# Bush

- **Event:** [Built with Opus 4.7: a Claude Code hackathon](https://cerebralvalley.ai/e/built-with-4-7-hackathon)
- **When:** Apr 21 at 12:00 PM – Apr 27 at 2:00 AM (EDT)
- **Where:** Online
- **Team:** [Jorge Fuertes](https://cerebralvalley.ai/u/horhebush)
- **GitHub:** https://github.com/horhebush/bantay
- **Demo video:** https://youtu.be/EGwUt_UPAaM
- **Gallery:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/117

Every Filipino has pointed at a potholed road or a flood drain that was never finished — and known, without proof, that something was wrong. Proving it has always required a lawyer, a journalist, or months of reading government documents nobody has time to read.

Bantay reads the entire Philippine government's infrastructure procurement dataset — 248,220 contracts, ₱6.4 trillion committed — and flags four kinds of anomaly: ghost projects marked 100% complete, contractors winning over 40% of a region's work in a single category, contractors whose licenses have been revoked, and timelines that do not survive arithmetic.

For 216 contracts in our Rizal pilot, Claude Opus 4.7 takes over as the analyst. It writes a verdict on each — but first it argues against itself, drafting the strongest non-anomalous explanation a contractor's lawyer could mount, then checking whether the data actually supports it. About one in eight contracts gets downgraded by this self-correction. The rule engine cites four laws; Opus cites six to eight per contract — specific sections of the Philippine procurement law, state audit circulars, and board resolutions — in plain language a citizen can act on.

Filipinos already have voice. Bantay gives them the receipts.

---

Markdown version of https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/117. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt