Deep Breath

A trading bot was the classroom. AI was the lesson.

I took a deep breath over the last five-plus weeks.

The TradeBot experiment finished its thirty days back in May, and then I went quiet. The bot still scans the market every day. It’s still in the black. The algorithm keeps the P&L positive with conservative parameters, but trades are rare now. Mostly it sits on its hands, which, it turns out, is the most honest thing it does.

I’ve learned, through trial and error, that I do not have the resources to find an edge in the world of trading. I’ll write more about how I know that. It took months of honest testing to be sure, but that’s the short version, and I’d rather say it plainly than dress it up.

Here’s the thing, though: failing to find an edge was a result. Learning to collaborate with AI was the real lesson.

Wrong Question, Right Lesson

Four months ago, I set out to build an AI-powered trading bot. The goal seemed straightforward. Learn AI by building something real.

I wasn’t interested in tutorials or endless courses. I’ve spent most of my professional life learning by doing. I built and ran a café for eight years. Later, I owned and operated a personal chef franchise. When I want to understand something, I build it.

TradeBot became my classroom.

Along the way I learned about large language models, prompt engineering, agents, APIs, workflows, automation, evaluation, and the strange strengths and weaknesses of AI systems. Some days were exciting. Some were frustrating. Most were both.

What surprised me wasn’t what I learned about trading. It was what I learned about AI.

The deeper I went, the less I cared whether a machine could help me beat the market, and the more I found myself circling bigger questions. What work will AI genuinely improve, and where does human judgment still matter? How will small businesses actually use these tools? What happens when people who aren’t programmers start building with them?

And maybe most of all: how do ordinary business owners make sense of any of it?

I’ve spent more than a decade running businesses, and I find I look at AI differently than a lot of the people writing about it. I’m not coming at it as a software engineer. I’m coming at it as a business owner, a writer, and someone who has spent years serving customers, managing employees, solving problems, and trying to make things work in the real world.

That perspective matters more, not less, as all of this accelerates.

So the publication is changing. TradeBot was the classroom, and the class is over. Going forward, I’ll be writing about AI through the lens of business, work, writing, and everyday decision-making. What these tools can actually do, where they fall short, and how small businesses and independent professionals can put them to use.

In a lot of ways this whole thing is a public record of one person figuring it out in real time. But I’ve come to realize my most useful contribution isn’t documenting a trading bot. It’s bringing together the three parts of my background that don’t usually sit in the same room: entrepreneurship, writing, and a growing, hands-on understanding of AI.

That’s what Human In The Loop is about from here.

Next up is a three-part series. I downloaded three AI models small enough to run on a MacBook M1 and worked with AI, in chat and in code, to build a fully local chatbot, then put those models through the same questions to see how they actually think. They don’t think alike, and that turns out to matter.

The tools are getting more capable. The reason to stay in the loop, awake to how they work, only gets stronger.

Thank you for following along this far.