Pricing a feature whose cost scales with use, when every pricing instinct in software was formed in a world where the marginal customer was free. The value metric question from Foundations returns here with a cost constraint attached and a much narrower set of right answers.
| # | Lesson | The question it answers |
|---|---|---|
| 01 | Why Seats Break | What happens when cost scales with usage and price does not? |
| 02 | The Value Metric for AI | Tasks, resolutions, outcomes or seats: what do you charge for? |
| 03 | Credits and Usage Pricing | How does a meter work, and how do you keep it from feeling hostile? |
| 04 | Outcome Based Pricing | When is it right, and when is it a trap? |
| 05 | Margin Under Load | What happens when the pricing model and the cost model disagree? |
| 06 | Free Tiers When Inference Costs Money | How much can you give away, and to whom? |
| 07 | Packaging AI Into Existing Plans | Add on, included, or a new tier? |
| 08 | Rate Limits as a Pricing Instrument | How do you cap cost without it reading as a punishment? |
| 09 | Repricing as Model Costs Fall | Token prices drop every year. Who captures that? |
| 10 | Communicating a Usage Price | How do you make a bill a customer can predict? |
| 11 | Enterprise Agreements for Usage Products | What happens when procurement wants a fixed number? |
| 12 | Worked: Pricing an AI Feature | From cost per task to a published price |