# Silico - comes from “in silico,” which means an experiment or simulation performed on a computer.

- **Event:** [The Recursive Agents Hackathon](https://cerebralvalley.ai/e/crusoe-recursive-agents-hackathon)
- **When:** Sat, Oct 17 at 9:00 AM – 10:00 PM (PDT)
- **Where:** San Francisco, CA
- **Team:** [Mahin Bharathwaj](https://cerebralvalley.ai/u/Mahinb)
- **GitHub:** https://github.com/Mahin2076/silico
- **Demo video:** https://www.youtube.com/watch?v=ad5UaVHeRBk
- **Gallery:** https://cerebralvalley.ai/e/crusoe-recursive-agents-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/crusoe-recursive-agents-hackathon/hackathon/gallery/49

Silico doesn’t just generate an answer. It checks whether the work meets the goal and shows what remains unresolved.

Silico lets you ask a question and explore how a simulated city might respond. For example, “Would more people buy a vegetarian lunch at $8 or $10?” You can compare the results on a map and click on synthetic residents to see their modeled responses. The population is based on Census data.

I built this for the “Done Means Done” challenge. Before running an experiment, the planner defines what it will test and what needs to be checked. The simulation estimates responses, then automated checks look for missing answers, invalid probabilities, and calculation errors. NVIDIA Nemotron reviews whether the results actually answer the question and whether the conclusions are supported.

If something is missing, the workflow can try a targeted fix or use You.com to find supporting research. It checks the new evidence before moving on. If it still cannot pass the checks within the retry limit, it stops and shows what went wrong. You can inspect the results and review history instead of having to trust a “done” message.

My hackathon work includes the experiment runner, independent review and repair loop, research integration, interactive results, and cloud deployment. I built on the existing Simtra map and population engine and credit that earlier work. The responses are hypothetical estimates, not real interviews or proven predictions of customer behavior.

Try it here: https://silico-livid.vercel.app/?view=city&city=sf&ws=railway-check

The backend runs on Railway without my laptop. Live runs share an API budget cap, and saved results stay available to explore.

## More from The Recursive Agents Hackathon

- [Silk Lab](https://cerebralvalley.ai/e/crusoe-recursive-agents-hackathon/hackathon/gallery/46)
- [EvoAction](https://cerebralvalley.ai/e/crusoe-recursive-agents-hackathon/hackathon/gallery/47)
- [Clueso](https://cerebralvalley.ai/e/crusoe-recursive-agents-hackathon/hackathon/gallery/48)
- [AGI Factory](https://cerebralvalley.ai/e/crusoe-recursive-agents-hackathon/hackathon/gallery/50)
- [Proofreaders](https://cerebralvalley.ai/e/crusoe-recursive-agents-hackathon/hackathon/gallery/51)
- [CampusLoop](https://cerebralvalley.ai/e/crusoe-recursive-agents-hackathon/hackathon/gallery/52)

---

Markdown version of https://cerebralvalley.ai/e/crusoe-recursive-agents-hackathon/hackathon/gallery/49. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt