Skip to main content

Managing Knowledge Indexes

Once you've created a Grounded Knowledge Index using the setup wizard, you can view its details, test queries, switch models, and manage its configuration.

Viewing Index Details​

To view an index's details:

  1. Navigate to Build > Knowledge
  2. Click on an index row to open its details page

Details Page Sections​

Overview Tab​

The Overview tab displays:

SectionInformation
ConfigurationIndex name, type, output mode, reasoning effort
Azure ResourcesConnected Azure OpenAI and Azure AI Search providers
StatisticsQuery count, last query time, document count
Component StatusStatus of index, knowledge source, and knowledge base
Knowledge SourcesSources linked to this index's Knowledge Base (for multi-source setups)

Topic Taxonomy Tab (Email/Corporate Knowledge Indexes)​

For email-type and corporate knowledge indexes, a second tab shows the configured topic taxonomy. See Topic Taxonomy for details.

Testing Queries​

You can test your index directly from the details page:

  1. Open the index details page
  2. Find the Test Query section
  3. Enter a sample query (e.g., "What are our return policies?")
  4. Click Run Query

Understanding Test Results​

The test results show:

  • Answer: The synthesized response from the Grounded Knowledge system
  • References: Source documents used to generate the answer
  • Confidence: How confident the system is in the answer
tip

If you get poor results, check that your documents have been indexed and that your query matches the type of content in your knowledge base.

Model Reconfiguration​

You can switch the Azure OpenAI model on an active index without re-provisioning. This is useful when upgrading to newer models (e.g., moving from gpt-4o to gpt-4.1-mini).

  1. Open the index details page
  2. Click the Edit button
  3. Follow the 4-step reconfiguration wizard

See Switching RAG Models for the full guide.

Multi-Source Knowledge Bases​

Knowledge Bases can aggregate multiple Knowledge Sources. From the index details page, you can:

  • Add Sources - Link additional Knowledge Sources to the Knowledge Base
  • Remove Sources - Unlink a Knowledge Source from the Knowledge Base

See Knowledge Bases for more details on managing multi-source configurations.

Index Status​

Indexes can have the following statuses:

StatusMeaningAction
ActiveFully operational and available as an agent toolNone needed
ProvisioningAzure resources are being createdWait for completion
ErrorSomething went wrong during provisioningCheck error message, retry or delete
DisabledManually disabled by administratorRe-enable if needed

Deleting an Index​

To delete a Grounded Knowledge Index:

  1. Open the index details page
  2. Click the Delete button (trash icon)
  3. Confirm the deletion in the dialog
  4. Optionally, check the Delete Azure resources checkbox to also remove the associated Azure AI Search index

What Gets Deleted​

When you delete an index:

  • The Outermind configuration is removed (index record, knowledge source, knowledge base)
  • The associated agent tool is removed
  • If the Delete Azure resources checkbox is selected, the Azure AI Search index is also deleted
  • Azure provider resources (OpenAI service, Search service) are not deleted, as these are shared resources that may be used by other indexes
warning

Deletion cannot be undone. Any agents using this index will lose access to it.

Troubleshooting​

Index Shows "Error" Status​

  1. Check the error message on the details page
  2. Common issues:
    • Azure credentials have expired or been rotated
    • Azure resource was deleted outside of Control Bridge
    • Quota limits exceeded on Azure services

Queries Return No Results​

  1. Verify documents have been indexed
  2. Check that the index has documents (see Statistics section)
  3. Try broader search terms
  4. Ensure the topic taxonomy is properly configured (for email indexes)

Can't Delete Index​

  1. Ensure you have the settings:write permission
  2. Check if the index is assigned to active agents (unassign first)
  3. If Azure resources fail to delete, they may need manual cleanup in Azure Portal

Best Practices​

  1. Regular Testing - Periodically test queries to ensure the index is working correctly
  2. Monitor Usage - Check query statistics to understand how agents are using the index
  3. Keep Credentials Fresh - Rotate Azure API keys periodically and update them in Control Bridge
  4. Clean Up Unused Indexes - Delete indexes that are no longer needed to reduce Azure costs
  5. Consider Model Upgrades - Use the model reconfiguration feature to upgrade to newer, more capable models as they become available