AI Studio¶

AI Studio is your unified workspace for data ingestion, data processing, knowledge bases, agents, and governance.

Use cases¶

A workspace is the basic isolation boundary for resources, members, and permissions.
The entry points available in AI Studio vary by your account permissions, deployment mode, and feature flags.
We recommend that you start with the shortest path: connect data → process data → build a knowledge base → use an agent.

Make sure you are signed in to the correct account and have switched to the target workspace.
Make sure your account has permission to view the target resources and perform the corresponding operations.
When external systems are involved, prepare network connectivity, a least-privilege account, and valid credentials in advance.

Item	Current implementation
Page entry	The workspace home page at `/$workspaceId/`; depending on `AUTH_MODE`, account flows are handled by the account center or built-in pages.
Service contract	`AUTH_MODE` supports `sso`, `builtin`, and `cloudsigma`; the workspace ID is included in page routes and the backend request context.
Core objects	Account session, Workspace, Effective role, Feature flag, Product entry

Open AI Studio and confirm the current account and workspace.
Select Platform administration, Data connection, Data processing, Knowledge bases, or Agents based on your business objective.
Use Monitoring and alerting to view execution status, and use User permissions to manage access boundaries.
Go to Genesis when you need to integrate models, or to MatrixOne when you need to manage database instances.
When you need programmatic access, go to the developer documentation and follow the authentication and API instructions for the relevant platform.

State or signal	Interpretation and action
Signed in / signed out	Determines whether you proceed to the account page or the workspace.
Workspace selected	Resource lists, permissions, and subsequent routes are all scoped to that workspace.
Entry visible / hidden	Determined jointly by permissions, the deployed product, and runtime feature flags.

The page shows a success status, and the object name, workspace, and update time match the operation you performed.
The result persists after you refresh the page, and the next operation that depends on the object can read its latest configuration.
If the operation fails, save the error message and request ID, then check permissions, network connectivity, credentials, and the status of dependencies.

Symptom	Recommended action
Redirect loop after sign-in	Check the account center URL, callback URL, cookies, and system time.
Workspace is empty	Confirm that your account has joined the workspace and that its effective roles have synchronized successfully.
Menu item is missing	Check feature flags first, then resource permissions and the deployed version.

Identity and permissions: Use a least-privilege account in the target workspace, and verify separately that the read, write, execute, and delete permissions meet the requirements.
Changes and rollback: Record object IDs, versions, execution IDs, and change times; prepare rollback paths for deletion, overwriting, and reindexing.
Capacity and stability: First use a small, representative batch to validate timeouts, concurrency, quotas, failure retries, and downstream impact, and then gradually expand the scope.
Observability: Correlate page status with service logs, task timelines, or operation logs instead of relying only on a single success message.

Note

This page is based on the current MOI 5.0 product model. If you cannot see an entry, first check the workspace, permissions, and feature flags.