RelationalAI CLI Reference#

The RelationalAI Command Line Interface (CLI) is used to initialize RelationalAI projects and manage project resources.


The CLI is bundled with the RelationalAI Python Package and can be installed using pip or your favorite package manager:

pip install relationalai

See Install the relationalai Python Package for details.

Basic Usage#

The CLI is invoked using the rai command. Execute rai without any arguments to see a list of all available commands:


If you encounter the error command not found: rai, you may need to activate your project’s virtual environment.

To execute a command, type rai, followed by the command name and any arguments you want to pass. For example, the following creates a new RelationalAI engine named my-engine of size S:

rai engines:create --name my-engine --size S

To view help for a command, use the --help flag:

rai engines:create --help


Configuration and Profiles#

All RelationalAI projects require a configuration file, created using rai init, that specifies the project’s settings, platform credentials, and other configuration options.

Use the following commands to manage your project’s configuration and profiles:


Engines are the compute resources used to run queries and models. Use the following commands to manage engines:


Imports are objects imported into RAI, such as data from a Snowflake table, for use in models and queries. Use the following commands to manage imports:


Exports are objects that are exported out of RAI for use in other systems, such as exporting query results into a Snowflake table. Use the following commands to manage exports:


This feature is currently disabled but will be available soon!


Transactions are operations that modify the state of a RelationalAI model. They are created when you import data and when you run queries. Use the following commands to manage transactions: