AI assistants are most useful when they can work with your actual data. Connect Insightful to a tool like Claude and you can ask about your workforce data in plain language - and if it's connected to your other tools too (Jira, SAP, your HRIS), it can combine them: in Claude Cowork, for example, one conversation can pull a team's utilization and check it against their open Salesforce deals.

The connection works through MCP (Model Context Protocol), an open standard that lets AI tools securely request data from other apps. You don't need to understand how it works - setup comes down to pasting a URL and signing in. Below you'll find the endpoint you'll need, followed by instructions for Claude, ChatGPT, and other MCP-compatible tools.

Before you begin

You'll need:

An Insightful account - during setup you'll sign in with your own Insightful credentials and approve the connection. The AI tool can only access data your account can access.
An AI tool that supports MCP - Claude, ChatGPT, or any client that can connect to remote MCP servers.

Find your endpoint

The endpoint is the URL your AI tool will use to reach Insightful. Use the one that matches the region your organization's Insightful data is hosted in - it's the same region as the app URL you normally use to log in.

Region	Endpoint
United States (default)	https://app.insightful.io/mcp
European Union	https://eu.insightful.io/mcp
Saudi Arabia	https://sa.insightful.io/mcp

Copy the URL for your region - you'll paste it in the steps below.

Connect your AI tool

Setting up Claude

Open Settings → Connectors.
Select Add custom connector.
Paste your regional endpoint URL and confirm.
You'll be prompted to sign in and authorize access to your Insightful organization. Approve the request.
That's it. Once connected, the Insightful tools (utilization, app usage, productivity labels, employees, teams, etc.) become available to Claude in that workspace.

Setting up ChatGPT

Turn the MCP connection On in Insightful

By default, the toggle for the MCP connection is Off in the Insightful app. For it to work, the organization's admin must first go to Settings → Connected Apps → Manage Access → Read Only MCP tools to turn it on.

As well, the Insightful page should be opened at the moment of processing the connection.

Setting up ChatGPT side

ChatGPT must be installed on the device, and the user must be an admin, as well the version must be paid.

Click your profile icon → Account → Plugins → Developer mode, and toggle Developer mode on.
Back in Plugins, select Create connector (or New connector).
Fill in: Name (e.g. “Insightful”), Description (optional), and URL - your regional endpoint from the table above.
Set Authentication to OAuth and save. Complete the sign-in and authorization flow when prompted.
In each conversation where you want to use it, click the + in the message box > More → Developer mode, then toggle the Insightful connector on for that chat.

Setting up MCP in ChatGPT app

Go to your profile icon → Settings → Plugins → MCPs.
Add a new MCP connection.
Enter the connection details:
- Name: Insightful
- Connection type: Streamable HTTP
- Server URL: https://insightful.io/mcp
Next to Insightful, you should see an Authenticate button. Click it and complete the authentication flow.
Once authentication is complete, close ChatGPT and refresh the Insightful dashboard.

In order for a customer to successfully set up the connection, when adding the new plugin they should go to Advanced settings and change the Client registration method from the drop-down to Dynamic Client Registration (DCR). Without this step, the connection may not complete correctly.

Open Advanced settings, and set Client registration method to Dynamic Client Registration (DCR).
Save the connection.
In Insightful, confirm ChatGPT appears under Settings → Connected Apps.

Setting up other MCP clients

Most MCP-compatible tools (Cursor, Windsurf, Cline, Claude Code, Codex, and others) accept mcpServers JSON. Add Insightful to your tool's MCP config file, making sure the url points to your regional endpoint:

{
  "mcpServers": {
    "insightful": {
      "url": "https://app.insightful.io/mcp"
    }
  }
}

The location of the config file differs from tool to tool, so consult your tool's documentation. Any client that supports remote (HTTP) MCP servers with OAuth should work - on first use you'll be asked to sign in to Insightful and approve access, just like in the steps above.

The example above uses the United States endpoint. If your organization is hosted in the European Union or Saudi Arabia region, replace the URL with the matching endpoint from the table.

Work Intelligence tools (early access)

In early access, Work Intelligence tools are also available through MCP: your AI assistant can list and read your organization's saved AI reports, and run the discovery flow to generate a new one - starting the analysis, polling for progress, answering any clarifying questions it asks, and asking follow-up questions once it's done. Report content comes back as plain text.

These tools only appear for organizations that have the corresponding permission enabled, separately from the general MCP toggle above.

You're all set

Once connected, simply ask your AI assistant a question about your workforce data - it will pull the answer from live Insightful data.

How to set up Insightful MCP server