Skip to content
Case study · e-governance

Municipal corporations · Government of Maharashtra

2 minto verify an application, down from 15

10 lakh applications a year, verified in 2 minutes

DocVerify combines computer vision, OCR and agentic AI to auto-verify citizen application documents — shifting officers from repetitive checking to the cases that actually need judgement.

10 lakh applications a year, verified in 2 minutes — Stark Digital
The challenge

Officers processed 10+ lakh applications a year across 130+ schemes — 50,000+ Shahri Garib and 10,000+ education applications among them — spending ~15 minutes manually cross-checking each one. The result: slow decisions, high error risk, no audit trail and long citizen waits.

What we built

The approach.

01Read form data — auto-extract names, dates and scheme details from the application
02Extract documents — advanced OCR reads printed and handwritten government documents
03Compare & verify — AI agents cross-reference form data against the documents
04Show match score — officers see an accuracy percentage; only flagged items need review

How a document becomes a decision

A four-step verification pipeline.

  1. 01IntakeIngest scans at scale
  2. 02OCRRead every page
  3. 03135-field extractionStructure each record
  4. 04Translate / summarizeNormalise content
  5. 05ValidateHuman review on low confidence
  6. 06HRMSCommit to system of record

Read form data → OCR the uploaded documents (Aadhaar, certificates, income proofs) → cross-reference with context-aware matching (e.g. ‘Pune’ vs ‘पुणे’) → surface an accuracy score, flagging only discrepancies for human review.

Results

The outcome, in numbers.

87%

projected reduction in time per application

7×

increase in daily application throughput

95%+

verification accuracy across all schemes

Built with

Technology & integrations

Computer VisionOCRAgentic AIAadhaarBilingual (Marathi)

Let's scope your next platform.

No obligation. A senior engineer reviews your requirement and maps it to a concrete, costed delivery plan.