在菲律宾推出了《书籍摘要》——了解到用户希望内容更深入三倍。
我几天前在Product Hunt上推出了Book Digest（AI书籍摘要）。<p>收到的反馈很明确：摘要太短（约800字）。人们期望达到Blinkist级别的深度（2500字以上）。<p>我花了两天时间调试OpenAI的JSON解析、Prisma数据库持久性和令牌限制。然后，我在一夜之间用改进的AI提示重新生成了450本书的摘要。<p>结果：摘要深度提升了2-3倍，包含详细的章节、见解和行动项。<p>演示（无需注册）：https://book-digest.com/books/6c8e5031-1c55-4bdd-8c11-ae5338f374c7?ts=1<p>关键经验：快速推出，但要倾听反馈并快速迭代。用户会告诉你缺少什么。<p>技术栈：Next.js、Postgres、OpenAI GPT-4o-mini、Stripe。<p>欢迎提问关于技术挑战的问题（数据库持久性错误尤其棘手）。
查看原文
I launched Book Digest (AI book summaries) on Product Hunt a few days ago.<p>Clear feedback: summaries were too short (~800 words). People expect Blinkist-level depth (2500+ words).<p>Spent 2 days debugging OpenAI JSON parsing, Prisma database persistence, and token limits. Then regenerated 450 books
overnight with an improved AI prompt.<p>Result: 2-3x deeper summaries with detailed chapters, insights, and action items.<p>Demo (no signup): https://book-digest.com/books/6c8e5031-1c55-4bdd-8c11-ae5338f374c7?ts=1<p>Key lesson: Launch fast, but listen to feedback and iterate quickly. Users will tell you exactly what's missing.<p>Tech stack: Next.js, Postgres, OpenAI GPT-4o-mini, Stripe.<p>Happy to answer questions about the technical challenges (the database persistence bug was particularly tricky).