Catch altered claim documents before you pay.
DocIntel reads uploaded claim documents — repair invoices, estimates and proformas — checks the numbers, compares them with what you already know, screens for signs of tampering, and gives your analysts a clear report with the reasons behind every flag.
Or email info@digikavach.com
Classify. Extract. Compare. Detect. Report.
Every document goes through the same fixed, auditable workflow — so analysts get consistent results they can explain.
Classify
Sorts each file: tax invoice, estimate, proforma, claim form or policy schedule.
Extract
Reads the text and pulls out invoice number, supplier, date, totals, tax and policy reference.
Compare
Checks line items, GST and totals, and matches details against reference values from your team.
Detect
Screens for hidden or altered text, repeated image regions, edit markers and reused photos.
Report
Gives a 0–100 risk score, highlighted page previews and reviewer tasks with reasons.
Built for Indian claim documents.
Safe intake
Every file is virus-scanned and fingerprinted (SHA-256) before anything reads it.
Reads Indian invoices
Understands GST components — CGST, SGST, UTGST, IGST and cess — and ₹ / Rs. number formats.
Maths & consistency checks
Recalculates line items, discounts, taxes and totals, and checks the supporting documents a claim should include.
Tampering screen
Compares a PDF's hidden text with what's visible on the page, finds repeated image regions, and checks metadata and save markers.
Photo-reuse detection
Spots claim photos that have already appeared in earlier submissions.
Analyst-first review
Analysts can correct extracted values, with every change kept in an audit trail. Each customer's data is kept separate.
Fits into your claims workflow.
- REST API with OAuth2 client-credentials sign-in, plus an OpenAPI specification
- Asynchronous processing: upload, check status, fetch the report
- Claim-pack upload of up to 30 files in one ZIP through the API
- Space to record registry results (such as GSTN or Vahan) from your approved sources
- Or use the ready-made Document Verifier web app — no integration needed
Data & security
- Processed in India, in the Mumbai cloud region
- Files virus-scanned and fingerprinted on arrival
- Each customer's data isolated from every other customer
- Retention agreed with you for each deployment
Supports PDF, PNG and JPEG files up to 10 MB. Text reading currently supports English-language documents.
DocIntel supports your analysts — it doesn't replace them. Its reports highlight risk and explain why. They do not certify that a document is genuine, and they do not approve or reject a claim. A clean result means no indicators were found, not that a document is proven authentic; carefully retyped or flattened edits can still evade automated checks.
See DocIntel on your own documents.
Built first for motor insurance claims. Talk to us about a pilot for your claims or fraud team.