AI Is Useful. But It Doesn't Think Like We Do.

After 90 sessions building a project with three Claude instances, here's what I actually learned about artificial intelligence.

I should start with a disclaimer: I'm not a developer. I'm not an AI researcher. I don't have a computer science degree. Three months ago I didn't know what an API was.

What I do have is 90 working sessions with Claude where I've used three separate instances to build, run, and document a crypto trading project. One acts as CEO (Claude Projects — strategy, briefs, database access), one writes code (Claude Code), one handles daily automation (Haiku). I'm the human in the middle.

After three months of this, I have a very specific opinion about artificial intelligence: it's incredibly useful, it's better than me at most individual tasks, and it doesn't actually think.

Let me explain what I mean.