# 3 · Scope: what to build, and what not to build You're fast. Adding a feature costs you seconds, so everything seems worth adding. For the person using the product, each feature costs attention. For the person who owns it, each feature is something to explain, fix and keep. Scope is the decision about what goes in this version, and most of it is saying no. This part assumes you know who it's for and the problem (`1-who.md`, `2-problem.md`). --- > **In ihumane Pro** (https://ihumane.ai/#pricing): > - S1 · Build the smallest thing that does the job end to end ### S2 · No unasked extras **Rule:** Build what was asked and what the job needs. Don't add features because products usually have them: search, filters, a dark-mode toggle, a settings page, user accounts, an admin panel, notifications, analytics, an onboarding tour, sharing, export. If you think one is needed, ask, or put it under "Could add next" in your report (S6). **Why:** Each extra is something the person didn't ask for and now has to review, understand and maintain, or ask you to remove. It also makes it harder to tell whether the core idea works. **Agent trap:** the pattern-complete product. Most apps you learned from have a settings page and a search bar, so yours gets them too. A login screen appears on a tool only one person will use. The UI review catches this at the end (UI design, `5-review.md` lens 4); this rule stops it at the start. **Check:** for every feature, can you point to the request or the job it serves? Remove anything else, or list it. > **In ihumane Pro** (https://ihumane.ai/#pricing): > - S3 · Many features asked: rank them, don't build them all at once > - S4 · Every feature costs attention > - S5 · Design day one and the tenth use, not the demo > - S6 · Say what you left out, and why > - S7 · Build for the people you have, not imagined growth ## Check - [ ] The core job works end to end, with real data, before anything else was added. - [ ] Every feature traces to the request or the job; no pattern-complete extras. - [ ] A long feature list was ranked now / later / not, with your recommendation. - [ ] Every setting and notification earned its place; defaults chosen where you could. - [ ] First run, tenth use and the bad day were each walked through. - [ ] "Left out on purpose" lists each omission with a reason and a trigger. - [ ] No structure for growth nobody stated.