StacksGPT
Customer Support

WhatsApp Business Chatbot Grounded in Your Documents

A WhatsApp Business chatbot that answers customer questions from your own documents stored in Google Drive.

The Problem

Customers ask product and troubleshooting questions on WhatsApp. Staff answer the same questions again and again, and generic bots give wrong answers.

How This Template Solves It

An electronics store loads its product guides from Google Drive into a vector database. When a customer asks a question on WhatsApp, the AI agent searches those documents and replies with an answer based on them.

How It Works

  1. 1Creates a Qdrant collection and loads documents from Google Drive, split and embedded with OpenAI.
  2. 2A webhook verifies your endpoint with Meta and receives incoming WhatsApp messages.
  3. 3Checks that each incoming event contains a user message.
  4. 4An AI agent with memory searches the Qdrant collection and replies through WhatsApp.

What You Need

  • A Meta developer app with the WhatsApp Cloud API
  • A Qdrant instance
  • A Google Drive folder with your documents
  • An OpenAI API key

What Stays Under Your Control

  • You choose which documents the bot can use.
  • Orders, refunds and sensitive requests should go to a person.

Watch Out For

  • WhatsApp limits when a business can send free-form replies. Check Meta's messaging rules for your use case.
  • Use the same URL for the verify and respond webhooks, one for GET and one for POST, as the notes explain.

Adapted from a community workflow template. Before publishing we removed credentials, account and file IDs, email addresses and sample data. The original author's setup notes are kept inside the workflow file.

More Customer Support Templates

Customer Support

Track Sentiment on Active Support Issues

Analyze the mood of comments on active Linear issues, track changes in Airtable and alert Slack when an issue turns negative.

Starts on: Airtable change, Schedule

AirtableOpenAISlack