Introduction to Workbench

Getting started with Workbench, Descartes Labs' hosted JupyterLab environment

Workbench is the best place to use the Descartes Labs Platform. If you navigate to your IAM landing page, you see that you can launch Workbench here.


Workbench is a cloud-based data science environment that combines the Descartes Labs Platform APIs, visualization tools, and sample models inside a hosted JupyterLab interface. We provide a 100 GB storage drive that persists across server restarts. When you first start a server, you will have several options available to you.


We generally recommend the first of these, which is the latest release and a CPU server. If you need a GPU server for performing graphics-intensive work, you can stop and restart your server with the GPU image.



Workbench is pre-loaded with some key code examples that we keep updated as our APIs and Python client evolve. These are located in the example_notebooks folder. In order to run these examples, and develop your own code, you will need to authenticate with our APIs.


The initial setup notebook takes you through authenticating in this way. If at any point in time you encounter an error stating that your client ID has not been found, you need to re-run this notebook.


The examples in Workbench provide a wealth of content generated by our Applied Science team. The two folders in the example notebooks directory, examples and guides, are there for you to run and examine as you learn how to use our platform.

Keep in mind that this example notebooks directory is effectively read-only; any files you save there will not be saved if the server restarts. It is recommended that you create a different directory, or directories, for your work.