# Archivist

- **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:** [Kamil Shirinov](https://cerebralvalley.ai/u/Shirinov)
- **GitHub:** https://github.com/kamil-shirinov/Archivist
- **Demo video:** https://youtu.be/iRVeZBQxxP4
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1686

Archivist turns scanned financial archives into a verified database. Every value is cited to its spot on the page, read twice (the second time blind, from a crop of the cell alone), checked by plain-code arithmetic, and sent to a person when anything is in doubt. No model grades its own work, and Archivist never changes a value: it flags source errors and proposes corrections for a person to decide. A new kind of table needs no code: Muse Spark drafts a template from a few pages, plain-code arithmetic tests it, and I approve it.

The problem is one I met at a bank internship. I built a parser for about 2,000 legacy memos, then found there was no archive to run it on: the oldest records existed only on paper, and loan applicants brought their earnings in notebooks that staff typed in by hand. A model reading pages like these alone gets some values wrong without saying which.

I pre-registered each test with dated commits before running it. On 157 synthetic pages (4,267 values), 0.36% of the values Archivist accepted were wrong, against 4.45% for a single read, and it caught 93.65% of wrong values. It missed two targets: it sent 21.58% of values to a person (target 20%) and flagged 23 of 26 planted source errors (target 90%). On 58 held-out pages of the US Treasury Bulletin, with 400 cells I labelled by hand, the main target was not met: 16.3% of accepted values were wrong against 27.8% for a single read, after a dash correction made by a rule written down first (36.9% against 41.1% as registered). On pages whose layout matches their template, the corrected figure was 0.0%, meeting the target there; the remaining errors are on pages whose layout changed after the templates were drawn. Every result, misses included, is published. Live demo: https://kamil-shirinov.github.io/Archivist/

## More from Meta Global AI Developer Hackathon

- [Rotebot](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1683)
- [TOMAN](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1684)
- [Nivetha](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1685)
- [JenX](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1687)
- [Aaditya GC](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1688)
- [Tempo](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1689)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1686. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt