DataQualifyUNSTRUCTURED
Document Intelligence

Review any document — rules or AI, your call

Point DataQualify at contracts, invoices, PDFs, scans and emails. Extract and validate them with a deterministic rule engine, an LLM, or both — and nothing leaves your machine unless you allow it.

Rule-based & AI Runs fully on-prem Nothing shared without consent
Sample document
MSA_Acme_2026.pdf2 · PDF
INV-2026-0417§12.3
•••• redacted€ 48,250.00
Review method

Deterministic patterns, dictionaries and layout rules — runs 100% offline.

Send to LLM
Not needed — rule review never leaves your machine.

Two ways to read a document

Deterministic where you need auditability, AI where you need understanding. Use either — or chain them.

Rule-based

Regex, dictionaries, checksums and layout rules extract and validate fields the same way every time. Fully offline, fully explainable — the answer never changes between runs.

  • Runs air-gapped, zero external calls
  • Deterministic & audit-ready
  • Checksums for IBAN, VAT, tax IDs

AI-powered

An LLM understands messy scans, free-text clauses and documents it has never seen — pulling out meaning, not just patterns. Point it at your own model, cloud or self-hosted.

  • Handles OCR noise & unseen layouts
  • Understands clauses & intent
  • Bring your own LLM — 8 providers
Your data, your rules

Nothing is shared without your permission

By default every document is processed on your own infrastructure. AI review only reaches an LLM after you explicitly turn on consent — and even then, PII is redacted first and every send is written to an immutable log. Deploy fully air-gapped and no byte ever touches the internet.

On-prem by default

Rule review and OCR run entirely on your machines. No cloud round-trip, ever.

Consent-gated LLM

A document reaches an LLM only when you flip consent on — per run, revocable, off by default.

Redacted before send

Detected PII is masked before anything is sent, so models never see raw identifiers.

Air-gap ready

Run the whole pipeline with self-hosted models and zero outbound network access.

Every send logged

An immutable audit trail records what was sent, to which model, and who approved it.

Bring your own LLM

Point AI review at your own hosted model — the data stays inside your boundary.

Built for unstructured documents

Contracts, invoices, forms, correspondence — structured or scanned.

PDFWord / DOCXScans & OCREmailsContractsInvoicesFormsReceiptsStatementsReports

Review your documents without giving them away

Run rules on-prem, add AI when you choose to, and keep every byte inside your boundary until you say otherwise.