I Knew Nothing About Code. I Built 7 Live Web Projects in 3 Weeks with AI. Here's How.

Let me be honest about my starting point, because it matters. I am not a programmer. I had never written a line of code I understood, never deployed a website, never touched a command line without fear. If you'd shown me a screen full of DNS settings three weeks ago, I'd have closed the laptop.

And yet, over about three weeks, I built and launched seven real projects on the open internet — sites and an app that actually work, that real people can use. I'm telling you this not to brag (most of it was clumsy, and I broke things constantly), but because if you're a non-technical person wondering whether AI tools have genuinely lowered the barrier, the answer is yes — with caveats I'll be straight about. This is the guide I wish someone had handed me on day one.

Where I started: completely lost

My first project was the most ambitious and, in hindsight, the hardest possible thing to start with: a local events platform with a website, a backend database, automated email, and even a bot that could read event posters from screenshots. It took days, I got tangled in moving parts I didn't understand, and I made every mistake going. But it taught me something crucial — the AI wasn't going to do it for me while I watched. It was going to do it with me, one step at a time, if I was willing to be the hands and ask a lot of questions.

By my last couple of projects — a finance calculator portal and a mobile app — I'd found a calm, repeatable rhythm. That shift, from drowning to a system, is the real story. It's available to anyone willing to stay curious and not panic when something breaks.

The mental model that unlocked everything

The single biggest thing that helped me was a simple analogy for how a website actually reaches people. Before this clicked, every technical step felt like magic words I was copying blindly. After it clicked, the steps made sense. Here it is:

Think of putting a website online like opening a shop.

• The domain (like yoursite.com) is your shop's street address. It's what you tell people so they can find you. You rent it for a year at a time from a registrar.
• The host is the actual building where your shop's stuff lives — the files that make up your site. Someone has to physically store them on a computer that's always on.
• DNS is the post office / directory that connects the address to the building. When someone types your address, DNS is what quietly points them to the right building.
• Deployment is simply moving your stuff into the building and putting it on the shelves so customers can see it.

That's it. Almost every confusing step I hit was really just one of those four things: registering the address, renting the building, telling the directory how to connect them, or moving the goods in. Once I had that map in my head, I stopped feeling lost. If you only take one thing from this article, take that analogy — there's a fuller walkthrough in my beginner's guide to domains, DNS and hosting.

What AI actually does for you (and what it doesn't)

Here's the honest division of labour I settled into, and it's the key to the whole thing:

The AI was my tireless expert. It wrote the code, explained concepts in plain English as many times as I needed, drafted the content, and — crucially — talked me through fixing things when they broke. I could ask "explain that like I'm five" endlessly and never get an impatient sigh.

I was the hands, the judgment, and the persistence. I clicked the buttons, ran the steps, made the decisions, and kept going when things failed. The AI couldn't log into my accounts, couldn't see my screen unless I showed it, and couldn't know what I hadn't told it.

AI didn't replace the work. It removed the gatekeeping — the years of training that used to stand between an idea and a live website.

That reframing is everything. I stopped thinking "the AI will build my site" and started thinking "the AI is the most patient teacher alive, and I'm the apprentice doing the actual building." Apprentices make mistakes. That's the job. For getting the most out of that teacher, see my notes on writing better prompts — clear instructions were half the battle.

The seven projects (and what each taught me)

Without naming the live sites, here's the range, because the variety is the point — these are very different kinds of things, and the same approach built all of them:

• A local events platform — my baptism by fire. Taught me that ambitious-first is hard, but you learn fastest in the deep end.
• A recipe-scaling tool — taught me about handling user uploads and keeping secret keys hidden safely on the server, never in the browser.
• A daily word game — taught me the most painful lessons about testing before publishing, and that "it works on my screen" isn't the same as "it works."
• A mental-maths learning site — taught me about user accounts and that some content genuinely needs a human expert, not just AI.
• An AI-tools guide (this very site) — taught me about content, SEO, and being honest rather than hyping.
• A finance calculator portal — by now I had a system: test the logic first, then build the interface. Taught me that calm and repeatable beats heroic and chaotic.
• A mobile app — taught me that the same problem-solving approach stretches well beyond websites.

Seven different problems, one mindset. None of them required me to become a "real developer." They required me to be willing to learn, willing to break things, and willing to ask the AI "why did that happen?" a hundred times.

What nobody tells beginners

A few honest truths that would have saved me a lot of stress:

Things will break, constantly — and that's normal

I had deploys fail, sites show blank pages, a domain that looked "expired" when it wasn't, and features that worked one minute and broke the next. None of it meant I'd failed. Breaking and fixing is the process. The skill isn't avoiding problems; it's not panicking when they appear.

The AI can be confidently wrong

More than once, the AI gave me an answer that sounded completely certain and was completely wrong. It's a tool, not an oracle. I learned to verify anything important and to treat its confidence as a suggestion, not proof.

Patience beats brilliance

I'm not smart in a technical way. What got me through was refusing to quit on a problem and being willing to go one careful step at a time. That's a temperament, not a talent — and anyone can choose it.

Honesty is a feature, not a weakness

On a couple of sites I was tempted to inflate numbers to look established. I learned that real platforms penalise that, and visitors smell it. Being genuinely honest about what something is turned out to be both the right call and the smarter one.

If you wanted to start today

Here's the simplest possible on-ramp, based on what actually worked for me:

1. Pick one small, real idea — something you'd genuinely use. Motivation matters more than ambition.
2. Open a free AI assistant and tell it exactly what you want to build and that you're a complete beginner. (New to that? Start with my AI for beginners guide.)
3. Ask it to teach as it goes — "explain each step, assume I know nothing." Go one step at a time.
4. Expect to break things, and ask "why did that happen and how do I fix it?" every time you do.
5. Learn the four-part map (address, building, directory, moving in) so the deployment steps make sense.

That's genuinely how I went from zero to seven live projects. Not talent — a patient teacher, a clear mental model, and a refusal to give up when the screen turned red.

The bottom line

Three weeks ago, "deploying a website" sounded like something only real engineers did. Now I've done it seven times. AI didn't make me a developer — it made the gate that used to keep people like me out swing open. If you've been on the outside of building things because you "don't know how to code," I'd gently suggest that barrier is lower than it's ever been. You still have to do the work, still have to be patient, still have to push through the broken bits. But the door is open. I walked through it knowing nothing. So can you.

Frequently asked questions