WhatsApp Automation: Scaling Demand Capture for a Supply Chain Client

Ashish joined a project for a supply chain client whose operations team was manually monitoring around 800 WhatsApp groups to capture demand signals. Associates were expected to read every message, identify relevant demand requests, and manually enter them into the company's system. At that volume, messages were inevitably missed, and the team's time was consumed by a task that left no room for higher-value work like building supplier relationships.

Finding the Integration Layer

The first challenge was ingestion. Standard approaches using web scrapers or browser automation tools like Playwright were unreliable. Ashish found Bailey's Library, an open-source solution that behaves like a WhatsApp Web session: once connected, it reads all incoming messages on a given number in real time. This solved the ingestion problem without requiring brittle scraping infrastructure.

Building the Classification Pipeline

With messages flowing in, he built an LLM-based classification layer that evaluated each message and determined whether it contained a relevant demand signal. Relevant messages were automatically routed to the client's system; irrelevant ones were discarded. The pipeline ran continuously, processing messages as they arrived.

Scaling to 800 Groups

The system initially ran across a small set of groups. As the client expanded coverage from roughly 10 to 50 groups and then toward 800, daily message volume jumped from 200 to 300 messages to around 6,000. The original architecture could not handle that load reliably.

Ashish redesigned the infrastructure using AWS Lambda functions that scaled dynamically with message volume. The system would fire Lambda instances based on incoming load, process each message independently, and take the appropriate action. The architecture was designed to handle 10,000 to 50,000 messages per day without degradation.

Once fully deployed, all demand signals were captured automatically. The operations team was freed from message monitoring entirely and redirected toward relationship-building and demand generation work.