DocuExprt Platform — Complete Walkthrough

DocuExprt is an agentic AI-powered, web-based document verification platform built to bring documents to life. It is fully customisable and deploys across any document-heavy sector.

• Sectors: Automotive, Banking, Tendering, Legal, Education, HR, Insurance, Government and more.
• Principle: Wherever there is a document, DocuExprt can extract, validate, verify and reason over it.

Every login is protected by 2FA. After username/password, the user enters a time-based authentication code before any module becomes accessible.

The post-login dashboard surfaces operational health at a glance. The left navigation includes Dashboard, Workspace, Integration, Credentials, Talk to Document, DocuExprt AI, DigiLocker Verification, Notifier Settings and Administration.

Top-level widgets:

• Counts of Workspaces, Workflows, Templates and Users
• Execution trends and statistical charts
• Billing overview (tokens / pages processed)
• Top Workflows and Top Templates
• Document forensic summary
• Trigger Notifications — flags missing documents in a document set and any document about to expire within a configurable threshold

A workspace is a logical container of workflows for a single sector or client engagement (e.g. Education, BFSI, Insurance). Inside each workspace you get Overview, Workflows and Templates tabs.

• Spin up unlimited workspaces — one per client, business unit or document programme
• Templates standardise the extraction format per document type
• Workflows orchestrate the steps a document goes through inside that workspace

Workflows are built visually on a canvas. The base node ingests the document, generates a template (the field schema), and chains into validation and conditioning nodes.

Supported via DigiLocker (out-of-the-box):

Additional document types — Passport, Driving License, PAN and others — are validated through dedicated government API integrations.

Example workflow (education): SSC + PAN + HSC are uploaded, each is validated against its DigiLocker source, then a conditioning node enforces a name-match across all three documents before producing the final pass/fail output.

When a workflow runs, results are stored masked by default. To read the extracted data, the operator enters their decryption key — the canvas then shows live data, blue conditioning nodes, and the orange document circles (uploaded vs. DigiLocker-fetched).

• Document detection: hologram, stamp, sticker, logo, photograph
• Document forensic verdict displayed on the same canvas
• Output can be translated into any Indian or international language
• Bulk results exportable to Excel

DocuExprt integrates with any client CRM or ERP and consumes documents directly from the client's S3-compatible storage. Processing happens through APIs and results are pushed back into the client's own cloud.

Talk to Document is a conversational interface over verified document data. The operator unlocks with the decryption key, picks a workspace, optionally narrows to a specific workflow, and then asks plain-English questions.

• Specific, on-the-fingertip answers — no SQL, no exports
• Switch workspace/workflow mid-session
• Start a new conversation thread or browse the full conversation history
• Export the entire chat for audit and record-keeping
• Decryption key required to copy messages — preserves data secrecy

The Copilot turns a single natural-language prompt into a complete, runnable workflow. It selects the document, infers the right extraction fields, suggests the validation conditions, wires the nodes, and creates the workflow on the canvas.

Demonstrated in the video: a prompt creates an SSC workflow that extracts candidate name, year of passing, percentage and subject-wise marks, and enforces a "percentage above 50" condition. The workflow runs immediately and the output displays subject marks, totals, the percentage condition result, and the final verdict.

The DigiLocker module orchestrates candidate-consented document retrieval directly from the government source. Setup flow:

• Pick the workspace + workflow (e.g. Education Document Validation)
• Configure the template with email, mobile and a unique identifier for each candidate
• System sends two emails: (1) pre-processing instructions for DigiLocker setup, (2) consent request to grant DocuExprt access to specified documents
• Candidate authenticates via DigiLocker OTP on registered mobile
• DocuExprt fetches only the consented documents and routes them into the workflow

The same flow scales to thousands of candidates — the template + workflow are reused; only the candidate identifiers change.

Each uploaded document is cross-matched against its DigiLocker counterpart, and a forensic engine scores tamper risk independently.

Source validation:

• Uploaded vs. DigiLocker copy compared field-by-field
• Conditioning nodes flag name mismatches, spelling drift, modified fields
• Visual canvas: orange = documents, blue = conditions, green = pass

Document detection: hologram, stamp, sticker, logo and photograph presence checks.

Forensic verdict per document:

• Risk Level: Low / Medium / High
• Fraud Score: 0–100
• Verdict: Modified / Not Modified

All outbound emails from DocuExprt are template-driven and fully editable:

• DigiLocker Pre-Processing — setup guidance for the candidate
• DigiLocker Document Remaining — reminder for incomplete document sets
• DigiLocker Verification — consent / access request
• CC / BCC support per template

The email queue dashboard shows delivery status per candidate per workflow — success, opens, action-required — so the operator can spot any stuck communication and resend.

Role-Based Access Control: Admin, Workspace Manager, Workflow Manager and custom roles. Each role's data visibility is gated by the decryption key — so even an admin cannot read masked data without authorisation.

User logs capture every login and action for audit.

Billing dashboard reports per workflow:

• Total input & output tokens consumed
• Total validation cost
• Per-page cost and total pages processed
• Target characters processed and total records
• Views: List, Grid, Chart — segmented day-wise, week-wise or month-wise

The walkthrough closes by tying the four pillars together:

• Document-to-document validation — cross-check fields across uploaded documents
• Source validation — DigiLocker and other government portals
• Document forensic — tamper, modification and identification-mark detection
• Talk to Document — natural-language reasoning over verified data

Together these let organisations automate the administrative work of verification and redirect their team's attention to the core business.