How I work

How we work together.

We go through three steps, you always talk to me, and two decisions are yours. Everything in between is my job.

  1. Listen

    We talk about your business, your people and where things get stuck. Afterwards I write down what the problem is and how we will know it has been solved. I write it in your words, without jargon, and the price is included.

    You say yes or no. That is your first decision.

  2. Build

    Then I build, together with my AI team. Early on you get something to click through, and you tell me what fits. Because nobody sits between you and me, your feedback goes straight into the next version.

    Before anything goes live, a separate AI reviewer checks the whole thing, one that had no part in building it. Then you see the result and sign it off. That is your second decision.

  3. Stay on it

    After launch I stay within reach. Whatever shows up in daily use, I take care of. What I learn along the way I write down, so the next round goes faster.

At an agency

Like a game of telephone.

At an agency, your request is handed from sales to the project lead, concept and design and on to development. With every handover the line gets shakier and paler.What you meanSalesProject leadConceptDesignDevelopmentWhat arrivesAt an agency, your request is handed from sales to the project lead, concept and design and on to development. With every handover the line gets shakier and paler.What you meanSalesProject leadConceptDesignDevelopmentWhat arrives

Your request gets passed on four times. Each time, a little of what you meant is lost, and you end up paying for the fixes too.

With me

No handovers.

With me, your request comes straight to me. Around me works my AI team: planning, concept, design, development, review and research.What you meanWhat you getPlanningConceptDesignDevelopmentReviewResearchWith me, your request comes straight to me. Around me works my AI team: planning, concept, design, development, review and research.What you meanWhat you getPlanningConceptDesignDevelopmentReviewResearch

I listen, decide and build with my AI team. It covers the roles an agency needs people for.

At an agency, your request passes through many hands. With me, it stays in one head.

Who decides what.

You decide

  • whether we start
  • whether the result is right

I take care of

  • how it gets built, in what order and with what
  • the check before anything goes live
  • making sure it runs, and answering for it

Why it is so fast.

At an agency your request passes through many hands: sales, project management, concept, design, development. Something gets lost at every handover, and a lot has to be agreed all over again. With me it all stays in one head. The work an agency hires people for is done by my AI team. It writes code, checks, researches and takes the busywork off my hands. That is why weeks are often enough where others plan in months.

What it costs.

You know the price before I start. If you would like to see how we work together first, start small, with a paid trial run on a manageable part of the problem.

Who decides what the AI does?

You decide what gets built. I decide how. The AI decides nothing and passes no judgement on people. Before every launch a separate AI reviewer checks the result, at the end you sign it off, and I answer for all of it.

What if you are ever out of action?

The code belongs to you. It is built on widely used technology that many developers know, so someone else can carry on with it.

If you would rather do it yourselves.

Not everyone needs someone to build. Sometimes what is missing is clarity: which processes can go digital, where AI really helps and where it doesn’t, how a team can switch over step by step without stopping the business. Then I advise, on product, team and approach. You get a plan your team can carry out, and me when it gets stuck.

Ask for advice

How my AI team works

I build almost everything with Claude and Claude Code from Anthropic. Claude is a language model, an AI that reads and writes text and code. Claude Code is the version that works directly inside my projects. In special cases I bring in Codex from OpenAI.

My team is several of these AI assistants working on different tasks at the same time. One plans, one builds, one researches. Before every launch a separate reviewer checks the complete result, and a second one looks at plans that would be hard to undo. What always applies is written in a contract a few pages long, which every assistant reads first.

I decide what gets built and when it is done. So that nothing slips past me, I built Kiesel, a small figure on my screen that shows me which assistant is waiting for me.

Everything I use every day

Contact

Which problem should disappear next?

Tell me what’s slowing you down. You’ll get an honest assessment from me personally: whether I can help, and how.