WhatsApp Automation Platform
by: Tomás Ospina.
Overview
Two complementary systems built around WhatsApp as an operational channel: an ingestion service that parses incoming messages into a database with a dashboard on top, and a bulk sender that delivers messages and media to a contact list with retries.
Every message makes a full round trip — submitted, parsed, stored, confirmed — so the person on the other end knows immediately whether their entry was accepted.
Ingestion & operations — WhatsAppServer
A Flask service receives messages through a webhook (both the WhatsApp Business API and an Android relay endpoint), runs them through a parser that validates the expected format, and stores the structured record in SQLite. Malformed messages get an immediate correction reply instead of silently disappearing.
On top of that sits a dashboard with session-based login, per-period metrics, CSV export and PDF report generation — so the people entering data over WhatsApp and the people reading the results never touch the database directly.
Bulk sending — whatsapp-sender
A Node.js service built on whatsapp-web.js with LocalAuth session persistence: it
reads the contact list from SQLite, sends text and media (PDF, video) to each contact,
and wraps every send in a retry helper so a dropped connection does not lose a recipient.
Technical Specifications
| Component | Stack |
| —————- | ——————————————————— |
| Webhook / API | Python, Flask — WhatsApp Business API + Android relay |
| Parsing | Format validation with an error reply path |
| Storage | SQLite (mensajes.db, contactos.db) |
| Dashboard | Flask, authenticated sessions, CSV export, PDF reports |
| Sender | Node.js, whatsapp-web.js, LocalAuth, headless Puppeteer |
| Deployment | Linux server, environment-based configuration |
My Role & Key Contributions
- Designed the message format and the parser that turns free-text WhatsApp messages into database records.
- Built the Flask webhook, the SQLite schema and the authenticated operations dashboard with metrics and exports.
- Wrote the Node.js sender with media support and a safe-send retry wrapper.
What I Learned
Validating at the edge — replying “formato incorrecto” the moment a message arrives — removed almost all of the data cleanup that would otherwise have happened downstream.
🔒 Private repository — happy to walk through the code on request.