read, extract, and route documents.
Agentic document processing — extraction, classification, redaction, QA, multi-step reasoning over docs. KYC documents, bank statements, ITR / Form 16, GST returns, NACH e-mandates, trade-finance docs (LC/BG/invoice), cheques (MICR/CTS-2010), SWIFT messages, CBS exception logs.
- Auto-classify + routeClassify incoming docs by type and route to the right pipeline. 8 document types at GA.
- KYC extractionAadhaar · PAN · passport · utility bill · voter ID. Cross-validation across documents.
- Bank statement parsingMulti-format. Cashflow categorisation. Existing-loan detection. Salary identification.
- Tax form extractionITR · Form 16. Indian tax form parser — income, deductions, sources.
- Trade finance docsLC · BG · invoice · PO matching. Sanction-list screening on entities.
- DPDP redaction pre-LLMPII stripped before any model call. India-region residency. On-prem deployable.
- Maker-checker handoffExtracted fields routed to human approval. Audit trail of every decision.
Document arrives via API, web upload, or email. Classifier auto-routes. Field extractor pulls structured data. Cross-validator checks across documents. PII redactor strips before any LLM call. Maker-checker workflow handles approval. Output written to immutable audit store.
↓
classifier → field extractor → cross-validator
↓
pii redactor (pre-llm) → maker-checker
↓
structured output · audit trail · downstream system
ga july 2026 · powers nehishdocs
DocAgent is the cross-vertical engine. NehishDocs is the BFSI overlay (banking document types pre-trained). General availability planned for July 2026.
start with a diagnose.
Two to three weeks. Five to ten lakh. A written scorecard with topology recommendation, cost ranges, and remediation plan. No commitment to build.
request diagnose