# ProposalIQ

- **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:** [Shourouk Dassouki](https://cerebralvalley.ai/u/shd)
- **GitHub:** https://github.com/shdassouki/ProposalIQ
- **Demo video:** https://youtu.be/I30iO8jjNbc
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3257

ProposalIQ: win the bid without overpromising.

I built ProposalIQ because I know what an RFP deadline feels like. During my time supporting Oracle presales at MDS for Computer Systems, I saw how much effort goes into answering one seemingly simple question: can we actually deliver what this customer is asking for?

The answer is usually scattered across product documents, old proposals, and conversations. Under deadline pressure, it is easy to miss a requirement or write a confident answer that the evidence does not support.

ProposalIQ turns that process into a reviewable workflow. Four agents extract requirements, retrieve capability evidence, draft responses, and audit compliance. Requirements retain their source references, proposal claims link to evidence, and coverage is calculated in code. Missing support stays visible for human review.

The fictional Harborview demo extracts 32 requirements and produces 71.9% coverage. That imperfect score is central to the product: it flags unsupported FedRAMP authorization, partial WCAG 2.2 support, and a gap in the requested ten-year migration history. A persuasive proposal should not hide those limitations.

Beyond drafting, ProposalIQ includes bid strategy advice, discovery-call transcription and insights, optional illustrative cover art, and PDF/CSV exports.

My goal is to give presales teams more time for customer conversations and solution design, while keeping commitments grounded in what they can deliver.

Try it: https://proposaliq-5j37.onrender.com/

The public demo runs in clearly labeled deterministic mock mode for reproducibility and free access. The Meta integrations were separately live-tested.

## More from Meta Global AI Developer Hackathon

- [Sanikammhatre's Team](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3254)
- [BlueTrace](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3255)
- [Siddhant Choudhary](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3256)
- [BenchSight](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3258)
- [drizzy765](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3259)
- [Spoty.dev](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3260)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3257. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt