Zhihu — Start with a narrow wedge, then earn the right to expand
Sangam Team
Nov 5, 2026
Early AI products rarely win by trying to serve everyone at once. The Zhihu-published interview about Granola’s startup story is useful because it frames product building as a sequence of focused bets: identify a painful moment, build a tool around that moment, then learn from the people who return. The practical lesson is not that every founder should copy a meeting-notes product. It is that a first user group gives a product its sharpest feedback. A narrow audience makes it easier to observe what people actually do, which parts they repeat, and what they would miss if the product disappeared. Those observations are more valuable than a broad list of hypothetical features. Positioning also becomes clearer when it is anchored in a job. “AI for work” is too wide to guide a roadmap. “Help a person prepare for and remember important meetings” is specific enough to shape onboarding, retention questions, and product boundaries. From there, adjacent use cases can be earned rather than assumed. Founders should treat early usage stories carefully. A source article can describe a company’s trajectory and philosophy, but it is not automatically a universal growth benchmark. The responsible approach is to extract the decision pattern: start with a concrete workflow, observe repeated behavior, and use evidence from real users to decide what to build next. Source attribution: This editorial draft is based on the original source and is an independent summary, not a reproduction. Original source: https://www.zhihu.com/en/article/1891583700526937028 Discussion questions: 1. What is the narrowest user problem your current product solves? 2. Which repeated behavior would prove that your first users genuinely need it? 3. When should a startup expand beyond its initial positioning?