Intuizi CLI
Build, size, and deliver audiences from your terminal.
The Intuizi CLI is one small program, intuizi. What you do in the Intuizi
console or with API calls, you can do with a command: type it yourself, or put
it in a script or a CI job.
Install
brew install intuizi/intuizi-cli/intuiziThen log in once, with the email and password you use for the Intuizi console:
intuizi auth loginIt runs on macOS, Linux, and Windows. Getting Started has the details, including how to use it in CI.
See it in action
# Find the brand you want to target
intuizi reference poi brands --search starbucks
# See how many devices the audience would hold, without creating it
intuizi audiences estimate create --type poi --brand starbucks --country USA \
--start-date 2026-09-02 --end-date 2026-09-09 --name "Starbucks - 1 week" --wait
# Build it, and wait until it is ready
intuizi audiences create --type poi --brand starbucks --country USA \
--start-date 2026-09-02 --end-date 2026-09-09 --name "Starbucks - 1 week" --wait
# Deliver it to one of your endpoint connections
intuizi activations create --audience-id <id> \
--endpoint-connection-id <id> --pricing-model-id <id> --datastream <id> --waitNames like starbucks are looked up for you, so you rarely need to find an id
first. The Quickstart walks through these steps one at a
time.
What you can do
Build audiences
From real-world behavior: places visited, apps, web and CTV activity, purchases, and more. Use brand and category names instead of ids.
Size before you build
Estimate how many devices an audience would hold, without creating it.
Deliver
Send an audience to your destinations: all of it, or only the devices seen on several days.
Bring your own data
Import your customer lists as cohorts, and submit your own locations.
Keep it fresh
Rebuild an audience every day, week, or month on a schedule.
Script it
JSON output, ids alone, and clear exit codes, for scripts and CI pipelines.
CLI, API, or MCP?
All three query the same Large Behavioral Model, with the same rules and limits:
- CLI - for people at a terminal, and for scripts and CI jobs.
- API - for your own code and integrations.
- MCP Server - for AI agents.
Every CLI command uses the documented API endpoints, so whatever works in the CLI works in the API too.