Build your first project
This is the part you came for. Pick one of the four directions below. Each one starts from something you already have, and comes with one I built so you are never staring at a blank screen. Do them in any order, or just do the one that sounds like you.
Your own data
You are almost certainly sitting on a rich set of data about your own life and have never really looked at it. Tesco and Amazon keep every order. Your ChatGPT history is downloadable. So are your bank statements, your emails, and your wearable stats. You understand this data, you actually care about the answer, and it never has to leave your computer. That makes it the best possible place to start.
Grocery, scored for processing
I mapped years of grocery shopping to the scale that ranks how processed food is: spend over time, category breakdowns, and how processed the basket really was. Built from my own Tesco export, shown here with made-up data.
ChatGPT history, mapped
What you actually ask AI for, how your topics shift across a year, and the hours you lean on it most. The same idea, a different question.
Wearables, combined
Sleep, recovery, and steps from different trackers in one place, with a tool to pick any two and see how they move together. The questions no single app will answer.
Do this with your own data
Pick one dataset you can download: your supermarket orders, your ChatGPT history, your bank export, a wearable. Choose the one you are most curious about.
Export it as a file. Most services have a "download my data" option in settings.
Hand it to Claude and ask for a simple dashboard that answers your question. Describe what you want to see, not how to build it.
It stays on your computer the whole time. That is the quiet benefit of building your own.
Build a need
Sometimes the app you want does not exist, or every version of it gets one thing wrong. The usual answer is to put up with it. The new answer is to build your own, exactly the way you want it.
I did this with a to-do app I call Auri. I had tried every task app going and none of them fit how my head actually works, so I built one that does. It is shaped around me: how I sort things, what I want to see first, the bits other apps got wrong. It is a personal tool rather than a public product, which is rather the point. The best version of this is the one made for an audience of one.
Do this for your own need
Name the thing no app gets right for you. The smaller and more specific, the better.
Describe it to Claude in plain words: what it should do, what it should show, how it should feel.
Build the smallest version that is useful. One screen that works beats ten that do not.
Use it for real, notice what annoys you, and ask for changes. Yours improves because you live in it.
Personal branding
Your work and your story are scattered: a CV here, some posts there, a half-finished profile, years of things nobody has ever pulled together. Doing something coherent with all of it is genuinely overwhelming when you are not a developer. That was a wall for me, and AI took it down. I could pull the scattered pieces into a real, working website, and keep it sounding like me rather than a template.
This very site is the example. It is built, hosted, and updated entirely through AI. I also keep a more formal site for my professional work, made the same way.
Build your own corner of the web
Gather the scattered pieces: what you do, what you have made, how you want to come across.
Decide the one thing a visitor should remember. A site with a clear message beats a busy one.
Ask Claude to build it, page by page, in your own words. Show it writing you like, so it sounds like you.
Put it online with a free host. The Resources page has the links to do that.
Just for fun
Not everything has to be useful. Some of the best things to build are the ones you make purely because you can. It is also the lowest-pressure way to learn, because nothing is riding on it.
I built a small app to solve a very specific annoyance: most curly-hair advice is written for the United States, and I could never tell what I could actually buy here in the UK. So I made a tool that checks ingredients and points to products I can find in British shops. I also build little browser games now and then, for no reason other than that I enjoy it.
Curly Girl, UK edition
An ingredient checker, a hair-type quiz, and a product guide built for what is actually on UK shelves, from Poundland to Boots. A real annoyance, solved for myself.
Make something just because
Pick something you would actually enjoy: a game, a generator, a daft little tool, a tribute to a hobby.
Describe it to Claude and let the first version be rough. Fun does not need to be polished.
Keep tweaking it until it makes you smile, then share it or keep it to yourself. Either is fine.
The thread
Every one of these starts the same way: with something you already have, your data, a need, your story, or just an idea, and a conversation with AI to turn it into something real. You do not need to be a developer. You need a reason to start. Pick a stream and go.
Get the links you needBuilt something? Take it further with me
You did the hard part on your own, which is exactly the point. If you want a hand turning your first build into the real thing you actually want, or you hit a wall along the way, we can do it together, one to one.
Work with me