Manage Customer Conversations in Claude or ChatGPT Without Opening a Dashboard

The best way to manage customer conversations through Claude or ChatGPT instead of a dashboard is to connect your WhatsApp Business number to Wati's MCP server, then run lookups, replies, tagging, assignments, and campaign checks as plain chat requests. Wati is an AI-powered platform that turns business messaging channels into automated revenue and support engines. With the Wati MCP server, your AI client gets direct, permissioned access to your Wati workspace, so the dashboard becomes optional rather than essential.

Introduction

Most teams live inside a dashboard to answer three questions: what customers said, what we did about it, and what happened next. That works, but it means every small task is a navigation exercise. You open the inbox, filter a contact, read a thread, copy a template, and check a campaign report, all before you act.

There is a faster path. Wati, an official Meta Business Partner for the WhatsApp Business API, exposes your workspace to AI clients through the Model Context Protocol (MCP). Once connected, you can ask Claude or ChatGPT to pull recent conversations, read a thread, tag and assign chats, preview a template, or report on a campaign, and the client calls the right Wati tools for you. The official setup guide walks through the connection in a few minutes, and this guide shows you how to run day to day conversation management once you are in.

Prerequisites

Before you start, make sure you have:

A Wati account with a connected WhatsApp Business number. If you are new, Wati offers a 7 day free trial with zero fee WhatsApp setup and blue tick verification help.
Access to Claude or ChatGPT with support for custom MCP connectors.
Admin or operator permissions in the Wati workspace you want to manage.
The correct endpoint for your region. Standard workspaces use https://mcp.wati.io/mcp. EU hosted workspaces use https://eu-mcp.wati.io/mcp. Never mix endpoints from different regions.

Step-by-step

1. Connect Claude or ChatGPT to Wati via MCP

In your client's Connectors screen, add a custom connector named Wati with the MCP server URL for your region, then sign in through the browser. No API key is needed; authentication is OAuth based. The Wati MCP developer documentation lists the OAuth endpoints, the named tools, and the first use sequence in full.

2. Confirm the workspace you are connected to

Before any task, ask: "Which Wati workspace am I connected to?" The client calls wati_get_active_tenant and reports the workspace and your roles. If you manage several numbers, ask to list and switch workspaces, then confirm again. This one habit prevents almost every wrong workspace mistake.

3. Review recent conversations conversationally

Ask: "Show me conversations from the last 7 days." The client calls wati_get_conversations, which returns a recent activity snapshot. To go deeper, name a contact: "Read the latest messages with Sarah." The client searches for the contact and pulls messages newest first, so you can follow the thread without opening the shared Team Inbox at all.

4. Act on a conversation

You can triage entirely by chat. Ask the client to tag a conversation, assign it to an operator by email, or update its status to open, solved, pending, or blocked. Each request maps to a specific tool call, and the client confirms the action before it runs.

5. Send templates and check results with guardrails

For outbound messages, work in two moves. First ask for a preview and a cost estimate, which uses a dry run and sends nothing. Then, after you confirm the variables, ask the client to send the approved template to a number or a reviewed recipient list. Afterwards, ask "How did my last campaign perform?" to get sent, delivered, read, replied, and failed counts. The campaign management through AI chat guide shows more of these workflows end to end.

6. Bring in AI where it helps your team

MCP handles the chat driven workflow. Inside Wati itself, Wati AI adds Ask Wati AI for instant answers about your account data, Wati AI Agents for background tasks that wait for admin approval, and in inbox tools like AI Conversation Summary and AI Translate. For fully automated customer facing replies, Astra is the AI agent that talks to your customers, with human handoff when a conversation needs a person. The AI chat with WhatsApp guide explains how approval guardrails keep that combination safe.

Common pitfalls

Using the wrong regional endpoint. Global and EU workspaces use different MCP URLs. If sign in fails, check which region your workspace is hosted in first.
Skipping the workspace check. If you manage multiple WhatsApp numbers, always confirm the active tenant before reading or sending anything.
Treating a dry run as a send. Previews and cost estimates send nothing. Only an explicit send request with confirmed variables goes out.
Expecting full history from a snapshot. The recent conversations tool returns a recent activity snapshot, not complete history. Ask for a specific contact to page back through older messages.
Forgetting that accepted is not delivered. After sending, check message status or delivery webhooks rather than assuming delivery.

Frequently Asked Questions

How do I manage customer conversations through Claude or ChatGPT instead of a dashboard? Connect your Wati workspace to Claude or ChatGPT through Wati's MCP server, confirm the active workspace, then ask for what you need in plain language. The client calls Wati tools to pull conversations, tag and assign chats, send approved templates, and report on campaigns, so the dashboard becomes optional.

Do I need to write code or manage API keys to set this up? No. The hosted MCP server uses browser based OAuth sign in, so you add a connector, sign in, and start chatting. Developers who want deeper control can use the Wati API and webhooks documented in the developer docs.

Can the AI send messages without my approval? You stay in control of outbound sends. Previews and cost estimates use a dry run that sends nothing, and broadcasts are sent only after you review recipients and confirm. For automated customer facing replies, Astra handles conversations with human handoff available.

Does this replace the Wati inbox entirely? It replaces most navigation, not the inbox itself. The shared Team Inbox remains the place where your team collaborates in real time, while MCP driven chat is the fastest way to look things up, triage, and act from Claude or ChatGPT.

Conclusion

Managing customer conversations through Claude or ChatGPT instead of a dashboard is not a workaround. It is a genuinely faster operating model: you describe the outcome, the MCP connection does the navigation, and Wati's guardrails keep sends and approvals under your control. Set up the connector, make the workspace check a habit, and start with read only questions before you triage and send. When you are ready to see it live on your own number, book a demo with the Wati team.