# Summa summarum, scientia !

- **Event:** [Built with Opus 4.6: a Claude Code hackathon](https://cerebralvalley.ai/e/claude-code-hackathon)
- **When:** Feb 10 at 12:00 PM – Feb 17 at 10:00 AM (EST)
- **Where:** Location TBA
- **Team:** [Jonas Walheim](https://cerebralvalley.ai/u/jwh), [Nives Rombini](https://cerebralvalley.ai/u/nives)
- **GitHub:** https://github.com/j-walheim/Critical-AI-Scientist
- **Demo video:** https://www.loom.com/share/5d8aa55b9dbf4183826cde255f98047d
- **Gallery:** https://cerebralvalley.ai/e/claude-code-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/claude-code-hackathon/hackathon/gallery/154

With AI being increasingly deployed for scientific tasks, we are seeing a surge in scientific hypotheses formulated by AI. However, in clinical research, evaluating hypotheses is expensive with the limiting factors being constraints of the physical world.

In software development we have tools for automatic reviews of pull requests that can make it very easy to triage code generated by AI (or humans) and makes it much easier to integrate AI into the coding workflow. Our idea here was that it would be great to have a similar kind of automatic review of scientific claims states in publications, investor decks, or even free text.

We used Opus 4.6 logical reasoning capabilities and developed a “Critical AI scientist” that searches and surfaces what is already known, and identifies and structures the right context to provide a nuanced view of a hypothesis across many axes.

DEMO: https://critical-ai-scientist.jonas-walheim.workers.dev
user: scicrit
pwd: Xo8gN4DciL1Oont56IDH2mp3

---

Markdown version of https://cerebralvalley.ai/e/claude-code-hackathon/hackathon/gallery/154. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt