Real estate data API, CLI, and MCP for AI agents

Ploti is now the real estate data layer for AI. Here is every way in.

Brennen Fountain

Founder & CEO, Ploti

Ploti gives AI agents real estate data through three interfaces: an MCP server for assistants like Claude, ChatGPT, and Gemini, a REST API at api.ploti.ai/v1 for programs and agents that call HTTP, and a command-line tool for agents that work by using a computer, like Meta's Muse. All three use the same tools and the same API keys, can request data Ploti does not have yet, and can watch datasets for changes.

Over the last few months we kept seeing the same thing: the people who got the most out of Ploti were asking for it from somewhere else. They worked in Claude or ChatGPT, built agents in Grok Bot or their own code, and wanted Ploti's data there, not in another tab.

So we rebuilt Ploti around that. Ploti is the real estate data layer for AI. Our data agents keep fetching and compiling public records from county and city governments, the parcels, zoning, permits, meeting records, and utility data behind real estate decisions, and every assistant, agent, and program you use can now reach all of it the same way.

This post covers what shipped: the MCP tools, the REST API, the CLI, and what they have in common.

One set of tools, three ways in

Everything Ploti can do is defined once and offered through every interface, so an assistant in ChatGPT, a script calling the API, and an agent using the CLI can find, query, and request exactly the same data.

  • MCP for assistants that support connectors: Claude, ChatGPT, Gemini, Cursor, Claude Code, Codex, and others.
  • The REST API for programs and for agents that call HTTP, such as Grok Bot and dots.
  • The CLI for agents that work by using a computer, running commands the way a person would, such as Meta's Muse.

They all sign in with the same API keys and work within your workspace's coverage areas.

MCP: Ploti inside your assistant

Add Ploti as a connector at https://mcp.ploti.ai/api/mcp and your assistant gets tools to:

  • Find data across the catalog and see which counties you cover.
  • Query it with Python and SQL in Ploti's sandbox, against the full dataset rather than a sample.
  • Show results as interactive maps and tables, or as files to download.
  • Request what is missing. If the data is not in Ploti yet, the assistant starts a research request, Ploti works out which datasets it needs, and sources them, including from permit portals and meeting archives, or by emailing the county.
  • Watch for changes. Ask it to follow a dataset and it can list what changed since the version you last saw.

In ChatGPT, Ploti also supports MCP events, so instead of asking again, the assistant hears when a data request is ready, a research report is done, or a dataset you watch is updated.

Setup guides: Claude, ChatGPT, Gemini, Cursor, Claude Code, Codex.

The REST API: Ploti for programs and agents

The API lives at https://api.ploti.ai/v1 and takes a Ploti API key as a bearer token. Finding data in a county is one call:

curl -X POST https://api.ploti.ai/v1/find-data \
  -H "Authorization: Bearer $PLOTI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"queries": ["flood zones"], "place": "Travis County, TX"}'

From there a program can read your coverage, run any of the MCP tools through POST /v1/tools/{name} (including Python and SQL analysis), start research requests for data you need, and follow the data requests Ploti files for them. Two features are built for programs that run on their own:

  • Webhooks. Register an endpoint and Ploti sends signed events when a data request is ready, a research report is done, or a dataset you watch changes.
  • Dataset watches. Watch a dataset, then ask for the changes since a given version, for example the new permits in a city's permit records since your last check.

The API describes itself for agents: an OpenAPI spec at /v1/openapi.json, a plain-text guide at /v1/llms.txt, and Ploti's skills at /v1/skills.

Agents without MCP: one line

Some agents, like Grok Bot and dots, work over HTTP rather than MCP. To connect one, tell it:

Set up Ploti for me by following https://api.ploti.ai/v1/setup.md

It asks you for an API key and keeps it as a secret rather than in the chat, checks which counties you cover, installs Ploti's skills so it knows how to use the API, and offers to register a webhook so it hears about updates.

The CLI: Ploti for agents that use a computer

Agents that work by using a computer, like Meta's Muse, are often more at home running a command than calling an API. The ploti command gives them the same operations. Sign in with ploti login: it shows a code, you approve it in Ploti and pick a workspace, and no key is pasted anywhere. Then every operation is a command, for example:

ploti find-data "flood zones in Travis County"

ploti commands lists what is available, ploti help <command> shows what a command takes, and ploti skills install <folder> saves Ploti's skills wherever the agent keeps them.

API keys that only do what they need to

Owners and admins create keys in Ploti under Settings, API keys, and choose what each key may do:

  • read: search the catalog, read coverage and requests.
  • requests: file data and research requests, which can spend credits.
  • webhooks: register webhooks and subscribe to events.
  • sessions: open sessions and run analysis in Ploti's sandbox.

A key behind a dashboard can read without being able to request new data, and a key for an agent that watches permits needs nothing more than read and webhooks.

What it costs

The API, CLI, and MCP tools are part of every Ploti plan, priced per county and starting at $50 a month, with a 7-day trial. Data requests use credits as the data is delivered.

Get started

Start a trial, pick the counties you work in, and connect your assistant from its setup guide above, or create an API key under Settings and point your program or agent at https://api.ploti.ai/v1. To see the kind of work this makes possible, read how a land investor pulled zoning across 15 counties, or what a real estate data layer for AI is in the first place.

Frequently asked questions

Does Ploti have a real estate data API?

Yes. The REST API at https://api.ploti.ai/v1 finds datasets, reads coverage, runs analysis in Ploti's sandbox, requests missing data, and sends webhooks when data changes. Its OpenAPI description is public at https://api.ploti.ai/v1/openapi.json.

What is the Ploti MCP server?

An MCP server at https://mcp.ploti.ai/api/mcp that gives assistants such as Claude, ChatGPT, Gemini, Cursor, and Codex tools to find, query, and map real estate data, request what is missing, and watch datasets for updates.

How do I connect an agent that does not support MCP?

Tell it: Set up Ploti for me by following https://api.ploti.ai/v1/setup.md. It asks you for an API key, checks your coverage, and installs Ploti's skills so it can use the REST API from then on.

Can API keys be limited?

Yes. Each key carries only the scopes you give it: read, requests, webhooks, and sessions. A key for a dashboard can read without being able to spend credits on new data requests.

Do I need a paid plan?

The API, CLI, and MCP tools work on any active Ploti plan, including the 7-day trial. Pricing is per county.