Get private revenue data into ChatGPT.

Option 1: add a custom MCP app in ChatGPT

In a ChatGPT workspace where custom MCP apps are enabled, open Settings → Apps or Connectors, enable developer mode if required, and create a custom app. Use VerifiedMRR as the name, https://verifiedmrr.com/mcp as the server URL, and OAuth as authentication.

Complete the browser sign-in and approve revenue:read. Start a new chat, enable the VerifiedMRR app from the tools menu, and ask: List my projects and compare net revenue for the current month.

Custom app controls and availability depend on your ChatGPT plan and workspace policy. If your workspace does not expose custom apps, use the Responses API method below or connect through Codex.

Option 2: use the OpenAI Responses API

Create a personal MCP token in VerifiedMRR Settings → Agent access and store it server-side. Pass the MCP endpoint and authorization token to the Responses API. Because every VerifiedMRR tool is read-only, the example disables per-call approval; keep approvals enabled when you add other servers that can write.

const response = await openai.responses.create({
  model: "gpt-5",
  input: "Compare net revenue by project this month.",
  tools: [{
    type: "mcp",
    server_label: "verifiedmrr",
    server_url: "https://verifiedmrr.com/mcp",
    authorization: process.env.VERIFIEDMRR_MCP_TOKEN,
    require_approval: "never"
  }]
});

Prompts that produce useful answers

Ask the model to state the date range and currency, distinguish gross from net revenue, and call get_sync_status before making a high-stakes comparison. Example: Check sync health, then compare project net revenue this month versus last month and explain the three largest changes.

For recurring reporting, make the model cite the project name, period, currency, and values returned by the tools. MCP supplies live context to the conversation; it does not permanently train ChatGPT on your revenue data.