AI Consulting

Hardly any topic is hyped as much as artificial intelligence — which makes a sober assessment difficult. AI consulting creates clarity: where does AI bring your company real value, which solution fits and what is realistic? As an independent consultant I give an honest recommendation — even when the answer is “this isn’t worth it (yet)”.

AI readiness: where is AI really worth it?

Not every process needs AI. In an AI readiness analysis I look at your workflows and identify the points with the biggest lever — tasks that are repetitive, text-heavy or data-driven. That way you invest where it pays off, instead of adopting AI for its own sake.

The right tool and model selection

The market of AI tools and models is confusing and changes fast. I help with the selection to match your goals, data and budget — from ready-made services to custom solutions — and assess cost, benefit and effort realistically.

Considering data privacy from the start

Especially in the DACH region, data privacy decides whether an AI project is viable. I assess data flows, providers and locations and show GDPR-compliant alternatives — including European or self-hosted models where the data requires it. That gives you a solid basis for your decision.

The AI workshop: one day, a clear outcome

For many companies a compact workshop day is the best entry point. In the morning we walk through your workflows and mark the spots where lots of text, data or repetition pile up. In the afternoon we score the candidates by benefit, effort and data quality. By the end of the day you have, in writing:

• a prioritised list of 3–5 use cases with realistic effort and cost estimates,
• a privacy assessment per use case,
• a clear recommendation on where to start — or whether to hold off entirely.

What a roadmap can look like

An example from a service company with five people in the back office:

1. Quick win (weeks 1–3): automatic classification and routing of incoming email — small effort, immediately noticeable relief.
2. Build-out (within the quarter): an FAQ assistant based on the company’s own content, delivered as an AI integration project.
3. Review after 3 months: look at the numbers first, then decide on further steps — instead of wanting everything at once.

When I advise against AI

Honest consulting includes saying no. Typical cases where I do:

• Too little volume: with five enquiries a day nobody needs a chatbot — a good FAQ page does the job.
• A messy data foundation: AI on incomplete data produces confident nonsense. Fix the data first, then pick the model.
• Legally binding statements without human sign-off — contracts, health or finance topics do not belong in unreviewed model output.
• A simple rule would do: a filter or an extra form field costs nothing and never hallucinates.