All resources
AI June 27, 2026 1 min read

How ML actually works (Module 1)

The mental model everything else hangs on: learning from data vs. writing rules, the train/loss/optimize loop, and why overfitting is the thing to fear.

Module 1 of 13 in the AI Track, a free, self-directed curriculum for building AI that moves revenue.

The idea

Machine learning is finding patterns in data instead of writing the rules yourself. That one shift is the whole thing. The loop never changes. Feed in data, let the model guess, measure how wrong it was, push it to be a little less wrong, run it again. That loop is training. Using the trained model on new data is inference.

A few words worth owning cold. Supervised, unsupervised, reinforcement. Features and labels. Parameters and weights. And the one that quietly kills projects: overfitting. That’s a model that memorized the training data and falls apart the second it sees something new. It’s why you hold back a test set the model never gets to study.

Why it matters

If you can’t explain this simply, you can’t sell it honestly. This is the vocabulary you need to talk to a technical buyer without bluffing. It’s also the ground every later call stands on, starting with the most useful one: is this even a machine learning problem, or a rules problem wearing a costume?

Build this

A 500-word piece called “ML for revenue leaders” that explains the loop with a sales analogy. If you can teach it in plain English, you understand it. If you can’t, you don’t. Publish it somewhere public.

How to work through it

  • Watch the first 3Blue1Brown neural networks video for the intuition
  • Work through the intro units of Google’s crash course
  • Write the explainer and get it clean
  • Write down the one thing that surprised you

Resources

Want this kind of work on your funnel?

Tell me where growth is stuck and I'll tell you where to start.

Get in touch