Applications in a Project#

Overview of Managed Applications#

Screenshot of the application section in the Project Container page. From left to right shows JupyerLab and a tutorial app.

Note

The best way to understand how applications work is work through the exercises in the Onboarding Project.

IDEs (JupyterLab, VS Code) or web apps like TensorBoard (or JupyterLab) are an important part of development environments.

AI Workbench lets you add and manage these kinds of applications to a project, and it has features for starting, stopping, monitoring, and logging through the UI.

You must configure an application in a project before it can be managed.

Applications are added on a project-by-project basis
Application dependencies must be installed in the project container
Application code must be in the repository
Once the dependencies are installed and the code is in place, you add an application by configuring launch and management instructions

Managed applications are divided into Classes and Types.#

Class	Type	Examples	Deployment	Requirements	Configuration
Web App	Project container	JupyterLab, TensorBoard	Runs in the project container	Dependencies installed in project container and relevant code in repository	Start/stop commands, port, and other configuration options
Process	Project container	Shell script	Runs in the project container	Dependencies installed in project container and relevant code in repository	Start/stop commands
Compose App	Multi-container	PDF-to-Podcast Blueprint	Runs in multiple containers	A compose file in the repository	compose yaml file
Native App	System application	VS Code	Runs locally with access to the project container	Varies by OS	Potentially requires manual steps

Adding Appplications to the Project Container#

Screenshot of the add application modal.

Add web-apps, processes and native apps in Environment > Project Container > Applications.

Web Apps and Processes#

Install application dependencies in the project container
Add the application code to the repository
Know the start/stop commands and the appropriate port number
Complete Environment > Project Container > Applications > Add > Create a custom app

Note

See Walkthrough: Add a Custom Application for an example.

Adding JupyterLab or VS Code#

If these applications are not already configured in a project, you can add them by clicking Add in Environment > Project Container > Applications > Add.

Configuring VS Code may require some extra manual steps depending on your container runtime or OS. See VS Code Integration for details.

Multi-Container Applications (Compose Apps)#

The Environment > Compose section allows you to configure, start, stop, and manage multi-container applications within AI Workbench.

Requirements to successfully configure a Compose application:

The Compose file must be in the top level of the repository
The Compose file must be valid and correctly formatted
The service name in the Compose file matches the name of the application

AI Workbench leverages service profiles to determine which services run within the application. If no profile is specified, the default profile is used.

For details see Multi-Container Environments.

FAQs#

Can I use the CLI to add and configure applications?#

Yes. The CLI has the same functionality as the Desktop App.

Is adding an application version-controlled?#

Yes. Adding web apps, processes or VS-Code creates an entry in the .project/spec.yaml file, and a Compose app requires a compose file in the repository.

Deleting an application in the Desktop App removes the entry from the .project/spec.yaml file, and removing the compose file from the repository removes the Compose app.