reddit · SideProject · 11-09 08:50 47 分 ← 返回列表

I built a tool that grills your hackathon project like real judges do (free, no sign-up)

I've won 5 hackathons, and the thing I kept noticing is that what decides the result is rarely on the rubric. It's whether the demo works, whether you can explain the project in one sentence, and how you handle the questions nobody practices. I built Would This Win?. You paste your repo. It reads your code, then three AI judges (an engineer, an investor and a very tired judge) ask you questions, 30 seconds each. Your odds of winning move live with every answer, and at the end you get a code

五维评分

付费意愿

需求真实度

个人可落地

竞争空间

新鲜度

AI 分析

痛点：黑客松参赛者不知道评委真正看重什么，决定结果的往往是rubric之外的东西：demo能不能跑通、能否一句话讲清项目、如何应对那些没人练习过的问题。

目标用户：黑客松参赛者（学生、独立开发者），以及高校、编程训练营、黑客松主办方

证据：what decides the result is rarely on the rubric. It's whether the demo works, whether you can explain the project in one sentence, and how you handle the questions nobody practices.

变现：C端免费做引流，B端向黑客松主办方、高校、训练营做付费（赛前模拟辅导、赛道评分分析、白标定制）；或订阅制解锁多轮模拟、答案复盘和赛后报告。

MVP：作者已有可用 MVP，独立开发者两周内可加：付费墙+团队/教练版、按赛道（AI/Web3/硬件）预置评委人格、赛后复盘报告 PDF 导出、每次模拟后的薄弱点清单。

竞品：通用 AI 模拟面试/简历工具（Interview Warmup 等）、AI mock interview bots

摘要：作者自称赢过 5 场黑客松，做了 Would This Win?：粘贴 repo 后三个 AI 评委（工程师、投资人、疲惫评委）各问 30 秒问题，胜率实时变化，最后给出代码评审式结论和赛前三条改进建议。技术栈 Next.js+Supabase+Claude Haiku，胜率由代码计算而非 AI，完全免费无需注册，发帖求反馈。原帖无用户量、无收入数据，也无任何付费意愿表述，变现能力尚未验证。