
Build
Start from the problem, not the prompt
An application's quality is not decided in the prompt, but in the problem you managed to define.
Most projects that fail are not badly built. They are well built on a badly defined need. I am often asked for "a dashboard," "a form," "a tracking app." These are screens. A screen is not a problem: it is the assumed answer to a problem that has not yet been formulated.
Describing the screen locks in the wrong solution
When you start from the prompt ("build me a page with a table and three buttons"), you get exactly that page. It displays, it looks nice, and three weeks later nobody uses it, because it answered an intuition, not a use case. You built fast in the wrong direction.
The four questions that stand in for a spec
Before the first line, I ask four questions. They are worth more than any prompt:
- What decision should this tool make possible, and for whom?
- What data comes in, where does it come from, and who is allowed to see it?
- What action is repeated manually today, and how many times a week?
- What happens when two people work on it at the same time?
Until these four answers are clear, no screen deserves to be designed. Once they are, the screen almost designs itself: it only has to serve the decision, the data, the action, and the shared work.
The tool is strategy made executable
An internal application is not a technical object. It is an organizational decision made concrete: who sees what, who approves what, in what order. That is why I never start with the tool: I start with the workflow and the profitability of the action we want to automate. The tool is only the last step, the easiest one, once everything else is clear.
What I recommend
Do not bring me a mockup. Bring me the problem: the exact moment things get stuck, what it costs you, and the decision you want to be able to make faster. Building is my job. Framing is where the project is won or lost.