Publio started from a very ordinary conversation.

My father had been getting more active online and at a family dinner we started talking about him having a proper website. A place for his writing and photography. My first instinct was to make the site for him. That’s the usual pattern: someone close to you needs a website and suddenly it becomes your problem.

But then a more interesting question came up: why should he need me for that at all?

By then, AI coding tools were already good enough that this no longer felt hypothetical. If he could describe what he wanted in plain language, maybe he should be able to create the site himself, shape it the way he wanted and update it whenever he wanted without having to bug me every time a paragraph changed.

That was the beginning of Publio.

What pulled me in early was how quickly the first version became real. In about a day, I had a basic flow working. People could write a couple of paragraphs about who they were and what kind of site they wanted and a few minutes later they had something that looked surprisingly good. The combination of speed and quality was the exciting part. From the first interaction to a usable website in under five minutes felt amazing.

That first version gave me one of those rare builder moments where the gap between idea and outcome suddenly feels much smaller than expected. People who saw it were genuinely impressed. So was I.

Publio also sharpened something I already suspected about where software is going. I was already deep enough into AI to know the capabilities were moving fast, but this project made the implication feel more concrete. A lot of the software we use every day may stop existing in its current form. Instead of fixed apps with rigid interfaces, I think more of our day-to-day tools will become interfaces created and managed by AI systems around a user’s intent. Maybe that won’t apply to safety-critical or high-risk software, but for a lot of normal use cases, it feels increasingly plausible.

Publio felt like a small glimpse of that.

The harder part, in the end, wasn’t building it. It was distribution.

Getting the first users was difficult and it pushed me into an area I’m not naturally good at and, at the time, didn’t have much patience or energy for: marketing. I tried a few things, but the next obvious step was going to be making content and testing channels like Instagram or TikTok ads. That may have been the right move, but it was also the point where my energy for the project started to thin out.

That changed how I saw Publio’s future. I still think the core idea is real. I still think there’s something powerful in letting someone create and maintain their own site just by describing it. But I’m also realistic about the market. This is the kind of space larger players can absorb very easily and that makes it harder to justify a huge investment unless there’s a very specific wedge or distribution advantage.

So I feel mixed about Publio now.

I’m still proud of how directly it answered a real need and of how fast the early product got to something that actually felt magical. But it also taught me that having a compelling product idea is only part of the story. If you can’t get it in front of the right people, or if you don’t have the energy for that part yet, the project hits a different kind of wall.

Still, I’m glad I built it. It started with a simple question about my father’s website and it ended up teaching me something larger about software, interfaces and the parts of building that are less fun but just as real.