My Adventures Building an AI Money Machine

Part Two: First Build

Almost nobody is using Codex, and that’s a real shame. It is, I’m now convinced, the best way to get serious work done with AI.

Chatbots are great for having conversations, asking questions, and getting answers. But your actual real work? That usually happens in files. Excels and PowerPoints and PDFs sitting in a folder somewhere on your computer. Those files need to be updated, concatenated, re-saved, reviewed and revised.

That’s how work happens in the real world, and chatbots can’t do that. Codex can.

It scared me at first, just like it scares a lot of other non-technical people. But I decided to give it a go, and it was a lot easier than I had anticipated. I didn’t want silly training exercises, so I committed to building something real:

I think online prediction markets are juicy. So I’m using Codex to arbitrage them.

I’m calling this approach the easy hard way to learn Codex. No silly training exercises. Just jump right in to build something that actually matters, but get a lot of help (from AI) along the way.

My primary motivation here isn’t the betting markets, although that does add some excitement. My primary motivation is to master Codex, because I think it’s the most important AI skill to learn right now. And I think new skills are best cultivated by doing real work.

In today’s post, I want to give you a full update on my project.

The cliff notes: I don’t know how to code. I only started using Codex a few weeks ago. And today, I just built this:

For my paid subscribers, I am going to share my step-by-step build process. If you would like to join me in a campaign to predict New York City rainfall better than the online betting markets, you’ll be able to replicate my build using this post.

Or, if you just want to learn Codex, and want to do it working on a project that is meaningful to you, this can serve as a field guide. I’ll share how I approached this, the help that I got from AI, the challenges I encountered, and the lessons I learned.

Note: this article is much more technical than my normal posts, as it is focused on building an analytical model to out-predict online betting markets.

Second note: nothing in this post constitutes financial advice, or should be interpreted as such. I’m just a guy building a thing and telling you about it.

Get your umbrellas ready!

---

For full access to this article, and everything on this Substack, please consider supporting my work with a paid subscription