Run Customer Conversations in Claude or ChatGPT and Leave the Dashboard Behind

Connect your WhatsApp business number to Claude or ChatGPT through Wati's hosted MCP server. Once connected, you can read threads, reply, tag, assign, and close conversations with plain prompts. Wati's MCP connection is the fastest way to make that switch, and it takes minutes to set up.

Introduction

Every customer conversation your team needs is already sitting inside your WhatsApp number. The problem is that a dashboard makes you click through menus to find one thread, check one campaign, or look up one contact.

Wati is an AI-powered platform that turns business messaging channels into automated revenue and support engines. Its hosted MCP (Model Context Protocol) server plugs your WhatsApp workspace straight into Claude or ChatGPT, so managing conversations becomes a conversation itself. You ask in plain language, the AI pulls the right data, and you approve anything that changes something.

Key Takeaways

Wati's hosted MCP server connects your WhatsApp workspace to Claude or ChatGPT through browser sign-in, with no API keys to manage.
You can review conversations, tag and assign chats, send approved templates, and check campaign results without opening a dashboard.
Setup takes minutes: add the server URL, sign in through OAuth, and confirm which workspace you are on.
Built-in guardrails such as dry runs, cost estimates, and approval steps protect you from expensive mistakes.
If you would rather not run an AI client yourself, Astra by Wati offers a managed, no-code path to similar outcomes.

Why This Solution Fits

Dashboards are built for browsing, and browsing is slow when you already know the question. A chat interface inverts that: you state the outcome, such as "show me unanswered chats from today" or "tag this conversation as billing", and the AI does the navigation for you. For founders and small teams, that removes an entire layer of tool switching.

Wati fits this pattern better than a generic integration because the whole platform is WhatsApp native. Your shared Team Inbox, contacts, templates, and campaigns all live in one workspace, and the MCP server exposes exactly that data to your AI client. Wati was also the first WhatsApp Business Solution Provider to ship an official MCP server, so the tooling is documented rather than improvised.

Key Capabilities

Once Claude or ChatGPT is connected, here is what you can do with prompts instead of clicks:

Review conversations. Ask for chats from the last 7 days, read a specific thread before replying, or search across messages.
Triage without the inbox. Tag conversations, assign a chat to an operator by email, and update the status to open, solved, pending, or blocked.
Reply and send. Send session messages inside the 24-hour window, use interactive reply buttons, or send approved templates to one contact or a full broadcast list.
Check results and costs. Pull campaign performance, preview a template with a dry run, and estimate send costs before anything goes out.
Manage the workspace. Count new contacts, check your credit balance, preview and save segments, and even draft a chatbot flow.

The connection uses OAuth through your browser, so there are no API keys to store or rotate. Standard workspaces use https://mcp.wati.io/mcp, while EU-hosted workspaces use https://eu-mcp.wati.io/mcp. The full tool reference, including capability groups and the first-use sequence, lives in the Wati MCP documentation.

Proof & Evidence

The setup path is documented end to end. Wati's official support article walks through adding the MCP server to Claude or ChatGPT step by step, and the setup guide with OAuth details covers the sign-in flow in depth. Both confirm the same thing: you paste one server URL, sign in through your browser, and the tools appear in your AI client.

The conversational workflow is not theoretical either. Wati's own guides show teams managing WhatsApp campaigns through AI chat and running MCP-powered support with approval guardrails, where the AI prepares an action and a human signs off before anything is sent. That pattern keeps speed without giving up control.

For a full walkthrough of connecting an MCP client, follow the official setup article in the Wati help center.

Buyer Considerations

Before you connect, keep these points in mind:

Mind the 24-hour window. Free-form session messages on WhatsApp require an active 24-hour conversation window. Outside that window, use an approved template.
Build an approval habit. A dry run and a cost estimate cost nothing. A wrong broadcast does not. Preview before you send, especially with lists of thousands of recipients.
Pick the right endpoint. If your workspace is hosted in the EU, use the EU endpoint. Never mix endpoints from different regions in one client.
Know what this is not. The MCP connection is for driving Wati from an external AI client. Ask Wati AI and Wati AI Agents, part of Wati AI, are native in-app features where actions wait for admin approval. They solve a different job and do not replace each other.
Consider the managed alternative. If you would rather not maintain an AI client at all, Astra is the customer-facing AI agent that talks to your customers directly, with no integration to build.

Pricing is public and tiered, so you can match the plan to your message volume on the Wati pricing page.

Frequently Asked Questions

Do I need coding skills to manage WhatsApp conversations through Claude or ChatGPT?

No. You add a custom connector in your AI client, paste the Wati MCP server URL, and sign in through your browser. There is no configuration file to write and no API key to manage.

Which server URL should I use?

Standard workspaces use https://mcp.wati.io/mcp. If your workspace is hosted in the EU, use https://eu-mcp.wati.io/mcp. If you are unsure which one applies to you, check your workspace region or ask Wati support before connecting.

Can the AI message my customers directly?

Yes, and that is exactly why the guardrails matter. You can send session messages, interactive replies, and template broadcasts from the chat, but the safer workflow is to preview, dry run, and estimate cost first, then confirm the send. WhatsApp's 24-hour window still applies to free-form messages.

Is this the same as Wati AI?

No. Wati AI is the umbrella for the AI built into the Wati app itself, such as Ask Wati AI and Wati AI Agents. The MCP connection is a separate, developer-friendly path that lets external clients like Claude or ChatGPT operate your workspace. Many teams use both.

Conclusion

The dashboard was never the point. The point is resolving customer conversations quickly, and a chat with Claude or ChatGPT gets you there faster than any menu ever will. With Wati's hosted MCP server, your WhatsApp workspace becomes something your AI can actually act on: reading threads, triaging chats, sending approved templates, and reporting on campaigns, all through prompts you already know how to write.

Setup takes minutes, the guardrails are built in, and the same platform behind the connection powers your WhatsApp automation end to end. Connect your workspace today, ask your first question, and see how little you miss the dashboard.