Skip to content
New: estimate and preview audiences from the CLI

Build audiences from real‑world behavior

Query Intuizi’s Large Behavioral Model, our flagship large quantitative model (LQM), trained on de-identified, real-world behavioral signals. Build audiences, size them, and deliver them to your partners - from your code, a terminal, or an AI agent.

Get started API Reference
How many coffee lovers visited a coffee shop in San Francisco last week?
curl -X POST https://console.intuizi.com/api/v2/analyses/audiences/estimate \
  -H "Authorization: Bearer $INTUIZI_TOKEN" \
  -H "Accept: application/json" -H "Content-Type: application/json" \
  -d '{"name": "Coffee lovers - SF - 1 week",
       "datasets": [{"type": "POI", "analysisdata": [101, 102, 103],
         "signal_providers": ["BID001"],
         "start_date": "2026-09-14", "end_date": "2026-09-20",
         "location": {"countries": ["USA"], "states": ["CA"],
                      "cities": ["San Francisco"]}}]}'

The coffee brand and signal provider ids come from the catalogs. The CLI and an agent look them up for you.

API v1 retires on December 31, 2026. Everything in these docs is API v2: start new work on v2, and plan your move well ahead of the date.

Three ways to build

One model, with the same rules and limits however you connect. Mix them freely: build an audience from the CLI and read it from your code.

The API

For your own code and integrations. One REST API, with examples in cURL, Python, JavaScript, and PHP.

An AI agent

Connect Claude, Cursor, or your own agent to the MCP server, and ask for what you need in plain words.

The CLI

Build, size, and deliver audiences from a terminal, a script, or a CI job, with names looked up for you.

How it works

Every audience follows the same path, whichever way you connect.

Step 1Look upFind brands, places, apps, geography, and every other catalog by name. Step 2BuildCombine datasets and filters into an audience, and estimate its size first. Step 3DeliverSend it to your partners' platforms: all of it, or only the devices seen most often. Step 4Get notifiedA webhook tells you the moment a build or a delivery is done.

Popular guides

Create an Audience
Build an audience from one or two datasets, and wait for it to complete.
Activate an Audience
Deliver a finished audience to an endpoint connection, and read the results.
Build a Lookalike Model
Grow a finished audience into a larger one of similar devices.
Cloud Cohort to Activation
From a file in your own S3 or GCS to a cohort, an audience, and a delivery.
Preview, then Activate
Count the devices a frequency range would deliver, then deliver exactly that.
Automate with an AI Agent
Look up, create, wait, and deliver, safely and unattended.

Explore the docs

Getting Started
From an Intuizi account to your first request.
Concepts
How Intuizi works, in eleven short pages.
API Reference
Every endpoint, parameter, and response.
Changelog
What changed in the API, MCP, and the CLI.
Questions about your account or your data? Talk to your Intuizi account team. Found a problem with the CLI? Open an issue on GitHub. Ready to build? Open the Intuizi console.