How to connect Claude, GPT, and other AI models to WhatsApp using the Model Context Protocol (MCP) with WhatsTeam.
Your support team answers the same 40 questions every week. Your sales reps copy-paste product specs into chat windows. Your operations manager manually routes conversations to the right department. All of this happens on WhatsApp -- and all of it can be handled by AI.
The missing piece has always been the connection layer. How do you get Claude, GPT, or any large language model to actually read and respond to WhatsApp messages in real time? The answer, as of 2026, is the Model Context Protocol (MCP) -- an open standard that lets AI models interact with external tools through a single, consistent interface.
This guide walks you through connecting AI models to WhatsApp using WhatsTeam's MCP server, built on the official WhatsApp Business API (WABA). No unofficial workarounds. No fragile browser automation. Just a production-grade API connection that Meta actually supports.
The Model Context Protocol is an open standard created by Anthropic that defines how AI models communicate with external tools and data sources. Think of it as a universal adapter: instead of building custom integrations for every AI platform, you expose your tools through MCP and any compatible model can use them.
Before MCP, connecting an AI model to WhatsApp required:
With MCP, that entire stack collapses into a single configuration block. WhatsTeam's MCP server handles the translation layer, authentication, and message formatting. Your AI model just talks to the MCP server, and the server talks to WhatsApp through the official Business API.
The flow is straightforward:
The MCP server exposes tools that your AI model can call directly: sending messages, reading conversation history, managing contacts, creating templates, and more. The AI model decides when and how to use these tools based on your instructions.
Before you configure anything, make sure you have:
That's it. No local Node.js installation. No Docker containers. WhatsTeam's MCP server runs over HTTP streaming, so the entire setup is a configuration change.
Open Claude Desktop's configuration file:
~/Library/Application Support/Claude/claude_desktop_config.json%APPDATA%\Claude\claude_desktop_config.jsonAdd this configuration:
hljs json{
"mcpServers": {
"whatsteam": {
"type": "http",
"url": "https://api.whats.team/mcp?key=YOUR_API_KEY_HERE"
}
}
}
Replace YOUR_API_KEY_HERE with the key you copied in Step 1.
Restart Claude Desktop. Open a new conversation and type:
"Send a WhatsApp message to +1234567890 saying: Hello, this is a test from my AI integration."
Claude will detect the WhatsTeam MCP tools and send the message through your WhatsApp Business API connection. You should see the message delivered within seconds.
https://api.whats.team/mcpwhatsteamAdd the server to your project's .claude/settings.json:
hljs json{
"mcpServers": {
"whatsteam": {
"type": "http",
"url": "https://api.whats.team/mcp?key=YOUR_API_KEY_HERE"
}
}
}
Add to your VS Code settings (JSON):
hljs json{
"mcp.servers": {
"whatsteam": {
"url": "https://api.whats.team/mcp?key=YOUR_API_KEY_HERE",
"transport": "http-streaming"
}
}
}
In Cursor's MCP settings:
hljs json{
"mcpServers": {
"whatsteam": {
"url": "https://api.whats.team/mcp?key=YOUR_API_KEY_HERE"
}
}
}
All of these use HTTP streaming -- no local server installation, no Node.js dependency, automatic updates when WhatsTeam adds new capabilities.
Once connected, your AI model gets access to a full set of WhatsApp Business API tools:
| Tool Category | Capabilities |
|---|---|
| Messaging | Send text, images, documents, audio, video, location, and contact cards |
| Templates | Create, manage, and send pre-approved message templates |
| Conversations | Read message history, search conversations, get unread counts |
| Contacts | Create, update, search, and segment contacts |
| Campaigns | Trigger broadcast campaigns to contact segments |
| Product Catalog | Share product listings and collections directly in chat |
| Labels & Tags | Organize conversations with labels for routing and reporting |
The AI model discovers these tools automatically through the MCP protocol. You don't need to teach it what's available -- it reads the tool definitions and knows what it can do.
Connect your AI model to WhatsTeam and let it handle first-line support. The AI reads the incoming message, checks conversation history for context, and responds with accurate answers based on your knowledge base.
How to set it up: Create an AI agent in WhatsTeam with instructions that describe your products, policies, and tone of voice. The agent uses the MCP connection to read and respond to messages automatically.
Expected results: Thousands of businesses using WhatsTeam report that AI agents handle the majority of first-response messages without human intervention, freeing support teams for complex cases.
Use AI to read incoming messages, classify intent (support, sales, billing, complaints), and route the conversation to the right team member or department using WhatsTeam's Team Inbox.
How it works: The AI model reads each new message, determines the category, and assigns a label or routes to the appropriate agent -- all through MCP tool calls.
Sales reps can ask the AI to draft personalized follow-up messages, product recommendations, or proposal summaries -- then send them directly through WhatsApp without leaving their AI tool.
Example prompt: "Draft a follow-up message for the customer who asked about bulk pricing on the premium plan yesterday, and send it via WhatsApp."
Use AI to analyze your contact segments and generate personalized campaign messages. The AI reads contact data through MCP, crafts tailored messages, and triggers sends through WhatsTeam's campaign tools.
Connect your product catalog to the AI model. When a customer asks "What do you have under $50?", the AI queries your catalog through MCP and sends formatted product cards directly in the WhatsApp conversation.
"MCP server not found"
https://api.whats.team/mcp?key=YOUR_KEY"Authentication error"
"Message failed to send"
AI model doesn't see WhatsTeam tools
mcpServers block is correctly formatted JSONMCP (Model Context Protocol) is an open standard that lets AI models like Claude and GPT interact with external tools. WhatsTeam's MCP server acts as a bridge between these AI models and the WhatsApp Business API, allowing the AI to send messages, read conversations, manage contacts, and more -- all through a single configuration.
No. The setup involves copying a configuration block into your AI tool's settings file. There is no code to write, no server to deploy, and no dependencies to install. The entire process takes under five minutes.
Any tool that supports the MCP standard works with WhatsTeam. This includes Claude Desktop, Claude Code, ChatGPT (Pro), VS Code Copilot, Cursor, and Windsurf. As more tools adopt MCP, they will work with WhatsTeam automatically.
Yes. WhatsTeam connects exclusively through the official WhatsApp Business API (WABA) provided by Meta. There are no unofficial connectors, no browser automation, and no risk of account bans. Your messages are delivered through Meta's production infrastructure.
The MCP server is included in all WhatsTeam plans. Pricing starts at $29/month for the Starter plan, $69/month for Business, and $119/month for Enterprise. WhatsApp conversation charges from Meta apply separately based on your message volume. Sign up to get started.
Ready to connect AI to your WhatsApp Business account? Create your WhatsTeam account and configure MCP in under five minutes. Your AI model is already waiting for the tools.
Step-by-step guide to integrating HubSpot CRM with WhatsApp Business API via WhatsTeam. Automate lead follow-up, deal notifications, and customer engagement.
Complete guide to integrating Salesforce with WhatsApp Business API. Connect via Zapier, Make, or REST API with WhatsTeam.
Step-by-step guide to connecting your WhatsApp Business App with WhatsTeam using Coexistence — requirements, setup process, pricing, and troubleshooting.