AI usage audit
A full picture of your AI spend, plus a list of savings ordered by impact.
May your AI budget fight harder for you
We help Belgian companies pay less for AI models and inference while the output stays just as good.
GDPR-first
NDA before we start
Belgian team, on-site
Teams we've worked with across Belgium
SAASFINTECHHEALTHTECHRETAILE-COMMERCEBIOTECHMEDIA
SAASFINTECHHEALTHTECHRETAILE-COMMERCEBIOTECHMEDIA
0%Average inference cost reduction in our tests
€0Cost of the first call
0%Commission from AI providers
0 citiesOne local team, close to you
The challenge
Make it work. Make it right. Make it cheap.
You've nailed the first two. The third is easy to put off: every feature you ship adds new calls and longer prompts, and the bill keeps creeping up. Keeping it down is a job of its own. We take that job off your hands.
- See where every token goesSee cost per feature, team and model, down to the single request.
- Tailored approach for each taskSimple work goes to smaller models. The big ones only get the requests that need them.
- Savings you can prove to your CFOWe measure every saving against a baseline you sign off on before we start.
Monthly inference spend
Live€41,280 ▼ 36%
SavedSpend after optimization
€14.8ksaved this month
How it works
We don't run the same playbook for everyone.
01
Look at your usage
We go through your usage data, invoices and workflows and work out where the money goes.
02
Find the leaks
Oversized models, bloated prompts, repeated calls, idle capacity. We rank what we find by how much it costs you.
03
Fix what matters to you
We build the fixes with your team and test them against your own quality checks before anything goes live.
04
Keep watching
Dashboards and a monthly report show what you saved and flag new leaks early.
Some of our approaches
Model routing and right-sizing
Send each request to the smallest model that does the job well, with automatic fallbacks.
Prompt and token optimization
Shorter prompts, tighter context and structured output. You get the same answers for fewer tokens.
Caching and batching
Stop paying twice for the same answer. Prompt caching, semantic caching and batch jobs handle that.
GPU and provider negotiation
Benchmarks and negotiation support for API contracts, reserved capacity and self-hosting.
AI FinOps dashboards
Live cost tracking per product, team and customer, with budgets and alerts.
…and many more
Where we work
Local experts in Antwerp, Brussels, Ghent and Leuven
One team, working across four cities. We come to your company and speak your language.
Based in another city? Let's find a solution
Security
Your data stays yours
GDPR compliant
Data processing agreements and data minimisation come standard.
EU data residency
Our analysis and tooling run on EU infrastructure.
Hands-on with your workflows
We work inside your real workflows to test what can change, with only the access you approve.
NDA by default
We sign the NDA before the first call.
Client stories
What clients say
★★★★★
"We finally understand our AI spend"
We knew our AI costs were growing, but not why. After the audit we had a clear list of fixes, and the first ones paid for themselves within weeks.
HE
Head of EngineeringSaaS scale-up · Antwerp
−41%
★★★★★
"Savings in the first month"
Model routing alone took a real bite out of our monthly invoice, and our users didn't notice a thing. That was exactly the point.
CF
CFOFinancial services · Brussels
−33%
★★★★★
"It felt like they were part of our team"
They sat with our developers, explained every trade-off and left us with dashboards we still use every week.
PL
Product LeadHealthtech company · Ghent
−38%
FAQ
Frequently asked questions
How does the AI usage audit work?
You give us read-only access to your AI billing and usage data. We come back with a breakdown of your spend and a prioritised list of savings. No cost, no commitment.
Will optimization affect the quality of our AI output?
Not when it's done right. Every change is tested against your own quality benchmarks before it goes live, and anything that falls short is rolled back.
Which AI providers and models do you work with?
We're independent. We work with the major model APIs and cloud platforms as well as open-source models you host yourself, and we don't take commissions from any provider.
How do you handle our data and security?
We work under an NDA and a data processing agreement, with the least access needed. Most of the analysis uses billing and usage metadata, not the content of your prompts.
How are your fees structured?
The audit is free. After that the price depends on what you need. Optimization sprints have a fixed price we agree on upfront, and ongoing FinOps is a monthly fee. We're also open to pricing based on results.
Do you sell a standard package?
No. AI bills grow for different reasons at every company, so we start by learning what's going on at yours. The plan we propose is written for your stack and your budget.
Let's see if we're a good match
Book a short call. We'll answer your questions and tell you honestly whether we can make a difference for you.
Book a call