
It was late, and I was staring at a portfolio I had not touched in five years. The work on it was...
It was late, and I was staring at a portfolio I had not touched in five years.
The work on it was real, but the site no longer looked like the person who made it. Five years ago, fixing that meant one of two things: pay a developer and wait, or carve out a month of evenings to wrestle with code myself. That night I did neither. I opened a blank project, and I started directing AI to build the thing for me.
By the time I went to sleep, meravish.com was rebuilt.
Most people describe AI as a helper. A tool that speeds you up. That framing is too small, and it keeps you in the passenger seat.
Here is the framing I work from instead: I design the product, and I make it. I define the vision, the design system, the user experience, the voice, and then I take it all the way to something live. AI is not my assistant and it is not my substitute. It is the team I direct so that the making stays in my hands. That is a different job, and it is the job I actually want.
For fifteen years I have been the person who decides how a product should look and feel. What changed is not my role. What changed is that the gap between my decision and a working product collapsed from weeks to hours.
I want to be honest about the stack, including the parts that were not smooth.
I used Claude to think through structure and write the code, Figma to hold the design decisions, and Cloudflare to ship it. None of this was a clean straight line. I redirected the AI constantly. I threw out a layout that looked impressive and said nothing. I rewrote the same hero section more times than I want to admit, because "good enough" is not a thing I accept on my own name.
That is the part people leave out of these stories. Directing AI is not magic and it is not hands-off. It is fast, but it still demands taste, judgment, and the willingness to say "no, again" until the thing is right. The AI gave me speed. The standards were still mine to set.
When I say there was no developer in between, I am not being dismissive of developers. I am describing a real change in how the work flows.
For most of my career, my design left my hands and entered someone else's. Something was always lost in that handoff: a spacing decision ignored, an interaction flattened, an intention I could see but could not enforce. The handoff was the place where my vision went to be negotiated.
Removing the handoff means the thing I imagine and the thing that ships are finally the same thing. That is freeing. It is also more demanding, because there is no one left to blame and no one left to translate. If a detail is wrong, it is wrong because I let it be wrong. I find that I work better this way.
Here is the part I did not plan. The case study for the rebuild basically wrote itself, because I was documenting the build as it happened. I was not reconstructing a project after the fact. I was capturing decisions in real time, which is the only way a case study ever feels true.
So the portfolio is not just a gallery of finished work anymore. It is proof of a way of working. The medium and the message finally match: a site about directing AI, built by directing AI.
The barrier was never the tools. The tools are good enough now, and they get better every month. The barrier is the mindset. It is the move from "AI helps me do my job" to "AI is how my job gets done, and I am the one directing it."
If you are a designer who already opens an AI tool every day, you are closer to this than you think. The next step is not learning to code. It is learning to direct: to hold the standard, to give clear creative instruction, and to keep saying "again" until what ships is exactly what you saw in your head.
I rebuilt my portfolio in one evening. The portfolio was the small win. The real win was proving, to myself first, that the person who designs the thing can also be the person who makes it.
You can see what came out of that evening at meravish.com.
Merav Shaked (Meravish) is a product designer and maker based in Israel. She designs products and then makes them, with no developer in between, drawing on 15+ years across UX/UI, branding, art direction and motion, including four years as Chief Design Officer at iCount. See her work at meravish.com or connect on LinkedIn.