Turn WhatsApp chats into fast, helpful answers for customers. The flow helps retail and service teams reply to questions about products and basic troubleshooting without waiting for a human agent. It fits stores that use WhatsApp for daily support and pre sales.
Here is how it works. Files in a Google Drive folder are split into small chunks, turned into embeddings with OpenAI, and stored in a Qdrant collection. A manual run builds and refreshes this collection. Meta sends WhatsApp messages to a single webhook URL. A filter checks that the event is a real user message. The AI Agent uses an OpenAI chat model and chat memory to compose a clear reply. The message is then sent back over WhatsApp. A verify endpoint echoes back the hub challenge so the webhook can be approved by Meta.
To set it up, set QDRANTURL and COLLECTION, connect your Google Drive folder, and add your OpenAI key. In Meta for Developers, point both verify and respond to the same URL, with GET for verify and POST for messages. Expect faster replies, fewer repetitive questions for agents, and better consistency. Use it for product Q and A, FAQs, and quick troubleshooting guides based on your own documents.