Skip to main content

Industry Solutions

Industry solutions are deployable AI workspaces designed around a business problem. A solution may include a Workbench, agents, workflows, tasks, datasets, sample records, knowledge configuration, and navigation that work together immediately after deployment.

Find a solution​

  1. Open Explore from the Platform Hub.
  2. Select the industry-solutions view.
  3. Review the solution's target user, business pain, outcome, value pillars, quick start, and primary workflow.
  4. Check which resources and sample data will be created.

Choose a solution based on the workflow and expected outcome, not only its industry label.

Deploy to a new space​

  1. Select Deploy on the solution card.
  2. Choose Create a new space.
  3. Review the deployment summary.
  4. Confirm.

Use this option when you want clean resource isolation and do not need to combine the solution with an existing team's data.

Deploy to an existing space​

You can deploy into an existing space that you own.

  1. Choose Existing space.
  2. Select the target space.
  3. Read the overwrite and conflict warning.
  4. Complete the second confirmation.

Back up or publish important Workbench drafts first. Existing-space deployment can reuse or update resources with matching solution identities.

Deployment progress​

Deployment runs as a background job. The progress dialog reports stages such as validation, space preparation, data resources, agents, workflows/tasks, Workbench, and finalization.

  • You may close the dialog; deployment continues in the background.
  • Returning to Explore restores a running job when possible.
  • A transient status-check failure does not mean the deployment stopped.
  • If deployment fails, the highlighted step and error identify where it stopped.
  • Retry reuses completed resources where the server can do so safely.

Do not click Deploy repeatedly while a job is running.

After deployment​

  1. Open the generated or updated space.
  2. Open its Workbench in run mode.
  3. Test the primary workflow with sample data.
  4. Open included agents in Chat and verify their authorized datasets or knowledge.
  5. Review Datasource and Dataset records.
  6. Check task status.

Manual and event-triggered tasks supplied by an industry solution are enabled by default so the solution is usable immediately. Review schedules and external integrations before enabling any additional scheduled task.

Language and sample data​

The selected catalog language controls the installed template language:

  • English solutions use English titles, labels, prompts, workflows, and sample data.
  • Chinese solutions use Chinese titles, labels, prompts, workflows, and sample data.

Field labels are stored with the solution and are not translated dynamically after installation.

Upgrade safely​

When a newer solution version is available:

  1. Open the solution or installed-application upgrade entry.
  2. Review the version notes and affected resources.
  3. Start the upgrade.
  4. Monitor the background upgrade job.
  5. Open the generated Workbench draft.
  6. Review component bindings, pages, navigation, and local customizations.
  7. Publish the new draft only when it is ready.

The upgrade keeps the currently published Workbench active until you publish the new draft. Use Workbench version history to compare or restore application configuration. Dataset contents and external database changes follow their own data lifecycle and are not rolled back simply by restoring a Workbench version.

Customization guidance​

  • Preserve stable resource identifiers used by bindings.
  • Add organization-specific fields through documented schemas.
  • Test agents after changing Dataset fields or vectorized source fields.
  • Keep required output fields aligned with the Dataset schema.
  • Use least-privilege Datasource operations.
  • Record local changes before applying a future solution upgrade.

Troubleshooting​

ProblemAction
Deploy button becomes available while status is uncertainReopen Explore and check the running deployment before starting another job
Dataset records are missingInspect the failed data step, model/service availability, and Dataset statistics
Workbench opens but a page is missingConfirm the draft was saved and the expected version was published
Agent cannot find expected recordsVerify Dataset binding, schema, vectorization status, output fields, and tool choice
Existing customization changed after upgradeReview the new draft and restore the earlier Workbench version if needed