Whatsapp OCR Bot for Business Cards

Project Flow Design: Agent Tap2Read by
Sellwell
1. Overview
Agent Tap2Read by Sellwell allows users to send a business card photo via WhatsApp. The bot
extracts contact details (Name, Phone, Email, Company, etc.), structures them, and replies back.
Users can also export data in formats like CSV/Excel or push to CRM.
2. Components
WhatsApp Business API (Meta Cloud API) Entry point for messages
Webhook Backend (Django/Node/Flask) Processes incoming messages
Storage (DB) Save parsed contacts (optional)
OCR Service Extracts text from card image
AI Parsing Layer (Gemini Pro) Converts OCR output into structured JSON
Response Formatter Formats details & sends back to WhatsApp
Export Integration Export contacts to Excel/CSV or CRM
3. High-Level Flow
1. User sends 'Hi' or a business card image via WhatsApp.
2. Meta API forwards the message to backend webhook.
3. Backend checks message type (text/image). If image, download media.
4. OCR extracts raw text from card.
5. Gemini Pro parses text into structured JSON.
6. Database saves structured contact (optional).
7. Bot replies with formatted contact details.
8. User can request export (CSV/Excel).
4. Technical Stack
Messaging WhatsApp Business Cloud API (Meta)
Backend Django REST / Node.js Express / Flask
OCR Free: Tesseract | Paid: Google Vision, AWS Textract
AI Parsing Gemini Pro
Database (optional) PostgreSQL / Firebase / MySQL
File Export Pandas (Python) / ExcelJS (Node)
5. Flow Diagram (Textual)
[User WhatsApp] | v [Meta WhatsApp API] ---> [Webhook Backend] | | | v | [OCR Extraction] | | | v |
[Gemini Pro Parsing] | | | v | [Database Save] (optional) | | v v [Bot Reply to User] <---- [Formatter +
Export]

Language : React +Node
Database : Postgradge or Mysql