How I build with AI – human plus tool
“AI-assisted” now appears on many offers. It rarely says what that actually means. So let me briefly take you into how I work, because the difference between “made by AI” and “made by a human with a strong tool” decides the quality.
What the AI takes over
For me, AI is a very fast colleague for everything that repeats: code scaffolding, test templates, quickly searching documentation, first drafts, suggestions for cleaning up existing code. This work is necessary, but rarely the part that makes a project good or bad. This is where AI saves real time.
What deliberately stays with the human
The decisions that matter, I do not hand off:
- What gets built at all and which problem it solves,
- the architecture and how the solution can later be maintained and extended,
- quality and security, meaning tests, data protection and an eye on the consequences.
The AI suggests, I take responsibility. A simple rule helps: code I do not understand myself does not go into production.
Why this is an advantage for you
The benefit is not that “an AI writes the code,” but the split: routine work goes faster, which leaves more time for your actual requirements and for clean implementation. The pace goes up without losing review and responsibility, because both stay with a human.
My learning game Arithia is a good example: built AI-assisted and still backed by more than 120 automated tests, fully working offline and production-ready. Built quickly, but not thrown together quickly.
So “AI-assisted” here means: a human makes the decisions and carries the responsibility, with a tool that speeds up the legwork. If you want to know what that would concretely mean for your project, feel free to reach out.