Transparency

How I work with AI

Last updated: 7 August 2026

I use AI tools. I'm writing that down here because clients deserve to know how a piece of work came about — and because a portfolio is only worth anything if you can believe it.

This website

Its structure, its code and a large part of its copy were produced with support from Claude. The requirements, the structure, the content and every decision are mine; I read, adjusted and tested every file before it went online. What stands here is my responsibility.

The portfolio projects

With the case studies, two things need to be kept apart:

• The assignment. The briefs — the requirement documents from the fictional clients — were generated with Claude. The reason is practical: a brief you write yourself only contains requirements you can already meet. A brief written by someone else asks questions you can't answer yet, and that is exactly where the learning happens.
• The build. I put the report together with Gemini as my assistant. Concretely: I had the path explained to me step by step, including concrete DAX formulas. My work was to follow each step, verify it in the model, hold it against the requirements in the brief, and correct it wherever the suggestion didn't hold.

That is a deliberate way of learning, not a shortcut. I make faster progress when a working result sits in front of me that I can take apart backwards: why is this function here, what happens if I remove it, where does it break? Start from zero and you spend the first week on syntax errors instead of on the data model.

What I explicitly do not claim is that I built this report freehand from memory. The standard I want to be measured by is a different one — I can explain every line of DAX in my projects: why it looks the way it does, what happens if you change it, and where it breaks. Ask me about it. Where I can't, it doesn't belong in the portfolio.

The data in the case studies

Every project shown so far runs on sample data. The companies are invented and so are the figures. That is stated at the top of each case study. These are not client projects and I don't present them as such.

The blog

Posts come out of my own practice. I use AI as a writing aid for phrasing and structure; the content, the examples and the code blocks come from my work and are checked by me.

What this means for you

• You get results I stand behind, regardless of which tool helped produce them.
• I don't claim client projects that never happened, or results I didn't achieve.
• You can ask about any piece of work and how it came about. The answer will be the same as here.

What AI doesn't take over

Judging whether a number is plausible. Deciding which metric actually answers the question. The three years in support where I learned how people describe problems and what they really mean. And the responsibility for the result.

← Back to the homepage