DocuLens AI

Enterprise document intelligence platform. Extracts, validates, and routes structured data from documents using AI with human-in-the-loop review.

Built by TDACorp

---

What It Does

Upload any document (invoice, ID, medical record, contract) and DocuLens AI:

1. Classifies — auto-detects document type
2. Extracts — pulls structured data using AI (field-level confidence scores)
3. Validates — runs business rules, flags low-confidence fields
4. Routes — auto-approves high-confidence docs, sends the rest to human reviewers
5. Logs — every action recorded in a compliance-grade audit trail

Key Features

• 6-stage serverless pipeline — Inngest step functions with real-time status via Pusher
• HITL review — side-by-side PDF viewer with editable extracted fields
• Field-level confidence — auto-approve/review/flag thresholds per tenant
• Hybrid search — semantic + full-text across all extracted content
• Multi-tenant — isolated data, RBAC, API keys
• Audit trail — every touch logged with before/after values
• Template system — schema-driven extraction for any document type

Tech Stack

Next.js 16, TypeScript, PostgreSQL + pgvector, Inngest, Gemini/OpenAI, Auth.js, MinIO/R2, Pusher, Tailwind + shadcn/ui, Recharts

Quick Start

# Prerequisites: Docker, Node 20+, pnpm

# 1. Start infrastructure
docker compose up -d

# 2. Install dependencies
pnpm install

# 3. Set up environment
cp .env.example .env
# Add your GOOGLE_AI_API_KEY (free) or OPENAI_API_KEY to .env

# 4. Set up database
pnpm db:push
pnpm db:seed

# 5. Start the app
pnpm dev          # Terminal 1: Next.js
pnpm inngest:dev  # Terminal 2: Inngest dev server

# Open http://localhost:3000

Demo Credentials

Role	Email	Password
Admin	admin@tdacorp.demo	demo1234
Reviewer	reviewer@tdacorp.demo	demo1234
Viewer	viewer@tdacorp.demo	demo1234

Documentation

• PRD — Full product requirements
• Architecture — System design and data flow
• Decisions — Architecture decisions with rationale

License

Proprietary. TDACorp Private Limited.