Forum Discussion

icassiem's avatar
icassiem
Post Prodigy
2 months ago
Solved

AI Copilot Narratives

Good day,  I got approval for an F2 sunscription 🙂 , and i want to embed the copilot summaries/naarratives and i see i new release it is dynamic and it auto updates 1. Can i do in depth summaraies...
Need Help
  • Parchitect's avatar
    Parchitect
    2 months ago
    Hi  , 
    Yes, you are on the right track, but I would separate the options a bit.

     

    1. Visual summaries
    For per-visual summaries, you do not normally build this as a “right-click hyperlink” type experience. The main options are: 
    - Use the Copilot pane to summarize report content / visuals.
    - Use the Narrative visual with Copilot if you want a narrative visual placed directly on the report canvas.

     

    The Narrative visual with Copilot can summarize selected visuals, pages, or report context, and you can use prompts to control the type of summary. 
    Reference: https://learn.microsoft.com/en-us/power-bi/create-reports/copilot-create-narrative

     

    2. Page narrative with slicers
    Yes, for a page-level narrative, use the Narrative visual with Copilot on the report page.
    Typical setup:
    - Add slicers, for example Client, Quarter, Half Year, Year.
    - Add the visuals that represent the performance analysis.
    - Add the Narrative visual with Copilot.
    - Prompt it to summarize the selected client/period/page.

     

    The narrative is grounded in the report canvas, so clearly named visuals, fields, axes, and measures are important. Microsoft notes that the Narrative visual pulls information from what is on the report canvas, not directly from the full underlying semantic model. 
    Reference: https://learn.microsoft.com/en-us/power-bi/create-reports/copilot-create-narrative

     

    Microsoft also announced “AI Narrative auto refresh” in the March 2026 Power BI update, so that is the feature to validate if you want the narrative to refresh dynamically as slicers/filter context changes. 
    Reference: https://community.fabric.microsoft.com/t5/Power-BI-Updates-Blog/Power-BI-March-2026-Feature-Summary/ba-p/5173928

     

    3. RAG / chatbot
    I agree with your thinking: I would not start with a full RAG/chatbot implementation immediately on F2 unless you have a clear use case and usage pattern.

     

    4. Performance essay / extended narrative
    Yes, this is a good use case for the Narrative visual with Copilot.
    Example setup:
    - Page slicers: Client, Year, Quarter, Half Year
    - Report visuals: revenue, margin, volume, variance, forecast, trend
    - Narrative visual prompt:
    • “Write an executive performance analysis for the selected client and selected period. Include revenue movement, margin movement, major drivers, risks, outliers, comparison to previous period, and suggested follow-up actions. Structure the response as an executive summary with bullet points and a conclusion.”

    The prompt/instruction is part of the narrative visual setup, so you can design the narrative once and reuse it as users change slicers. I would test the exact refresh behavior in your tenant because AI Narrative auto refresh is a newer feature and may depend on rollout/settings.

     

    5. Enabling Copilot

    At a high level, your admin needs to check these:
    - You need a paid Fabric capacity. F2 or higher is supported for Fabric Copilot Capacity; trial SKUs are not supported.
    - The Fabric/Power BI admin must enable Copilot tenant settings.
    - The workspace must be assigned to a supported Copilot-enabled capacity.
    - If your region requires it, the admin may need to enable cross-geo data processing for Azure OpenAI.
    - Users must have the required Power BI/Fabric permissions and licenses for the content.

     

    References:
    https://learn.microsoft.com/en-us/fabric/fundamentals/copilot-enable-fabric
    https://learn.microsoft.com/en-us/power-bi/create-reports/copilot-enable-power-bi
    https://learn.microsoft.com/en-us/fabric/enterprise/fabric-copilot-capacity

     

    My suggested rollout for you on F2:
    1. Start with Narrative visual with Copilot on one report page.
    2. Test slicer-driven client/period summaries.
    3. Validate AI Narrative auto refresh behavior.
    4. Monitor capacity usage.
    5. Only then test Fabric Data Agent for a more advanced conversational/RAG-style experience.

     

Parchitect's avatar
Parchitect
Solution Sage
1 month ago
Hi icassiem,

 

No problem at all.
For your two last questions:

 

1. What does the Azure / Fabric admin need to enable?
There are a few separate steps. Some are done by the Azure admin, and some by the Fabric / Power BI admin.

 

Azure admin / capacity owner:
- Create or confirm a paid Fabric capacity, F2 or higher.
- Make sure the capacity is in a supported Fabric region.
- Assign the correct capacity admin(s).
- Make sure the workspace that contains your report is assigned to that Fabric capacity.

 

Fabric / Power BI admin:
- Go to the Fabric Admin portal.
- Enable “Users can use Copilot and other features powered by Azure OpenAI”.
- Scope it to a security group first if you want a controlled rollout.
- If required for your geography, enable the setting that allows data sent to Azure OpenAI to be processed outside the capacity’s geographic region/compliance boundary.
- Enable/confirm Copilot at the capacity level if delegated capacity settings are being used.
- Make sure users have the required Power BI/Fabric access to the report/workspace.

 

Useful Microsoft links:
Enable Copilot for Power BI: https://learn.microsoft.com/en-us/power-bi/create-reports/copilot-enable-power-bi

 

Enable and configure Copilot in Microsoft Fabric: https://learn.microsoft.com/en-us/fabric/fundamentals/copilot-enable-fabric

 

Fabric Copilot Capacity: https://learn.microsoft.com/en-us/fabric/enterprise/fabric-copilot-capacity

 

Fabric region availability: https://learn.microsoft.com/en-us/fabric/admin/region-availability

 

Since you mentioned North Europe / Ireland: North Europe is listed as supporting Power BI and Fabric workloads. However, Copilot also depends on the Copilot tenant settings, capacity setup, and Azure OpenAI processing settings. So I would still ask the admin to validate the Copilot settings directly in the Fabric Admin portal.

 

2. Do you need F4, or is F2 enough?
F2 is enough to enable Copilot features from a licensing/capacity prerequisite perspective. Microsoft documents Fabric Copilot Capacity as supported from F2/P1 and above.

 

So you do not need F4 just to make Copilot narratives available.

 

However, F2 is the smallest SKU. That means it is fine for a pilot or controlled rollout, but you should monitor capacity usage carefully. AI narrative generation, Copilot interactions, and Fabric Data Agent requests consume capacity. If many users start generating narratives or asking AI questions at the same time, F2 can run out of headroom faster.

 

My recommendation:
- Start with F2.
- Enable Copilot for a small security group first.
- Test one report with Narrative visual + slicers.
- Validate AI Narrative auto refresh behavior.
- Watch the Fabric Capacity Metrics app.
- If usage is consistently high or users experience slow/throttled behavior, then move to F4 or higher.

 

So short answer:
- F2 = enough to start and validate the use case.
- F4 = not required for feature availability, but may be needed for more users, more frequent AI usage, or better operational headroom.

 

For your case, I would not panic yet. Start with F2, prove the narratives and executive summaries work, then use capacity metrics to justify F4 if needed.
 

🔍Parchitect
Solutions Architect · Microsoft Fabric Specialist

đź’ˇHelpful? Kudos are appreciated.
✔️Solved? Mark as Solution so others can find it faster.