its own, with nothing to subscribe to afterwards.