# 3000

- **Event:** [RAISE Summit Hackathon](https://cerebralvalley.ai/e/raise-summit-hackathon)
- **When:** Jul 4 at 9:00 AM – Jul 5 at 7:00 PM (GMT+2)
- **Where:** Paris, France
- **Team:** [yashwanth krishna](https://cerebralvalley.ai/u/yashwanth)
- **GitHub:** https://github.com/yashwanth-3000/patient-information-software-ai
- **Demo video:** https://youtu.be/dV-YGESVPkc
- **Gallery:** https://cerebralvalley.ai/e/raise-summit-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/raise-summit-hackathon/hackathon/gallery/30

My dad is a homeopathy doctor with two assistants - one packs medicines, the other spends her entire day manually typing handwritten prescriptions into the clinic's 20-year-old Patient Information System, and the backlog still takes days or weeks to clear. The clinic has trusted that one software for two decades, so instead of building a new app and confusing them, I patched their existing legacy software with a small API bridge and put a crew of five AI agents on top of it. This is a real multi-step agent workflow, not a single retrieve-then-answer call: the agents plan, retrieve repeatedly, call tools, and decide - OCR with alternates for every ambiguous digit, live patient verification against the clinic's actual database, medicine grounding with VultronRetriever rerank, and a final ready-or-review decision. Everything runs on Vultr end to end: agent reasoning on Vultr Serverless Inference (DeepSeek), grounding on VultronRetriever, and the whole backend - the agent crew, the PIS API bridge, and the clinic queue - hosted on Vultr Compute. No value is ever invented: every field traces to OCR evidence, a live database record, or a corpus citation, and a human approves each entry with a swipe before anything is saved. It was tested on my dad's real handwritten prescriptions against his live 20-year patient database - take a photo, and minutes later the entry is inside the same old software the clinic has always used. Full architecture and agent pipeline details on GitHub: https://github.com/yashwanth-3000/patient-information-software-ai

## More from RAISE Summit Hackathon

- [Tara Chand](https://cerebralvalley.ai/e/raise-summit-hackathon/hackathon/gallery/27)
- [The Peak](https://cerebralvalley.ai/e/raise-summit-hackathon/hackathon/gallery/28)
- [Zavalab](https://cerebralvalley.ai/e/raise-summit-hackathon/hackathon/gallery/29)
- [OldManGrooving](https://cerebralvalley.ai/e/raise-summit-hackathon/hackathon/gallery/31)
- [Falcon](https://cerebralvalley.ai/e/raise-summit-hackathon/hackathon/gallery/32)
- [DSync](https://cerebralvalley.ai/e/raise-summit-hackathon/hackathon/gallery/33)

---

Markdown version of https://cerebralvalley.ai/e/raise-summit-hackathon/hackathon/gallery/30. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt