ABHILEKH

अभिलेखRFP 3.8TRL 7

Document Intelligence & e-Office Automation

Devanagari OCR (VLM), intelligent data extraction to structured JSON, contract intelligence, file movement analytics with bottleneck detection.

Hero Demo Flow

Bad Khasra scan → OCR → 14 structured fields → swap OCR provider live in API Master

VLMLLM

Configurable OCR Spine Swap provider live

The Devanagari OCR engine is routed through the platform's API Master. Open the /master page and toggle the VLM/OCR provider (Z.ai Vision → Azure Document Intelligence → Bhashini OCR) — ABHILEKH picks up the new provider on the next request without code changes. Model Card: ABHILEKH/devanagari-ocr (TRL 7).

Document Image Upload

दस्तावेज़ छवि अपलोड करें — JPG/PNG · max 5 MB

Drag & drop document image here

or browse files

हिंदी: यहाँ दस्तावेज़ की फोटो खींचें या क्लिक करके चुनें

Extraction Parameters

Document Type

VLM call to ABHILEKH.OCR prompt template

Recent Extractions

Letter

26 Aug 2026

Review

OCR results will appear here

Upload a document image, select its type, and click Extract.

Field-level confidence Low-conf → human review GovDocument persisted

Prototype — MP-contextualised demonstrator · Synthetic data only · AI routed through API MasterAnchor depts: Revenue · GAD · Finance