Insights
Working notes from production.
Opinion and method from building and running AI products: what they cost, how they get found, and what breaks.
- 01
Cost · Unit economics · Software and SaaS engineering
What an AI product actually costs to run
Why the demo lies about cost, the five lines on the ledger, the multipliers to model before you build, and the governance that keeps the bill flat.
- 02
Growth · Search · Growth engine
AEO: how to get cited by ChatGPT
How answer engines choose what to cite, and the answer-first pages, entity consistency, crawler access and structured data that get a site quoted.
- 03
Engineering · Agents · AI products and agents
Agents in production: what breaks
The gap between the demo and the long tail: what breaks when agents meet real traffic, what holds, when multi-agent helps, and the order to ship in.