All services

AI that earnsits place.

AI where it pays off

Most companies do not need an AI strategy. They need two or three specific tasks automated well, and a clear-eyed view of what the model gets wrong and what that costs.

Write to us

This fits if

  • Your team spends hours a week reading, sorting or rewriting the same things.
  • The board has asked what your AI plan is and you would rather answer with evidence.
  • You have documents, tickets or records nobody has time to search properly.
  • You tried a chatbot pilot and it quietly stopped being used.

What is included

  1. Opportunity mapping

    We watch how the work actually happens, then rank candidate tasks by hours saved against risk and effort. Most of the list gets rejected, deliberately.

  2. Proof of value

    One task, built properly and measured against how it is done today. Two to four weeks, with a clear decision at the end.

  3. Cost and feasibility

    What each candidate needs in data, integration and ongoing spend - and which of them a spreadsheet would solve more cheaply.

  4. Governance and the EU AI Act

    Where your data goes, what gets logged, where a human belongs in the loop, and what your obligations amount to.

What you get

  • A ranked list of opportunities with estimated hours saved
  • A working proof of value on one task
  • An evaluation set so you can measure quality over time
  • A data-handling and governance note
  • A recommendation - including where not to use AI

What we work with

  • Claude · OpenAI · open-weight models
  • RAG & vector search (pgvector, Qdrant)
  • Python · TypeScript
  • MCP · agent tooling
  • Evaluation & tracing
  • Self-hosting for sensitive data
  • EU AI Act readiness
  • GDPR-compliant data flows

Tools are chosen to fit the problem, not the other way round.

AI strategy

Questions about this service

Will our data be used to train someone else’s model?

Not if it is set up properly. We use enterprise agreements that exclude training, and for genuinely sensitive data we can run open-weight models on infrastructure you control.

How will we know whether it actually works?

We build an evaluation set from your own real examples before building the feature, so “better” is a number rather than an impression.

Is it going to replace our people?

In our experience it removes the part of the job people already dislike. The headcount conversation is a business decision, not a technical one - and we will tell you plainly what it can and cannot take over.

Sound like your situation?

Write us a couple of sentences and we will tell you whether we can help.