Reading
Following questions and ideas that might change how I see a problem.
Updated October 2026
What I am on about
If this page is stale, so is everything on it.
Where the hours go
Three
Product discovery
Looking for unmet needs and deciding which are worth making a product for.
Working with agents
Directing multiple agents in parallel, defining tasks, and reviewing what comes back.
Working bets
- The distance between a demo and adoption is becoming a first-class engineering problem.
- Agent products will be bounded by management, not by model capability.
- Cheap infrastructure buys the runway to be wrong twice.
Open questions
- What does an agent product owe a user when it silently reconstructs the past?
- Where does taste stop being tacit and start being specifiable?
- Is a productised service a better first move than SaaS in a market this young?
Any of these yours?
Then the email writes itself.