00Insights
Notes from the work.
Plain-language writing on deciding what to build, what it should cost, and what it takes to keep software working after the launch applause dies down.
01Index
N-15 Building got cheap. Deciding didn’t. AI collapsed the cost of writing code. The expensive part is now knowing what to build — and that part got quieter, not easier. N-14 Launch is a midpoint, not a finish line The four ongoing costs of custom software that almost nobody budgets for — and how to plan for them before you build. N-13 The demo is not the product Why the prototype that dazzled on Thursday dies on Monday, and what actually changes in between. N-12 Ship the smallest useful thing Every feature in version one is a bet placed with the most expensive money in the project. Sequencing is how you win the bet. N-11 Security without a security team A small company can’t do everything about security. It doesn’t need to — it needs four things, done consistently. N-10 The intern test Hand any system to the newest person and watch. What breaks in their first week is the software’s real report card. N-09 Someone has to own it Software without an owner works, mostly, and nobody knows why. Every system needs a name on it — not a team, a name. N-08 The integration is the product The screens are the easy fifth of the work. The rest is making your systems agree with each other — and that part decides everything. N-07 Your company runs on email The most important software at most companies is the inbox. Approvals, prices, and decisions live there — which works, until it doesn’t. N-06 Why the cheap quote is expensive Send one brief to three builders and the quotes come back far apart. The gap is rarely efficiency — it’s what each quote decided not to count. N-05 Judge a builder by their failures Every portfolio is a highlight reel. What a builder says about the projects that went wrong tells you more than any demo. N-04 The model is the easy part AI projects don’t usually fail at the model. They fail at the data — the one part that is specifically yours. N-03 When not to build custom Custom software is the most expensive way to solve a problem that already has a product. Sometimes that’s the point — usually it isn’t. N-02 The spreadsheet that won’t die Every company has one. It refuses to die because it works — and that deserves respect before anyone tries to replace it. N-01 Automate the boring thing first Everyone points AI at their most impressive problem. The right first project is the dull one — here is how to tell which is which.