I am not a developer. For most of my working life, that felt like a wall between me and building anything real online.
It turned out the wall was lower than I thought.
Where I started
Before I started using AI tools, I had been doing online marketing for 20 years. When I needed traffic for a website, I mostly bought it through Google AdWords. If I needed a website built, I paid someone else to do it. The only thing I could do myself was change a few words of text inside a simple HTML page. I didn’t know what a database was, I didn’t understand CSS, and I couldn’t read any of the front-end code. All I really knew how to do was register an account. Then AI came along, and everything changed in a way I didn’t expect.
The turning point
This happened just last month. I was talking with Google Gemini in a normal conversation, and I told it: I want to build a website that runs for the long term with zero money invested, but I don’t understand code. So Gemini started guiding me, step by step, to build the site on GitHub. As I kept exploring, I slowly got familiar with how a development project actually works—the workflow, the files, the rhythm of it—and along the way I built a few really practical systems: a project-development knowledge-base system, and a couple of front-end content-generation tools that work through conversation.
What I got wrong
Plain chat mode is not the most convenient way to build a real project. The better way is local mode—give the AI a certain level of permission to read and write files on your machine. It saves you from copy-pasting everything back and forth.
You have to build your own set of interaction rules that match your style and your working habits. Tell your AI assistant how you like to work, set the red lines it must not cross, and build up an AI working knowledge base of your own. Without that, the AI is just guessing.
The biggest pitfall I kept hitting: the context kept getting interrupted. I’d lose the thread of what we’d already done, and then I’d have to spend time explaining the earlier stuff all over again. The fix is rules: you have to use rules to keep your workflow from breaking and to keep the context from being lost. It can be done—but only if you give your AI a way of working that actually fits you.
What actually worked
Google Gemini, used in conversation, was the thing that actually moved the needle. It walked me through building on GitHub when I couldn’t read code at all.
Ghost CMS, about a year ago: I used it to stand up a website and an email subscription, and it took me roughly three months of trial and error to get it working the way I wanted.
Describing what I wanted in plain words, then steering the result until it matched the picture in my head. Slow at first. Much faster now.
What I’d tell you
I can do this. You can do it too. Believe in yourself.
Transparency
I used AI tools to build this site and the things behind it. Some links on this site may be affiliate links; I only point to tools I have actually used. Where I am wrong, I say so. Your results will differ from mine, because your starting point and your project are your own.