Most “AI + BI” products put a chatbot on top of a chart.
CustomerDashboard’s remote MCP server goes further: it lets your AI assistant build and publish the dashboard itself—including secure logins for each customer.
You can connect data, inspect schemas, write queries, add charts, create customer accounts, configure branding, and publish a working client portal without leaving ChatGPT, Claude, or Cursor.
The client-dashboard problem
Many businesses still share customer data by emailing spreadsheets. The alternative is often a custom portal that takes months to build.
What sounds like “a few charts behind a login” quickly becomes a substantial software project:
- Authentication and account management
- Database connections and queries
- Charts, tables, and KPI widgets
- Customer-specific filtering
- Row-level data isolation
- Branding, hosting, and sharing
- Ongoing maintenance
CustomerDashboard packages those capabilities into one hosted dashboard platform.
Connect your existing data and create the views your customers need. For a multi-customer dashboard, write queries using the :customer_id parameter. When a customer signs in, CustomerDashboard supplies their ID so they see only their own rows.
What MCP adds
The Model Context Protocol, or MCP, gives AI assistants a standard way to work with external tools.
CustomerDashboard’s remote MCP server makes the same dashboard operations available from the assistant you already use. Instead of navigating through every screen manually, you can describe what you want:
Create a client dashboard with monthly revenue and a conversion table, filtered for each customer.
The assistant can inspect the available tools, build the dashboard, preview its queries, create customer access, and publish the result.
Because it is a hosted remote MCP server, there is nothing to install locally. You connect using the MCP URL and sign in through OAuth—no CustomerDashboard API token to copy or manage.
If you are completely new to CustomerDashboard, create your workspace once on the website before connecting your assistant.
Connect your assistant
The remote MCP endpoint is:
https://app.customerdashboard.io/mcp
ChatGPT
Add CustomerDashboard as a connector using the remote MCP URL. When ChatGPT asks you to authenticate, sign in to your CustomerDashboard account and approve access.
Claude or Cursor
Add the remote endpoint to your MCP configuration. In Cursor, this is typically configured in mcp.json.
Your client should discover the server’s authentication requirements and prompt you to sign in. The initial authorization challenge—sometimes shown as a 401 response—is part of the normal OAuth flow.
Once connected, your assistant can discover the available CustomerDashboard tools and begin working with your workspace.
Walkthrough: build your first multi-customer dashboard
Here is what an end-to-end workflow can look like.
1. Create your account
Sign up for CustomerDashboard and start the 14-day trial. No credit card is required.
If this is your first time using the product, create your workspace on the CustomerDashboard website. After that, you can handle most of the dashboard-building workflow through MCP.
2. Connect your database
Ask your assistant to connect the database that contains the data you want to expose.
For example:
Connect this Postgres database and list the tables I’ll need for affiliate payouts.
Depending on your database and network configuration, you may need to allow CustomerDashboard’s connection IP through your firewall.
3. Inspect the schema
Before building anything, ask the assistant to inspect the available tables and columns. It can help identify relevant data, relationships, customer identifiers, timestamps, and useful metrics.
This is especially handy when you understand the business question but do not know every detail of the database schema.
4. Create a customer dashboard
Create a dashboard in customer mode, then add the views your customers will navigate between—for example:
- Overview
- Revenue
- Conversions
- Payouts
Customer dashboards are designed for signed-in external users, with data filtered separately for each customer.
5. Write and preview customer-scoped queries
Queries on a customer dashboard must use :customer_id.
For example:
SELECT
month,
revenue,
conversions
FROM affiliate_performance
WHERE customer_id = :customer_id
ORDER BY month;Ask the assistant to preview and validate each query before adding it to the published dashboard.
6. Add widgets
Turn the query results into the format that best answers each question:
- KPI cards for headline numbers
- Tables for detailed records
- Line charts for trends
- Bar charts for comparisons
- Other supported visualizations for the data at hand
You can describe the desired outcome without specifying every setting:
Add a KPI for total revenue, a monthly revenue chart, and a conversion table sorted by the newest activity.
7. Add customers and logins
Create each customer using the identifier that matches your database records, then give the appropriate people access.
For example:
Add customer Acme with ID acme-123 and create a login for jane@acme.com.
When Jane signs in, queries using :customer_id are scoped to Acme’s ID.
8. Theme and publish
Ask the assistant to apply your branding, review the dashboard, publish it, and return the share URL.
Your customers now have a hosted portal they can sign in to—without your team building and operating a separate dashboard application.
Things you can ask CustomerDashboard MCP
You can use natural-language requests throughout the process:
Connect this Postgres database and list the tables I’ll need for affiliate payouts.
Create a client dashboard with revenue by month and a conversion table, filtered per customer.
Add customer Acme with ID acme-123 and a login for jane@acme.com.
Audit this dashboard for widgets that could leak data across customers.
Publish it and give me the URL.
You can also ask informational questions such as:
What widgets are available?
Explain the current plans and limits.
Public product information can be explored without signing in. Actions involving your workspace require authentication.
Internal dashboards and TV displays
The same MCP server can also create an internal dashboard for your own team.
Internal dashboards are useful for operational metrics, executive reporting, office displays, and TV dashboards. They can be shared publicly or protected with a password, depending on the audience. Views can also auto-rotate for unattended displays.
The workflow is the same: connect data, create queries and widgets, configure access, and publish.
Safety is part of the workflow
Giving an AI assistant operational access to a dashboard platform requires clear safeguards. CustomerDashboard’s MCP integration includes several of them:
- Credentials are redacted so sensitive connection details are not returned in ordinary tool responses.
- Customer dashboards require
:customer_id, helping prevent widgets from accidentally returning unscoped data. - Named deletes use a two-step process, reducing the chance of removing the wrong resource.
- OAuth permissions are conservative by default, so clients request only the access they need.
You can also ask the assistant to audit a dashboard for possible cross-customer data exposure before publishing it.
Who it is for
CustomerDashboard is a strong fit for organizations that need to expose data to customers without turning a dashboard into a full software-development project, including:
- Affiliate networks
- Publishers
- Agencies
- Marketplaces
- Managed service providers
- New SaaS companies that need a client panel now
It is not intended to replace your core product. It is also not designed around maintaining separate Sheets-only portals for every customer.
It is for the space in between: when customers need a polished, secure view of their data and you need to deliver it quickly.
Get started
Start with a 14-day trial—no credit card required—and connect the assistant you already use.
- Product: customerdashboard.io
- MCP documentation: app.customerdashboard.io/docs/mcp
- Remote MCP endpoint:
https://app.customerdashboard.io/mcp
The assistant is not a sidecar chatbot sitting beside your dashboard. It is another way to operate the same dashboard platform your customers already log into.
