Skip to content
Model Match Help Desk

AI / MCP

Model Match exposes its mortgage and real estate data through MCP (Model Context Protocol), an open standard that lets AI assistants call live tools instead of relying on whatever they already know. Once connected, an assistant can query originators, agents, lenders, and loans by name — sourced from the same data that powers the app, not the open web — and the results respect your existing Model Match workspace roles and permissions.

Every connection authenticates with OAuth against your Model Match account, so there’s no API key to generate or paste anywhere. Pick your tool below.

Claude

Section titled “Claude”

Claude supports Model Match on claude.ai, Claude Desktop, and Claude Code. Free Claude accounts are limited to one custom connector.

  1. In Claude, go to Settings → Connectors (claude.ai/settings/connectors on the web).

  2. Select Add custom connector, name it Model Match, and paste the server URL: https://mcp.modelmatch.com/mcp.

  3. Click Add, then Connect, and sign in with your Model Match credentials when prompted.

  4. Start a new chat and open the tools menu (the + icon) to confirm Model Match is available.

Claude Desktop uses the same Settings → Connectors flow. If you’d rather configure it by hand, add this to claude_desktop_config.json:

{
  "mcpServers": {
    "modelmatch": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.modelmatch.com/mcp"]
    }
  }
}

For Claude Code, run claude mcp add --transport http modelmatch https://mcp.modelmatch.com, then run /mcp inside Claude Code and select modelmatch → Authenticate to finish sign-in in the browser.

ChatGPT

Section titled “ChatGPT”

Model Match requires a ChatGPT Plus, Pro, Business, Enterprise, or Education plan — custom connectors aren’t available on the free plan.

  1. In ChatGPT, go to Settings → Apps & Connectors, open Advanced settings, and turn on Developer mode.

  2. Back in Apps & Connectors, click Create, name it Model Match, paste https://chatgpt.modelmatch.com/mcp, and keep authentication set to OAuth.

  3. Click Create, then Connect, and sign in with your Model Match credentials. Allow pop-ups if the authorization window doesn’t appear.

  4. Start a new chat and type @Model Match find real estate agents in Charlotte, NC to confirm results come back.

Reference Model Match in any chat by starting your message with @Model Match followed by your request in plain English.

Cursor

Section titled “Cursor”

  1. Open Settings → Customize → Configure MCP servers in Cursor, or edit the config file directly — ~/.cursor/mcp.json for all projects, or .cursor/mcp.json in a project root for just that project.

  2. Add a modelmatch entry (the key must stay lowercase for the tool picker to pick it up):

    {
      "mcpServers": {
        "modelmatch": {
          "url": "https://cursor.modelmatch.com/mcp"
        }
      }
    }

  3. Save, reopen MCP settings, select the modelmatch entry, and authorize through the browser-based Model Match sign-in.

  4. Ask the agent something like “Search Model Match for the top loan officers in Austin, TX this year” to confirm production data comes back.

Replit

Section titled “Replit”

  1. Go to replit.com/integrations and find the MCP Servers section.

  2. Click + to add a new server, and enter Model Match as the display name — Agent shows this exact name whenever it logs a tool call, so a clear name helps you track what it’s doing.

  3. Paste the server URL: https://replit.modelmatch.com/mcp. Leave Advanced settings untouched — Model Match authorizes over OAuth, so there are no custom headers to add.

  4. Click Test & save, then complete OAuth sign-in in the browser window Replit opens.

Ask Agent a Model Match question — top loan officers in a market, for example — to confirm the connection. If Agent can’t reach Model Match, check the MCP Servers pane: an “unauthorized” status means sign-in didn’t finish.

Lovable

Section titled “Lovable”

  1. In Lovable, go to Connectors → All and find the Custom MCP card.

  2. Click Connect your own MCP.

  3. Enter Model Match as the server name and paste https://lovable.modelmatch.com/mcp. Leave authentication set to OAuth — there’s no bearer token or API key to enter.

  4. Select Add & authorize and sign in with your Model Match credentials.

Connections in Lovable are per-user, not workspace-wide, so each teammate needs to authorize with their own Model Match login. Every tool call then respects that person’s existing workspace permissions. To verify, ask the agent to search Model Match for the top loan officers in a market and confirm named originators with production data come back.

Perplexity

Section titled “Perplexity”

Model Match requires a Perplexity Pro, Max, or Enterprise plan.

  1. Go to Settings → Connectors, select + Custom connector, and choose Remote.

  2. Name it Model Match and paste https://perplexity.modelmatch.com/mcp.

  3. Open Advanced settings and set Transport to Streamable HTTP.

  4. Set authentication to OAuth and leave the client ID and secret fields empty — Model Match supports OAuth discovery and dynamic client registration, so none are needed.

  5. Check the acknowledgement box, click Add, and sign in with your Model Match credentials when prompted.

On an Enterprise plan, a workspace admin needs to enable Allow members to add custom connectors before teammates can add one, and share organization-wide connectors from the Permissions screen before others can use them. To verify your connection, ask Perplexity to search Model Match for the top loan officers in a market.

Microsoft Copilot Studio

Section titled “Microsoft Copilot Studio”

  1. Open the Tools page for your agent in Copilot Studio.

  2. Select Add a tool → New tool → Model Context Protocol.

  3. Fill in the server details: name it Model Match, write a clear description of what it can answer (this affects when the agent decides to use it), and set the server URL to https://copilot.modelmatch.com/mcp.

  4. Set authentication to OAuth 2.0, choose Dynamic discovery, and leave the client ID, secret, and callback URL blank — Model Match supports OAuth 2.0 dynamic client registration with discovery.

  5. Select Create, then Next. On the Add tool screen, choose Create a new connection, authenticate with your Model Match credentials, and select Add to agent.

Power Platform data policies govern this connection — an admin manages access restrictions through connector policy settings, not agent configuration. To test, ask the agent to search Model Match for the top loan officers in a market and confirm named originators with production data return.

What’s next

Section titled “What’s next”