A mastery-based course for business students enabled by AI

Good Decisions at Any Scale

Making good decisions is a skill. Like any skill, you build it one honest rep at a time, working at the edge of what you can do today. This course trains five of these skills. Each one builds on the one before it.

1. Lane 1

   Put your work in code. A decision you write as a program keeps working while you sleep. It can run a thousand times a day, and it never gets tired or forgets a step. But a program will repeat a bad decision just as tirelessly as a good one. So first get the tools running, then learn what makes a decision good.

2. Lane 2

   Reason under uncertainty. What makes a decision good? Not the outcome. You have to choose before you know how things will turn out. Probability is the set of rules for choosing well anyway.

3. Lane 3

   Simulate what you can't work out by hand. Real problems have too many moving parts to reason through step by step. So let the computer play out thousands of possible futures and watch how each choice tends to end.

4. Lane 4

   Draw what causes what. A simulation is only as good as the story behind it. Causal diagrams make you draw that story, so your assumptions are out in the open and the data cannot quietly mislead you.

5. Lane 5

   Make it land. A sound analysis changes nothing until someone acts on it. Visualization turns your reasoning into something a busy decision-maker can see, believe, and act on.

Put all five together and no decision is out of reach. It can affect one person or millions, happen once or repeat automatically all day long. That is what it means to make good decisions at any scale.

This course runs like a swim class: 22 stations across 5 lanes. Work the station in front of you until you master it. Mastery means passing a timed skill test with numbers generated for you alone. Each lane ends with you deploying a live, automated decision-maker to your own website. Your grade is the highest lane you complete. Nothing is averaged; nothing is hidden; everything is retryable.