# LoanDesk

- **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:** [Daniel Jung](https://cerebralvalley.ai/u/dandan002)
- **GitHub:** https://github.com/dandan002/secfin-muse
- **Demo video:** https://youtu.be/r7Nd6mN5Fqo
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4999

LoanDesk AI is an agentic desk for securities lending. On a typical trading desk, a stock-loan trader works across broker chats, multiple execution venues, a books-and-records processor, and a front-office trading app. The trader connects them by hand, reading a chat, retyping it, checking both venues, sending the order and reconciling the booking. LoanDesk AI does the work itself.

It reads every chat message in its conversation and turns it into a structured request, quote, acceptance, recall or return. It clears easy "locates" (indications a stock is able to be borrowed) under a standing policy, reasons through the harder ones and sends the hardest to the trader with a suggested partial fill. It picks a venue for each order and explains the choice, then books the fill, checks it against the processor and settles it with the custodian. It also handles recalls, returns, re-rates and fails, and after the close it builds the next day's short list.

All the money maths is plain code, and every action that leaves the desk passes through one approval gate where the trader approves, edits, rejects or takes over. A kill switch freezes everything at once. This platform is meant to make traders' lives easier while also streamlining the traditionally opaque securities lending market.

## More from Meta Global AI Developer Hackathon

- [HamzaHacks07](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4996)
- [Walter Guzman](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4997)
- [Hi Agencia](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4998)
- [KyHoang](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5000)
- [NeuroMuse](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5001)
- [2 leches](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5002)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4999. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt