Fraud · live

Capture DB + Scam Patterns

A continuously synced research database of UAE-relevant fraud signals — articles, Reddit reports, observed typologies — feeding the askhoras corpus.

Why

Askhoras only gets sharper if the corpus underneath it stays current. The Capture DB is the research substrate: every credible signal — articles, Reddit threads, advisory notices — gets triaged, tagged, and either filed as raw intel or promoted into a Scam Pattern v2 entry the LLM can reason with.

How

  • Inoreader feeds + manual capture flow into Notion (Capture-Info DB)
  • Daily syncs and pattern extraction; 246+ articles synced in the last month alone
  • Patterns versioned (v2 schema), tagged by archetype and region
  • Sync state tracked with a Date Synced field so re-runs are incremental