# Reflexes

- **Event:** [The Harness Engineering & Model Wrangling Hackathon](https://cerebralvalley.ai/e/mongodb-nyc-hackathon)
- **When:** Sat, Sep 26 at 9:00 AM – 10:00 PM (EDT)
- **Where:** New York, NY
- **Team:** [Amey Borkar](https://cerebralvalley.ai/u/ameyborkar771), [Meet Bhanushali](https://cerebralvalley.ai/u/MeetBhanushali)
- **GitHub:** https://github.com/Amey771/reflexes
- **Demo video:** https://drive.google.com/file/d/1jdCyouNxa7JHIP45sKMwOq8E8JSqFSij/view?usp=drive_link
- **Gallery:** https://cerebralvalley.ai/e/mongodb-nyc-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/mongodb-nyc-hackathon/hackathon/gallery/18

Reflexes: agents that grow reflexes.

The problem: AI agents send every small decision (classify, prioritize, route, check policy) to a slow, expensive LLM, forever. Cost grows with every request and latency never improves.

What it does:
• Shadow: each decision starts on an LLM while Jev, TypeSafe AI's System One model, answers in parallel. Every outcome is stored in MongoDB.
• Graduate: once Jev proves reliable on a decision, the harness promotes it to a ~250 ms reflex, with a confidence floor it sets itself.
• Recall before acting: Atlas Vector Search checks the alert resembles cases where that reflex has proven itself. Novel, low-confidence or unproven cases go back to the LLM.
• Demote and rewrite: audits and confidence collapse demote reflexes. An evolver rewrites the reflex's own question (new categories, context policy), and it relearns. Every change is a versioned harness document in MongoDB.

Results (SOC alert triage, 6 decisions per alert, 360 alerts):
• 81% of decisions moved off the LLM within ~24 alerts
• 2.6× lower cost per alert
• 38% of alerts handled with no LLM call, at 257 ms
• Accuracy 96% → 92% on held-out labels the engine never sees (a trade we show openly)
• When a new campaign of attacks on AI agents appeared, the harness flagged it as unfamiliar, handed it to the LLM, and added a new category to its own taxonomy, with no human in the loop.

Live app: https://reflexes-omega.vercel.app (replay the run, paste an alert into Try it, or follow one alert through the full engine in the Sandbox)

Tracks: Recursive Harnessing (primary), Long Horizon Engineering.

## More from The Harness Engineering & Model Wrangling Hackathon

- [Converge](https://cerebralvalley.ai/e/mongodb-nyc-hackathon/hackathon/gallery/15)
- [Levy](https://cerebralvalley.ai/e/mongodb-nyc-hackathon/hackathon/gallery/16)
- [CtrlAltElite](https://cerebralvalley.ai/e/mongodb-nyc-hackathon/hackathon/gallery/17)
- [MongoHive](https://cerebralvalley.ai/e/mongodb-nyc-hackathon/hackathon/gallery/19)
- [Ada](https://cerebralvalley.ai/e/mongodb-nyc-hackathon/hackathon/gallery/20)
- [PolicyPilot](https://cerebralvalley.ai/e/mongodb-nyc-hackathon/hackathon/gallery/21)

---

Markdown version of https://cerebralvalley.ai/e/mongodb-nyc-hackathon/hackathon/gallery/18. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt