Available now
AI
AI put to work on your own documents and your repeated tasks
Through avanserv, a one-engineer practice in Namur, Belgium, I put AI models to work for companies that are asked what they are doing about AI and want an answer that pays for itself. That covers retrieval over your own documents, automation of the work a team repeats every day, and models inside a delivery pipeline, and I say so when a plain script would do the job. I also help development teams master AI harnesses and tools, so that they deliver more and what they deliver is better. I work in English or French, remotely or on site anywhere in the European Union.
Available now
Where a model earns its place#
An AI project starts from one task, chosen because the time it takes today can be measured. I build the smallest thing that does that task on your real data, and the measurement decides whether it goes any further. When an answer comes out of your documents, it cites the passage it came from, so a person can check it.
Before any of that, the scope names the model, who hosts it and where your data is processed. That is decided with you and put in writing, because it is the first thing the person signing off will ask.
With a development team, the work is on how the team itself uses AI. I help it master the harnesses and tools it writes code with, so that it delivers more and the quality of what it delivers goes up. It is also how I work: every change to this site goes through a model review before it is merged, and the build log describes that gate.
Questions#
What does AI work cost?
The price follows the engagement, which is one of three separate kinds:
- A fixed-scope build has a fixed price, set from its written scope.
- An advisory block, for finding out whether a model fits a task at all, is billed by the hour, at €120.
- A retainer reserves days each week, from €720 a day.
What the models consume is billed on top of any of the three: either by the provider we choose together, directly to you and at its own price, or through my platform, at a rate set per model and per token. Prices exclude VAT.
How does an engagement start?
You describe the task on the scope page, in eight questions and about three minutes. Within two working days I send a written scope with its assumptions and its exclusions, or I tell you plainly that the task does not need a model.
How long does it take?
The duration is in the written scope, with what it assumes about your data and your systems, since those decide most of it. A retainer has no fixed term: the days remain yours each week, and changing or ending it takes a month's notice.
Which models, and where does my data go?
The model, its host and the place your data is processed are chosen per project, with you, and written into the scope before anything is built. Nothing of yours is sent anywhere that document does not name.
In which languages?
English or French, for the meetings and for everything written.
On site or remote?
Either one, anywhere in the European Union. My office is in Namur, Belgium.
From the work
AI agents a non-developer can deploy, on European infrastructure
CTO and lead developer at a Belgian company implementing AI and Odoo: providers that keep client data in Europe, and agents a non-developer can deploy.
Build log
Start with the problem, not with the proposal
A written scope comes back within two working days, with its assumptions and its exclusions stated. It costs nothing, and it is yours whether or not you hire me.
Tell me about your project, or the problem you want solved.
Or write to info@avanserv.com, and I will come back to you with a scope.