# TraceMint AI Analyst

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Rishi Sahoo](https://cerebralvalley.ai/u/rishilifts)
- **GitHub:** https://github.com/rishilifts/tracemint-ai
- **Demo video:** https://drive.google.com/file/d/1j8hoG3iJvs9REs-im79fwEt8OfAazzAl/view?usp=sharing
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/140

Grounded weekly revenue answers from raw payment events - deterministic numbers, worded by Meta's Muse Spark.

TraceMint answers the question every solo founder asks: where did the money go this week? Raw commerce events (charges, refunds, disputes) are ingested with their acquisition source. A deterministic TypeScript engine computes gross, refunds, net, refund rate and per-source breakdowns. Then Meta's Muse Spark (muse-spark-1.2) via the Meta Model API writes a plain-English weekly digest from those numbers. Figures are checked against the computed summary before the digest is shown; a digest that fails the check is withheld. The model receives the computed summary, not raw events.

## More from Meta Global AI Developer Hackathon

- [Vivek's Team](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/137)
- [FORGE-0](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/138)
- [masasi katano](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/139)
- [Twix](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/141)
- [PicoCare](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/142)
- [mabarint](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/143)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/140. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt