Workspaces

Workspaces contain all the data and Package and Plugin instances within Steamship. It is highly recommended to create an instance of a Package in a new Workspace to avoid data conflicts.

Workspaces can be used in two ways:

As the backing store to Packages. Each Steamship Package instance is bound to a Workspace, giving it an isolated environment to store state and model parameters.
As a cloud environment for language AI projects. You can use Workspaces on their own from Jupyter notebooks or your own application code.

Creating Workspaces

Workspaces are created (or resumed) when creating a new Steamship client. In your client’s constructor, specify the workspace you’d like. If it doesn’t exist, it will be created for you.

from steamship import Steamship
 
client = Steamship(workspace="my-workspace-name")

import { Steamship } from "@steamship/client"
 
const client = Steamship({workspace: "my-workspace-name"})

curl \
    --header "Content-Type: application/json" \
    --header "Authorization: Bearer $STEAMSHIP_KEY" \
    --request POST \
    --data '{"handle":"your-handle", "fetchIfExist":true}' \
    https://api.steamship.com/api/v1/workspace/create

Every workspace-specific operation this client performs will now take part in that workspace. This includes file uploads, plugin training, plugin inference, queries, and any infrastructure associated.

If you need to switch your workspace, you can either create a new client or type:

client.switch_workspace("new-workspace-name")

Data Files