Three years of vibecoding: building a one-person portfolio with AI (2026)

Three years ago, AI changed what I could build. But that’s not the part that mattered. The real shift was quieter and more useful: I stopped thinking like a maker and started thinking like an investor of my own effort. This is what three years of building with AI actually taught me — written from the chair of someone with an economics background who treats time and energy as the only capital a solo operator really has.

The first thing AI fixed wasn’t code

Before any of the building, there was a mess to clear. Burnout, a business partnership that had run its course, the honest question of where the energy was even going to come from. Oddly, the first thing AI did for me wasn’t technical — it was a thinking partner through that. Hours of working out, in plain words, what I actually wanted and why. I don’t romanticise it, but I won’t pretend the human layer wasn’t the real starting line. Only once that was cleared did the building make sense.

Then I did the unglamorous thing: a portfolio review. Pulled every idea I’d been carrying for years, looked at each one honestly, chose a set worth backing, and started shipping. Some returned something. Some I closed. That’s not failure — that’s the job.

The unlock was leverage, not coding

Here’s what people get wrong about “AI built it.” The win wasn’t that the machine wrote code. The win was leverage — dozens of hours of routine I used to grind by hand, gone. When the cost of running an experiment collapses, you can place more bets. An economist would say AI lowered my cost-of-experiment to near-zero; in plain terms, I could try ten things in the time one used to take.

What that let one person actually do, in practice: wire up banking and card payments, ship products and landing pages in 15 languages for a global audience, and put out project after project — the kind of work that used to require an agency or a small team. The full map of where AI fits in that workflow is in how to use AI to run a one-person business, and the builder’s stack itself in the best AI website builders and the best AI tools for solopreneurs.

Treating effort like capital

This is the frame that changed everything. Your time and energy are the only capital you’ve got as a solo operator, and a portfolio approach beats a perfect-idea approach:

• Place several small, cheap bets rather than one big one. AI makes each cheap to run.
• Expect most to underperform. A few will carry the rest — that’s normal portfolio maths, not a personal failing.
• Kill the ones with no signal without sentiment, and concentrate effort on what’s working.
• Favour assets that compound — things that keep paying after the work is done — over pure time-for-money. (The how-few-customers-you-need version is in the mathematics of a solo business.)

It’s the same discipline an investor applies to money, pointed at your own hours. The goal isn’t to ship the most; it’s to allocate limited effort where the expected return is highest — toward either passive income that compounds, or a clean exit: selling the asset you built (which only works if the books are clean — the logic is in how solopreneurs make money).

What AI didn’t fix — and this is the whole point

Here’s the honest ending, and the reason this site exists. AI made building almost free. It did nothing for the two things that actually decide whether a solo project earns: judgement (building the right thing — AI will confidently build the wrong thing all day if you let it) and distribution (getting it found and paid). The bottleneck didn’t disappear; it moved. When everyone can build, the value shifts to deciding what to build and getting people to it — which is exactly the wall most solos now hit.

That’s the quiet truth under the AI hype: the cheap part got cheaper, and the hard part got harder to compete on. Three years of vibecoding didn’t make me a better coder. It made me a better allocator — of attention, of effort, of which bets are worth a human’s scarce time. And it’s why a one-person business in 2026 is less about building and more about thinking like an investor who happens to ship their own projects.

For the practical model — the best vibecoding workflow, and how it differs for a beginner versus an experienced projectologist — see vibecoding for solopreneurs.

Pick the bets, build them cheap with AI, then put the real effort where it counts — find your path.