Original Summary

Back Me Up is a tiny science-ish machine that finds published papers that back your argument. It stemmed from a conversation with a friend around life sciences, where we kind of landed at &quot;surely there’s a paper confirming most arguments&quot;, i.e. coffee is bad, coffee is great, etc.<p>You enter a claim, e.g. &quot;drinking wine is good for the heart&quot; -&gt; the system does a semantic search across a corpus (PMC, bioRxiv, medRxiv and arXiv) -&gt; pulls in relevant quotes if they fit your claim, and gives you a verdict.<p>Frontend was done with a combination of Sol + Opus, I spent a bit of time exploring different designs and landed on this slightly ridiculous printer that spits out your evidence.<p>Some notes: - I have no accuracy evals, this was a throwaway thing I built, so no clue on reliability. It does pass the vibe test, so give it a whirl! - Initially built a RAG for a tiny set of PMC, but scaling that up seemed like a lot of work for an evening project. Luckily found an API that does the semantic search across the different sources.

  • 情报分类：技术学习与提效
  • 分类依据：内容涉及技术、AI、软件工具或工程实践
  • 信息来源：Hacker News 新项目
  • 发布时间：2026/9/15 16:57:08