Loading page...Загружаем страницу...正在加载页面...
AtlasRepo

Your Policies Are Out of Date: How I Built a Sanity AI Agent to Catch Fact Drift

Описание и применение

Система Fact Ledger автоматически обнаруживает и исправляет несоответствия в политике, используя модели фактов в Sanity и AI‑генерируемые патчи, обеспечивая единый источник истины и аудит изменений.

Задача

When a policy value changes in a CMS, hard‑coded copies across many pages remain stale, causing fact drift and potential legal risk.

Результат

Fact Ledger achieves 100% precision and recall on a seeded dataset of 31 violations, automatically propagates policy changes, and provides a single‑click remediation workflow with audit logging.

Рабочий процесс

  • Edit a fact document in Sanity Studio
  • Webhook triggers deterministic scanner (R1–R5)
  • Scanner identifies stale or contradictory text and creates finding documents
  • AI agent drafts Sanity patch mutations for each finding
  • Editor reviews diffs in Studio or dashboard and approves fixes
  • Client.transaction() atomically patches pages, updates findings, publishes remediation, and logs audit event

Доказательства

Fact Ledger fixes this at the data model level.

Источник

Run the benchmark against the seeded dataset of 31 planted violations: 100% precision, 100% recall.

Источник

The core principle, distilled to four words per step: Rules flag. AI drafts. Human approves. Sanity remembers.

Источник

  • Дата записи: 2026-10-03T22:40:53.737Z