Use case assessment
Which processes genuinely lend themselves to AI, with estimates of impact, risk and effort. And which to leave alone.
AI consulting
AI consulting
We help you decide what's worth doing with AI, what isn't, and in what order. Then, if it makes sense, we build it: we're the same people who write the code, so the advice has to survive contact with reality.
30 minutes to frame the problem. No sales deck.
- Consultants who build, not just advise
- Independent of vendors and licences
- GDPR and EU AI Act compliant
- Documentation always included
When it helps
When AI consulting is actually worth it
Not every situation needs a consultant. These do, because the cost of the wrong decision is high.
- You need to decide whether to buy an existing solution or build a bespoke one, and the two differ sharply over time.
- You've received proposals from several vendors and lack a technical yardstick to compare them.
- An AI project is already running and isn't delivering what was promised.
- You want a realistic roadmap instead of a list of good intentions.
We don't advise on AI for marketing or content generation: our ground is systems that enter operational processes.
What you get
Material you can decide on
Every consulting engagement closes with concrete documents, written to be read by management and used by engineers.
A reasoned technology choice
Which approach and which models fit your case, including the alternatives we ruled out and why.
Build vs buy comparison
What to buy and what to build, accounting for running costs, data constraints and vendor lock-in.
Audit of existing solutions
If an AI system is already in place: what works, what doesn't, what costs more than it should, and which fixes pay back fastest.
A prioritised roadmap
A sequence of steps with dependencies and checkpoints, sized for the people you actually have.
How we work
A short, verifiable consulting engagement
We prefer few steps with clear outcomes over a months-long analysis project.
Frame the problem
We establish objectives, constraints and what has already been tried. Often this is where it emerges that the real problem differs from the stated one.
Technical analysis
We examine your data, systems and existing integrations, and test feasibility in practice rather than in theory.
Recommendations
We present the options with costs, risks and timelines, plus an explicit recommendation. We don't leave the whole choice on your desk.
Validation or build
If the path is clear we build it; if a significant doubt remains, we settle it with a targeted proof of concept.
Expertise
Where our opinion is worth something
We only advise on what we have built and maintained in production.
LLM architectures
When a commercial model makes sense, when an open one does, when fine-tuning is justified and when good prompt design is enough.
RAG and knowledge bases
How to make company knowledge reliably queryable, with source citation and error control.
Agents and automation
Where system autonomy genuinely pays off, and where it introduces risks not worth taking.
Classical machine learning
Many problems are better solved by a traditional predictive model than an LLM. Knowing which is half the work.
Data and integration
Quality, accessibility and structure of data: the variable that determines project outcomes more than any other.
Compliance and risk
GDPR and the EU AI Act translated into concrete architectural choices, not a document to file away.
Working together
How an engagement is structured
Different formats depending on whether you need to decide, to verify or to build.
Fixed-scope assessment
An analysis with a defined subject and question, closing with written recommendations and a roadmap.
When you need a well-founded decision, quickly.
Ongoing advisory
A regular presence alongside your team: reviewing decisions, weighing vendors, unblocking technical problems as they surface.
When the internal team exists but lacks specific experience.
Consulting and build
The consulting continues into delivery with the same people: whoever proposed the solution also answers for how it behaves.
When the path is clear and you need someone to walk it.
Cost depends on scope: the depth of the analysis, the number of systems involved, and the duration of the support. That's why we don't publish rate cards — we define the scope on the first call and the proposal arrives with explicit numbers.
FAQ
What people ask about consulting
Agencies mostly work on marketing and content produced with third-party tools. We design and write software that enters operational processes and integrates with business systems. They're two different trades: if your need is the former, an agency suits you better.
No, and we hold no reseller agreements. The choice between commercial and open models, cloud and on-premise, is made case by case. When an off-the-shelf product is the right answer we say so, even though it means less development work for us.
Yes, we're asked this often. We assess architecture, output quality, running costs and compliance, and return a list of interventions ranked by payback. Sometimes the conclusion is that the system is sound and only needs better tuning.
Yes. In that case the consulting also covers how to govern the system over time: what needs oversight, how often, and which skills are worth building internally rather than depending on us.
A scoped assessment typically closes within a few weeks. Ongoing advisory is a periodic collaboration instead, sized to the pace of the decisions your team has to make.
Related services
If you're after something more specific, these paths start from a different need.
AI for business
You don't need to reinvent the company to use AI. We start from a process that costs you hours every week, test it on real data within weeks, and scale only once the numbers justify it.
AI consultant
Sometimes you don't need to hand over a project — you need someone to think with before deciding. A senior engineer who knows your context and is consistently available, without hiring a full-time role.
AI software development
Most AI projects stop at the demo. The distance between a prototype that works in a meeting and a system that survives daily use is made of integrations, error handling and monitoring — that's the part we do.
Let's talk
Bring us a decision you need to make
Half an hour to frame the problem and work out whether it needs an analysis, a validation, or nothing at all.