Connect Centralize to Claude via MCP

Use the Centralize MCP connector to bring your account intelligence into Claude, ChatGPT, and other AI tools. Ask questions in plain English and get live answers from your Centralize data. Your AI tool can also combine this data with your other connected tools, for example Notion or PostHog.

Contents

• Before you start

• What you can do

• How to connect

• Read vs. write access

• Available tools

• Troubleshooting

• Need help?

Before you start

• MCP is available on the Trial, Pro, and Enterprise plans. It is not available on the Free plan.

• In Enterprise workspaces, users with the Partner role cannot use MCP.

• Claude sees only the data that you can see in Centralize. If your team can see only some accounts, Claude sees only those accounts.

• Each AI tool has its own rules for adding connectors. For example, in a company ChatGPT workspace, your ChatGPT admin may need to enable Centralize before you can connect it. See the steps for your tool below.

What you can do

After you connect, you can ask Claude questions like these:

• "What accounts are at risk of churn this quarter?"

• "Summarize the latest activity at Acme Corp."

• "Who should I multithread at Envoy, and do we have any warm intro paths?"

• "What is the recommended next action for our Stripe opportunity?"

• "Show me the call strategy for my meeting with Figma tomorrow."

• "Which contacts at Acme are suggested for outreach based on our personas?"

• "Which of our contacts changed jobs in the last 30 days?"

Claude gets live data from Centralize. It can combine this data with context from your other connected tools.

How to connect

The Centralize MCP URL is https://app.usecentralize.com/api/mcp.

Option 1: Claude.ai

Use this option to connect your own Claude account. You do not need an API key.

1. Open Claude.ai and go to Settings.

2. Click Connectors.

3. Click Add custom connector.

4. Enter Centralize as the name.

5. Enter https://app.usecentralize.com/api/mcp as the URL.

6. Click Add.

7. When Claude asks you to connect, log in to your Centralize account to authorize the connection.

Note: Anthropic does not control third-party connectors. Only use connectors from developers that you trust.

Option 2: Admin-managed Claude workspace

On a Claude Team or Enterprise plan, your Claude admin can add the Centralize connector for your whole workspace. In this case, you do not add the connector yourself. The first time that you use the connector, log in to Centralize to authorize it.

Option 3: Other MCP clients

You can also connect other MCP clients that support remote MCP servers, for example Cursor. Add a server with the Centralize MCP URL. If your client supports OAuth login, log in to Centralize when the client asks you to.

If your client does not support OAuth login, use an MCP API key. To create a key, see Creating and Managing MCP API Keys. Then send the key in the Authorization header as a bearer token. The configuration format is different for each client. This example uses the common mcpServers format:

"mcpServers": {
  "centralize": {
    "url": "https://app.usecentralize.com/api/mcp",
    "headers": {
      "Authorization": "Bearer YOUR_MCP_API_KEY"
    }
  }
}

Note: Keep your API key secret. Anyone with the key can get your Centralize data through MCP.

Option 4: ChatGPT

Centralize is available as a plugin in ChatGPT, so you can add it without any custom setup. The plugin includes the Centralize app and a set of skills for common workflows like warm introductions, pipeline reviews, and org chart building.

1. Open the Centralize plugin in ChatGPT, or click Plugins in the ChatGPT sidebar and search for Centralize.

2. Click Install plugin.

3. Log in to your Centralize account if you're asked to.

4. On the Authorization Request page, review the access ChatGPT is requesting (Centralize - Read and Centralize - Write), then click Authorize.

5. You'll return to ChatGPT, where you can start asking questions about your Centralize accounts, or try one of the suggested prompts on the plugin page.

Read vs. write access

There are two access levels. When you create an MCP API key, you choose one of them:

• Read-only: Claude can search and read your Centralize data. Claude cannot make changes.

• Read & write: Claude can also use the tools marked Write in the tables below. For example, Claude can update org charts, set buying roles, or create saved views.

If Claude does not have write access, Centralize declines requests to use a write tool. Claude cannot create or delete contacts, accounts, opportunities, or notes through the connector. This is true for both access levels.

Available tools

Claude has access to 34 tools through the Centralize MCP. A check in Read means that the tool works with a read-only key. A check in Write means that the tool can change data in Centralize and needs a Read & write key.

Search and discovery

Account and deal context

Contacts and relationships

Activity and timeline

Org charts

Recommendations

Outreach requests

Saved views and account settings

Configuration

Troubleshooting

Need help?

If the Centralize connector does not authorize, does not show in your AI tool, or cannot get your data, contact support@usecentralize.com.