CopilotScopeWork in progress

StudioLens documentation

    Fabric setup guide

    The Fabric build lands Copilot Studio transcripts and message-credit consumption as Delta tables in a Lakehouse for scheduled ingestion, larger volumes, longer history, and the PPAC message-credit pages. Notebooks do the parsing in Spark; the Power BI template is a thin client over the Lakehouse SQL endpoint.

    1. Create a Lakehouse

    In a Fabric workspace on a capacity (F2+ or trial): create a Lakehouse and note its SQL endpoint (<workspace-guid>.datawarehouse.fabric.microsoft.com) and the Lakehouse Name.

    2. Permissions

    Register an Entra app with Microsoft Graph application permissions (admin-consented), and grant environment access per notebook:

    For Permission
    Transcript parser Dataverse Read on ConversationTranscript (System Customizer / Environment Maker)
    Credit consumption Global or Billing Administrator to export PPAC MCSMessages reports
    Agent 365 registry (optional) CopilotPackages.Read.All + Application.Read.All

    3. Run the core notebooks

    Import each notebook, attach and pin the Lakehouse as default, fill the # === CONFIG === cell, and run:

    Notebook Output tables (dbo)
    Copilot_Agent_Transcript_Parser.ipynb agent_sessions, agent_turns, agent_errors, agent_subagents, agent_catalogue, agent_performance
    Copilot_Credit_Consumption_Ingester.ipynb credit_consumption_tenant, credit_consumption_agent, credit_consumption_user

    Org data is optional (copilot_org_data) for department slicing.

    IDSL-FABRIC-001

    CaptureCapture the Fabric Lakehouse explorer after the StudioLens core notebooks have run, with the notebook CONFIG cell visible using synthetic values. Show the agent_* and credit_consumption_* tables; exclude workspace names, lakehouse names, tenant identifiers, environment URLs, secrets, and customer records.

    Proposed altFabric Lakehouse with StudioLens transcript and credit tables after the core notebooks run.

    Optional preview only. Copilot_Agent365_Registry_Ingester_PREVIEW.ipynb writes dbo.agent365_catalog, which is not the agents_365 table the current model reads. It is optional preview/reference material — it is not part of the core Fabric run and does not power the current report. Do not run it as part of normal setup.

    4. Handle secrets

    Keep the app’s client secret in Azure Key Vault and reference it via notebookutils rather than hardcoding it. Use synthetic placeholders only — never commit real tenant IDs, environment URLs, or secrets.

    5. Connect the template

    Open StudioLens - Fabric.pbit and set:

    Parameter Value
    Fabric SQL Endpoint <workspace-guid>.datawarehouse.fabric.microsoft.com
    Lakehouse Name Your Lakehouse name
    Enable_Dataverse Include (agent pages)
    Enable_Consumption Include (Credit Consumption page)

    The correct toggles are Enable_Dataverse and Enable_Consumption. Click Load, then Publish.

    IDSL-FABRIC-002

    CaptureCapture the StudioLens - Fabric template's Edit Parameters dialog using synthetic values. Show Fabric SQL Endpoint, Lakehouse Name, Enable_Dataverse = Include, and Enable_Consumption = Include; exclude real endpoints, workspace names, tenant identifiers, credentials, and secrets.

    Proposed altStudioLens Fabric template parameters with Enable_Dataverse and Enable_Consumption set to Include.

    6. Schedule the refresh

    Schedule the notebooks (or a Fabric pipeline) to run on a cadence, then enable dataset Scheduled refresh in the Service so the report tracks new transcripts and credit data.