Skip to content
English
Contact Support
  • There are no suggestions because the search field is empty.

TractIQ AI Connector (MCP) Technical Overview

What is the TractIQ Connector?

The TractIQ Connector is an integration that allows AI assistants like ChatGPT and Claude to access TractIQ to gather data. Instead of exporting data manually from TractIQ and pasting it elsewhere, a user can ask the AI a question in plain English and the AI will fetch live TractIQ data on their behalf.

The Connector uses MCP (Model Context Protocol), an open standard originally published by Anthropic and now adopted across the major AI platforms (OpenAI/ChatGPT, Anthropic/Claude, and others). MCP is a universal plug-in format for AI assistants. When ChatGPT is configured with the TractIQ Connector, it gains the ability to call a fixed set of TractIQ tools, nothing more.

What is a "tool" in this context? A tool is a single, pre-defined action the AI can call. Each tool has a specific purpose (e.g., "look up a facility by ID," "get demographics for a 3-mile radius"), a defined set of inputs it accepts, and a defined set of outputs it returns. The AI cannot invent new tools or call TractIQ in ways the Connector does not explicitly support. The full list of 10 tools is in the "Available Tools" section below.

What the Connector is not:

  • It is not software that gets installed on the customer's network or endpoints.
  • It is not a general-purpose API or database connection.
  • It does not have write access to the customer's systems. Data is returned to ChatGPT only. Anything the user does with that data downstream is governed by ChatGPT's own permissions and connected apps.

 

Security Summary

The Connector is designed to be low-risk from an IT approval standpoint for the following reasons:

  • Hosted endpoint, no software to deploy.

The Connector is a hosted service at https://app.tractiq.com/mcp. Adding it to ChatGPT is a configuration change inside ChatGPT. Nothing is installed on the customer's devices or network.

 

  • Fixed tool set, not a general query interface.

The Connector exposes a curated set of tools with defined inputs and outputs rather than an open-ended query surface. This meaningfully reduces the risk of malformed or injected queries producing unexpected behavior.

 

  • Authentication uses the user's existing TractIQ login.

Each user authenticates with their own TractIQ credentials via a standard OAuth flow. No separate provisioning, shared service accounts, or API keys are required. Rate limits and quotas are tied to the individual user account. TractIQ does not receive any new PII from ChatGPT or Claude during authentication. The flow uses the credentials the user already has with TractIQ, and the AI platform is not a source of identity data for us.

 

  • Minimal data exposure from the customer side.

TractIQ does not receive the user's full ChatGPT conversation. We only receive the specific inputs required to fulfill each tool call (e.g., a set of coordinates, a facility ID, a radius in miles). We do not see the user's prompts, uploaded files, or other conversation context.

 

  • Prompt injection surface area is limited.

The Connector returns structured market data from TractIQ's own database, not scraped or user-generated content. The one data field worth calling out is facility website URLs, which are returned as part of facility records. In practice, this doesn’t create any “new risk” that wouldn’t have otherwise existed if you have web access enabled and the URLs are only accessible to the AI if independent web access enabled in the first place. If web search is disabled in the customer's ChatGPT or Claude configuration, the model has no mechanism to fetch those URLs, and they function as plain text in the response.

 

  • Data validation through the tool design.

Each tool enforces sensible inputs and outputs, and includes front-matter instructions that tell the AI how and when to use it. This reduces the likelihood of the AI misinterpreting results or pulling incorrect data.

 

If you use TractIQ's API, receive a bulk data export, or are using our AI Connector, we also encourage you to review the API License and Datafeed Agreement.

 

How the Data Flow Works

A typical request looks like this:

  1. The user asks ChatGPT a question (e.g., "What facilities exist within 3 miles of 500 Elm Street in Dallas?").
  2. ChatGPT decides which TractIQ tool(s) to call and with what parameters.
  3. ChatGPT sends a structured request to the TractIQ Connector endpoint (e.g., “geocode_search(query="500 Elm Street, Dallas")).
  4. The Connector authenticates the request against the user's TractIQ account, runs the query, and returns structured JSON.
  5. ChatGPT uses that data to formulate its answer to the user.

               

 

What ChatGPT sends to TractIQ: Only the parameters required for the specific tool call. Typically a coordinate pair, an address string, a facility ID, or a radius. ChatGPT does not send the user's full prompt, uploaded files, or conversation history.

What TractIQ returns to ChatGPT: Structured JSON containing the requested market data (facility records, demographics, pricing, etc.). No customer PII is returned because none is stored. The data is about self-storage facilities and their markets.

 

Technical Specifications

Protocol: Model Context Protocol (MCP), an open standard for AI tool integration.

Endpoint: https://app.tractiq.com/mcp

Authentication: Individual TractIQ user login via OAuth. Per-user rate limits and daily quotas.

Rate limits:

  • 60 requests per minute (infrastructure-level protection against excess load).
  • Daily tool call quotas by subscription tier: Pro (25/day/user), Pro+ (50/day/user), Max (500/day/user).

Data freshness: Supply data refreshes daily. Pricing data is collected on an ongoing basis, with each data point timestamped (YYYYMM format).

Transport: HTTPS (TLS-encrypted) from the AI platform to the TractIQ endpoint.

 

Available Tools

The Connector exposes 10 tools, grouped into four categories. Each tool has defined inputs and outputs, and none of them accept arbitrary queries beyond their defined scope.

Location tools

  • Geocode Search. Converts an address or place name to latitude/longitude coordinates. Inputs: address string. Outputs: coordinates, formatted address, confidence score.
  • Reverse Geocode. Converts coordinates to an address. Inputs: lat/lng. Outputs: nearest street address.

Facility tools

  • Facility Search. Finds self-storage facilities near a point. Inputs: lat/lng, radius (max 25 miles), optional filters. Outputs: list of facilities with basic attributes (name, address, size, owner, management type, distance).
  • Facility Lookup. Returns detailed information on a single facility. Inputs: facility ID. Outputs: 60+ fields covering physical attributes, ownership, tax and parcel data, construction details, and current pricing.

Market analysis tools

  • Market Summary. Returns a supply/demand snapshot for a trade area. Inputs: lat/lng, radius. Outputs: facility counts, square footage metrics, key supply ratios (sqft per capita), basic demographics.
  • Demographics. Returns detailed census-level demographics for a trade area. Inputs: lat/lng, radius. Outputs: population, household, income, and age breakdowns.
  • Pricing Trends. Returns historical pricing data for a trade area. Inputs: lat/lng, radius, months of history (max 36), optional unit size filter. Outputs: monthly rate data by unit size and climate control.
  • Contour. Generates radius polygons around a point as GeoJSON. Inputs: lat/lng, radii. Outputs: GeoJSON polygons for mapping.

Deal management tools

  • List Deal Lists. Shows the user's saved deal list folders (view of the user's own TractIQ account).
  • List Deals. Shows the user's saved deals, with optional filters by list and pipeline status (view of the user's own TractIQ account).

 

Data TractIQ Receives vs. Returns, Concrete Example

If a user asks ChatGPT "Show me what facilities exist in Austin, TX," here is what TractIQ actually sees:

Sent to TractIQ:

  1. geocode_search(query="Austin, TX")
  2. facility_search(lat=30.2672, lng=-97.7431, radius_miles=3)

Returned to ChatGPT:

  1. Coordinates for Austin, TX.
  2. A list of self-storage facilities near those coordinates, with attributes like name, address, size, owner, and management type.

TractIQ does not see the user's original question, their ChatGPT conversation, any uploaded files, or anything else outside these two structured tool calls.