Assessment
The bill nobody can assign, assigned.
Fixed price, net. Credited against the gateway rollout.
€3,000
For companies whose AI bill runs through API keys or Bedrock and who cannot assign it to anyone. Built from your providers' billing data and three to four conversations.
How it runs
What you spend on AI per key, team and model, how it developed over the months and where it is heading.
Who actually carries the usage. Usually a few who move a lot, and many accounts that do nothing and still cost.
Who hits the limit, when in the week, and what happens next: a pause, a workaround or a private account.
Which model carries how much, and where the choice obviously does not fit the task.
The ones nobody in the building knew about. They almost always turn up.
An amount and an intervention per finding. Not a measurement, and we do not call it one. But the basis for deciding whether the gateway comes next.
Levers
Budget
A budget per team and application, derived from actual usage. Those who move a lot get room. Those who do nothing stop costing.
Routing
Picking the model that fits the request, escalating only when the answer falls short. Decided by measurement, not by feel.
A cache
A layer in front of the model that recognises what has been asked before and reuses the answer.
Rebuilding the interaction
Often it is not the model that is too expensive but the dialogue that is badly built. That is design work, not infrastructure.
An evaluation track
Fixed test cases that every change is measured against: accuracy, speed, cost. Without them, every optimisation is a guess.
Guarantee
If we find less, you get the €3,000 back.
Found means savings potential documented in the report, each finding with an amount and a measure. Applies from €10,000 a month in AI spend. Refunded on request within 14 days of handover, no reason needed.
Fixed price, a clear result, no obligation.
An answer within 24 hours.
