Before you begin¶
By the end of this tutorial, you will have loaded a small set of files into MOI, made them searchable, and used an agent to produce a downloadable result. Complete these five checks first so you can follow the tutorial without interruptions.
What you need |
Ready when |
|---|---|
An MOI address and account |
You can sign in and see the current workspace name in the top-left corner |
A workspace |
The object names used by the tutorial are still available |
Permissions |
You can create import tasks, workflows, and knowledge bases, and run an agent |
Models and Credit |
Default models are available in the workspace and the account has enough Credit |
Tutorial data |
The data package has been downloaded and extracted into two files |
1. Access MOI¶
If you use MOI Cloud, open MOI Cloud and sign up or sign in with your email address or phone number. If your organization runs its own MOI deployment, use the address provided by your administrator.
Enter your own sign-in credentials. After you sign in, the current workspace name appears in the top-left corner.
Open the user menu from your avatar in the top-right corner, then go to Account Management. Make sure the email address or phone number you use to sign in is verified.
2. Prepare a workspace¶
A workspace is the top-level boundary for resources in MOI. The import task, workflow, and knowledge base created in this tutorial will all belong to the workspace you are currently using.
The workspace switcher in the top-left corner shows the current workspace. If you do not have one yet, or want a separate workspace for the tutorial, open the switcher and click Create Workspace. For details, see Workspaces.
The tutorial uses the names moi_quickstart, moi_quickstart_pipeline, and moi_quickstart_knowledge. Use a workspace in which these names are not already taken.
3. Check your permissions¶
The tutorial creates a import task, a workflow, and a knowledge base, then runs an agent. Each action is controlled by your workspace role.
Open the relevant page and check its Create action. If the action is disabled or missing, ask a workspace administrator to assign the required role. See Users and Roles and permissions.
4. Confirm models and Credit¶
The tutorial uses three types of models: a document processing model to parse files, an embedding model to create vectors, and a language model for the agent. Leave these fields at their workspace defaults throughout the tutorial.
Genesis provides these model services. To see the models available in your environment, open the Genesis Model Catalog.
Model calls consume Credit. Before you begin, open Billing Center from the account menu and check the balance. For details, see Credit and pricing.
5. Download the tutorial data¶
Download and extract the tutorial data package. It contains the two text files you will upload in step 2.
Start the tutorial¶
When all five checks pass, continue to Build your first data application. Each step ends with a completion check so you can confirm the result before moving on.