Kadrin Docs Kadrin Batch

Get started with Kadrin for AI

Kadrin for AI lets an AI assistant look up facts about places while you talk with it. Ask "How many people live within a 10-minute drive of Boston City Hall?" and the assistant asks Kadrin, then answers with the number, its source and its margin of error (how far the true number could reasonably be from our estimate).

It's a connector that uses the Model Context Protocol (MCP), a standard way for AI assistants to use outside tools.

1. Make a key

Sign in to Kadrin Batch, open Keys, and make a key. A test key is free and gives made-up numbers, which is enough to check the setup. Make a live key for real numbers.

2. Add Kadrin to your assistant

For now the connector takes your key in a header. In Claude Code, run this in a terminal, with your key in KADRIN_API_KEY:

claude mcp add --transport http kadrin https://mcp.kadrindata.com/mcp --header "Authorization: Bearer $KADRIN_API_KEY"

Other apps that can connect to an MCP server over HTTP with a header work the same way. Give them:

Apps that add connectors only by signing in, rather than with a key, need OAuth (a standard way to let one app act for you in another). It arrives in January.

3. Ask about places

Ask in your own words. For example:

The assistant picks Kadrin's tools itself: finding an area by name, choosing measures, getting the numbers, comparing places. Each answer it gets lists the sources and years, so ask it to cite them if it doesn't.

What it costs

The same credits as the API: one credit for each group of measures for each place, and 2 more to draw a drive-time area. Finding places, searching the measures and checking your credits are free.

Next