A real person, with a real portfolio.

Who I am

I'm Ben Dixon. I've been investing my own money for over a decade, primarily in UK equities, with a focus on smaller companies and situations where the market hasn't fully priced in what I think I can see. I'm not a professional fund manager. I don't manage anyone else's money.

I started experimenting with AI tools in my investment process because I was sceptical of them — and I find the best way to form a genuine view is to actually use the thing under real conditions. This site is where I write up what I find.

Real decisions, not hypotheticals

The experiments on this site are run against my actual portfolio. When I write about a position, it's a real position. When I test whether an AI agent is useful for a buy decision, there's real money attached to that decision.

I'm not going to publish my full allocation or specific position sizes here — that's not the point. The point is to document the process: what I asked the model, what it said, whether the output was actually useful or just plausibly-worded noise.

How I think about this

• — Evidence before opinion. Any AI output that skips straight to a verdict without first surfacing the facts it's drawing on is useless. I build prompts to force the evidence step.
• — Scepticism is the feature. I don't use AI to confirm what I already think. I use it to argue against me, find the downside, and surface what I might have missed.
• — Small scope, real output. The most useful automations I've built do one specific job well. The ones that try to do everything produce impressive-sounding nothing.
• — Write it down. The discipline of writing up an experiment — what I asked, what I got, what actually happened — is where most of the learning happens.

What this site is not

This is not financial advice. Nothing on this site should be taken as a recommendation to buy or sell anything. I'm documenting my own process for my own purposes; if you find it useful, good, but you're responsible for your own decisions.

I'm also not claiming that AI is reliably useful for investing. My working hypothesis is that it's useful in narrow, well-defined situations with careful prompting, and actively misleading in others. The experiments here are how I try to tell the difference.