I'm not the technical one anymore. AI is how I still ship.
I started as a developer, but these days I spend my time on marketing and product. Field notes on how AI fills the gap, and where I still refuse to let it think for me.
Here is a slightly funny confession for someone who started his career writing code: I do not think of myself as the technical one anymore.
My days are mostly marketing, product decisions, writing, and talking to people. The deep technical work is not where my head lives now.
And yet I still ship things. Pages, fixes, experiments, small tools. The reason is simple: AI.
The gap AI fills
Every founder has a gap between what they can imagine and what they can build themselves.
For a long time, that gap meant one of three things: learn the skill yourself, hire someone, or let the idea die. Most ideas died.
AI changed that math. Now a lot of the “I wish I could just…” ideas get built, because the cost of trying dropped close to zero.
I use Claude more than any other tool for this. Not as magic, but as a very fast collaborator that does not get tired of my questions.
What I actually use it for
Most of my use falls into a few buckets:
- Building small things. Website changes, layouts, small fixes. I describe what I want in plain words, review what comes back, and iterate.
- Thinking out loud. Before a decision, I talk it through. Getting pushback on a half-formed idea is surprisingly useful.
- First drafts of boring work. Structure, outlines, checklists. The parts that need to exist but do not need my personality.
- Learning fast. When I hit something I do not understand, I ask until I do.
Where I do not let it think for me
This part matters more than the list above.
AI is great at producing. It is not responsible for the result. I am. So there are a few places where I keep my own hands on the wheel:
- Opinions. What I believe about product, marketing, and building has to come from me. Otherwise, what is the point of reading me?
- Stories. My experiences are mine. AI can help me tell them more clearly, but it cannot live them for me.
- Final calls. It can suggest. I decide. Especially for anything a customer will see.
If you outsource your thinking, you stop having anything worth saying. If you outsource the busywork, you get more time to think.
The new skill is direction
The most important skill with AI is not prompting tricks. It is knowing what you want.
Clear direction gets good results. Vague direction gets vague results, very quickly. That is true for AI, and honestly it was always true for working with people too.
This is why I think founders who come from product or marketing are well positioned right now. We spend our days defining problems, writing briefs, and judging whether something is good. That is exactly what working with AI requires.
What this means for founders
You do not need to be the best engineer in the room to build. You need to be clear about what you are building, and honest about whether it is working.
AI made building cheaper. It did not make judgment cheaper. That is still the scarce thing.
Post thirteen of writing in public. If you are a non-technical (or formerly technical) founder figuring out how to build with AI, email me. I am happy to share what has worked for me.