Insights
Notes from the workbench
What we're learning while building — written for people who make product and engineering decisions, without the content-marketing filler. New pieces land as they're ready, not on a content calendar.
Built to be operated, not merely launched
Launch day is the cheapest day your software will ever have. The engineering that matters is aimed at year three.
7 min readWhere AI genuinely improves a product
A field guide to separating the AI features users will rely on from the ones they'll screenshot once and never open again.
8 min readWhen should a mobile app work offline?
Offline-first is neither a checkbox nor a religion. A practical framework for deciding how much offline your product actually owes its users.
7 min readOn the workbench now
Pieces in progress — the thinking is underway, the writing follows.
- Why AI prototypes fail in production Applied AI
- Native or cross-platform: how we actually decide Mobile product engineering
- Designing human approval into AI workflows Responsible technology
- The hidden operational cost of “simple” software Engineering decisions
- How to scope a prototype that answers a real question Product strategy
- Evaluating AI features beyond demo quality Applied AI
- Privacy as a product-design constraint Responsible technology
- Building a company in public, carefully Building in public
- Choosing boring technology for interesting products Engineering decisions
Prefer the applied version?
Everything here comes from real build decisions. Bring us yours.