Installation
Installation
Parlant (available on both GitHub and PyPI) has been tested on multiple platforms (Windows, Mac, and Linux), so you should find the installation to be nice and easy.
Note that Python 3.10 and up is required.
Local Installation (Recommended)
We recommend installing Parlant using pipx, to isolate your operating system's package versions from user-installed ones. This is probably the healthier choice for your system.
If you don't have it yet, follow their installation instructions to set it up. It should take you less than a minute, and it's a really great tool to have on your system!
When you're ready, go ahead and install Parlant. Don't worry if it takes a few minutes!
$ pipx install parlant
Then choose your provider and run the server:
$ parlant-server --help
$ export OPENAI_API_KEY="..."
$ parlant-server --openai
Global Installation
Alternatively, use pip
to add parlant
to your global Python environment.
$ pip install parlant
Then choose your provider and run the server:
$ parlant-server --help
$ export OPENAI_API_KEY="..."
$ parlant-server --openai
Verifying Your Installation
To test your installation, head over to http://localhost:8800 and start a basic interaction with the default agent. You can add more agents later.
Creating Your First Guideline
While the server is running (in another terminal), try adding a guideline for the default agent using the CLI client to see how it responds in a new conversation session.
-
$ parlant guideline create \
--condition "the user greets you" \
--action "offer a drink" - Now head over to http://localhost:8800, start a new session, and greet the agent. You should expect to be offered a drink!

Installing Client SDK(s)
To create a frontend app that interacts with the Parlant server, we recommend installing our native client SDKs. We currently support Python and TypeScript (also works with JavaScript).
$ # For building a frontend client in Python
$ pip install parlant-client
$ # For building a frontend client in TypeScript/JavaScript
$ npm install parlant-client
For other languages—they are coming soon! Meanwhile you can use the REST API directly.