# BenchmarkWatcher

- **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:** [Albert Nikanorov](https://cerebralvalley.ai/u/albertn)
- **GitHub:** https://github.com/alikatgh/benchmarkwatcher/tree/ui/ux-polish-20261027
- **Demo video:** https://youtu.be/igxH8F9W2HA
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4337

BenchmarkWatcher Claim-proof board turns a financial assertion into a reproducible investigation. An analyst, journalist or curious reader can ask whether a company's free cash flow or operating income kept pace with revenue, inspect the exact filing rows behind the result, and compare the same claim across a selected group of companies.

Muse Spark selects a bounded research plan and relevant evidence. Application code then calculates every growth rate, margin, gap and verdict from retained SEC filing data. The evidence stays visible: reported values, fiscal periods, filing accessions, source links and snapshot hashes. Users can recheck the arithmetic and change cash capital expenditure in an explicitly hypothetical scenario.

A concrete example is Microsoft FY2024 to FY2025: revenue rose 14.93%, while calculated free cash flow fell 3.32%. The board shows how operating cash flow and cash PP&E purchases produce that result, and makes the distinction between reported figures and calculated metrics explicit.

The hackathon contribution is a new claim-proof engine, cohort investigation, model-selected evidence, audit/recheck flow and interactive board added to the existing BenchmarkWatcher project. The existing product is disclosed as the baseline in HACKATHON.md.

The current scope covers 14 companies, each company's own FY2024/FY2025 annual periods, and two supported growth comparisons. Fiscal-period differences and metric definitions remain visible. Unsupported questions, such as isolating AI-only spending from consolidated filings, are declined. Scenarios are labelled arithmetic exercises. The board supports financial research and education; it provides no forecasts or investment advice.

## More from Meta Global AI Developer Hackathon

- [KV_ca$h](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4334)
- [Pixel Societies](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4335)
- [Ali Lin](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4336)
- [Foresight-AI](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4338)
- [Real One](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4339)
- [SoloPaddlerLab](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4340)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4337. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt