Set Up AI Query for Essbase

To enable and use AI Query in Oracle Essbase with Oracle AI Database 26ai, you must set up connections and vectorize the outline. For AI powered data analysis, you can then submit natural language queries (NLQ), which generate valid MDX queries you can save as reports.

Note:

Oracle Essbase includes features powered by artificial intelligence (AI) and Generative AI (Gen AI). While Oracle strives to provide accurate results, Gen AI outputs may be incorrect, incomplete, or inappropriate. Information provided by Gen AI features should be verified by users before taking action or making decisions based on it. Oracle is not liable for damages or losses arising from the use of Gen AI features.

Prerequisites

  • Install Oracle AI Database 26ai.
  • Deploy Essbase 21.8.1.0.0 or higher using the OCI Marketplace.
  • Only Service Administrator can create AI connections.

  1. Create Global Connections in Essbase

    In the Essbase web interface, under Sources:
    1. Create a Connection:

      Setup a connection to Oracle AI Database 26ai. For detailed steps, see Create a Connection and Datasource to Access Oracle Database.

    2. Create a GenAI Connection:

      Select GenAI for the connection type and complete the setup. For detailed steps, see Create a GenAI Connection.

  2. Associate an AI Connection with Your Application

    Note:

    Only users with Application Manager permission can associate or dissociate an AI connection with an application.
    1. Select your application, click the Disabled status next to AI Status, choose your AI connection from the drop-down menu.
    2. Click Agree to accept the terms and conditions before proceeding.
    3. Click Associate to complete the connection.
  3. Vectorize Outline
    1. Run the Vectorize Outline job to convert your database outline to vectors. For complete instructions, see Vectorize Outline.

      Note:

      You can track the job progress in the Jobs section of the Essbase web interface.
  4. Interact With AI Query

    Once the vectorization job is complete:

    1. Open your Application and select your Database.
    2. Click Analyze Data.
    3. Select Reports.
    4. Click AI Query to open the AI Query panel.

    Additional Options in the AI Query Panel

    When you launch the AI Query panel, you can select from system-generated (auto-populated) starter questions.

    AI Query features:
    1. Repopulate this query: Regenerates the answer based on the same question.
    2. Show MDX query: Displays the generated MDX query.
    3. Copy MDX query: Copies the generated MDX to your clipboard.
    4. Explain MDX: Provides a natural language explanation of the generated MDX query, making it easier to understand the query structure and purpose.
    5. Analyze query in Smart View: Opens the report in Oracle Smart View for further analysis in Excel.
    6. Clear conversation: Resets the session or conversation history.
  5. Natural Language to Report Generation (Smart View Reports)
    1. In the AI Query panel, click the Create Report icon.
    2. Enter a name for the report in the Name field and click OK. The report will appear in the Reports section in the left pane.
    3. Select the generated report to start working on it directly in the spreadsheet.
    4. To edit or delete a report, click the action menu next to the report and select Edit or Delete as needed.
  6. Natural Language to MDX Query Generation
    1. In the left pane, click Create, then select Use Natural Language Query.
    2. Enter your query in the Natural Language Query field and click Generate MDX.
    3. Click Validate to review the generated MDX.
    4. Once validated, click Save. The report will appear in the Reports section in the left pane.
    5. Select the generated report to start working on it directly in the spreadsheet.
    6. To edit or delete a report, click the action menu next to the report and select Edit or Delete as needed.
  7. Dissociate an AI Connection
    1. To dissociate an AI connection, open the application page and click the Disassociate connection icon next to AI Status, and then click Dissociate to confirm.

      Note:

      This action will start a background job that deletes all vector tables and related AI data created during outline vectorization in your Oracle AI Database schema, ensuring all associated data is removed.