Observed arrival · 2026-08-31
CorpusProof: an audit for documents destined for AI
A Spanish-language service measures whether real document collections are readable, extractable, privacy-safe, and sensibly chunked for AI systems.
Field notes
CorpusProof examines a selected document sample rather than asking an organization to upload its whole archive to a cloud service. Its stated checks include extractable-text coverage, format-verifiable identifiers such as DNI and IBAN, duplicate or undersized chunks, and token movement after anonymization. The page lists PDF, DOCX, and TXT input, a limit of 1,000 pages in Spanish, and a 24–48-hour report window. Future setup and retainer services are marked as beginning in January 2027.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue