I built this whole portfolio with AI. You can too — you just have to start.

Thirteen case studies. Almost no hand-written code. Six months of evenings, and about $40 of compute. Built entirely with AI tools.

I'm not telling you that to impress you. I'm telling you because most designers I talk to believe a thing that isn't true: that building with AI is a technical skill they don't have, a wall they can't climb without learning to code first. It isn't. The wall is mostly in your head. The only real difference between the people shipping with these tools and the people watching from the sidelines is that one group started and the other is still waiting to feel ready.

You will not feel ready. You learn this by doing it, getting it wrong, and doing it again. That's the whole method. This post is the encouragement and the proof.

Try it, don't study it

Here's the trap. You watch someone build something with AI and it looks like magic, so you assume there's a body of knowledge between you and them — frameworks, terminals, commands, things that break. So you decide you'll learn all that first, and then you'll build. And "first" never arrives, because there's always more to learn, and the learning has no end and no stakes.

That's backwards. You don't study your way in. You build your way in.

When I started, I didn't know how half of it worked. I still don't know how a lot of it works. What I knew was what I wanted the thing to look like and feel like — and that turned out to be the part that actually mattered. The AI handles the how. Your job is to know what "good" looks like and to keep asking until the screen matches it.

You describe what you want in plain English. You look at what comes back. It's wrong in some way. You say what's wrong. It tries again. You repeat that loop until it's right. That's it. That's the skill. There's no secret step you're missing.

Being bad is the point

The first case study I tried to build was rough. The spacing was off, the colors drifted, things didn't line up. If I'd treated that as evidence I "wasn't technical enough," I'd have quit and gone back to Figma, and you wouldn't be reading this.

Instead I treated it as the first rep. The second case study was better because I'd already made the mistakes once. The third was better than that. By the time I'd done a few, I had a feel for how to ask, what to ask for, and what to do when it came back wrong.

None of that came from a tutorial. It came from being wrong, in the actual thing, and fixing it. You cannot shortcut that with reading. You can only get it by starting and being a little bad at it for a week.

That's not a flaw in the process. That is the process. Every person who's good at this now was bad at it at some point, and the only thing they did that the quitters didn't was keep going past the bad part.

Write your rules down

There's one lesson from all those reps that I'd hand you directly, because it'll save you the frustration that almost made me quit around case study three.

The mistakes started repeating. A color I'd fixed in one case study came back wrong in the next. Widths I'd lined up perfectly drifted again. I kept fixing the same things over and over, and it felt like the tool was ignoring me.

The tool wasn't the problem. I was. I was keeping all the rules in my head and explaining them slightly differently every time. The AI can't read your mind, and it doesn't remember last week's conversation.

So I did something simple: I wrote the rules down in one plain document. The exact colors. The spacing. How I wanted things worded. The order sections should go in. Then I told the AI to follow that document every time.

The repeating mistakes stopped.

That's the real unlock, and it's not technical at all. The AI is the same for everyone who uses it. What makes your work yours is the set of rules you've written down. Two people using the identical tool get wildly different results, and the difference is just how clearly each one has decided what they actually want. Your taste, written down, is the whole game.

You don't need to know this on day one. You'll discover it the same way I did — by getting annoyed that the same mistake keeps coming back, and finally writing it down. I'm just telling you early so you recognize it when it happens.

What I got wrong

I made plenty of mistakes. A few worth naming, not because you need to avoid them — you'll make your own — but so you see that "getting it wrong repeatedly" really is the normal shape of this.

I let the AI put words in my mouth. Early on it wrote punchy little quotes in my voice that sounded like me but that I never said. I caught nine of them in one case study on a re-read and cut every one. Lesson I wrote down: never let a quote ship unless I actually said it.

I over-decorated. These tools can add gorgeous animations and effects, and at first I added all of them. The site started looking like an effects demo instead of a portfolio. I deleted almost all of it. It got faster and clearer immediately. The good version came from what I removed, not what I piled on.

I buried my best work where no one would see it. I kept putting my strongest evidence deep inside long pages a busy reader would never reach. The fix was just to move the good stuff up front. Obvious in hindsight. Invisible until I'd done it wrong a few times.

Every one of those came from doing the thing, not from planning the thing. You will not think your way to these lessons. You'll bump into them, and bumping into them is how you learn.

It's cheaper than you think

People assume this costs a fortune. The whole portfolio — thirteen case studies, the home page, the about page, this post — cost less than one afternoon of a freelance engineer. A single case study runs me about an hour of my time and a few dollars.

One small habit keeps it cheap: use the simple, fast version of the AI for the simple work, and save the powerful, expensive version for the moments that actually need judgment. You don't need to overthink this. Most of what you do is routine, and the cheap setting handles routine perfectly well.

The point is that money isn't the barrier and time isn't either. The barrier is starting.

If you're a designer

You can ship. Not "someday, once you learn to code." This week.

Here's the entire plan:

Pick the smallest thing you're working on — a single card, one screen, a simple page. Build it directly with an AI tool instead of in Figma.
It'll come out wrong in some way. Tell the AI what's wrong. Let it try again. Repeat until it's right.
When the same mistake shows up twice, write the rule that prevents it into a plain document, and have the AI follow that document from then on.
Put the finished thing on the internet at a real link.

That's it. That's the whole method, and you already have everything you need to start step one today.

Your advantage over the AI isn't technical — it's taste. You know what good looks like. The tool doesn't, until you tell it. That's not a weakness in your position. That's the entire job, and it's the one part the tool can't do for you.

The close

The only thing standing between you and a portfolio built this way is the decision to be bad at it for a little while.

I was bad at it for a week. Then I wasn't. Nothing about me made that possible except that I kept opening the tool after it disappointed me the first time. There was no moment where I "became technical." There was just attempt after attempt, each one a little less wrong than the last.

You will not know whether you can do this by reading about it — not this post, not any post. You'll only know by trying it, badly, tonight, and slightly less badly tomorrow.

So stop reading. Open a tool. Build the smallest thing you can think of. Watch it come out wrong. Fix it. That's the whole secret, and it's been waiting for you to just begin.