The Fabric data agent answers natural-language questions about your data and returns visuals, such as charts and graphs, alongside its text responses. Those visuals now render using Fabric Visuals, bringing higher-quality, more consistent visualizations to your responses.
What's changed
Previously, the data agent rendered its charts with a separate, standalone visualization approach. The data agent now generates its visuals using Fabric Visuals, the same visualization foundation used in Fabric Apps and increasingly across Fabric. This means the charts you get in a data agent conversation are built on the same technology that powers visuals in other places in the product.
The move to Fabric visuals delivers improvements in chart fidelity and polish, including clearer formatting, more readable legends, better tooltips on hover, refined line thickness and markers, and smarter axis scaling. The result is a visual experience that is more polished and consistent with other Fabric experiences. The following example shows a data agent response rendered with Fabric visuals, including a multi-series line chart, clustered bar chart, and stacked area chart, generated from natural-language questions.
Fabric data agent responses rendered with Fabric Visuals, showing improved formatting, legends, and tooltips.
Why consistent, high-quality visuals matter
A visual answer is only useful if you can trust it and read it quickly. By standardizing on Fabric visuals, the data agent gives you:
- Higher chart quality. Charts are cleaner and easier to read, with improvements to formatting, legends, tooltips, markers, and axis scaling.
- Consistency across Fabric. Visuals in the data agent now look and behave in line with the broader Fabric experience, so there's less to relearn as you move between surfaces.
- Faster insight. Clear visuals help you identify trends, patterns, comparisons, and outliers at a glance, without leaving the conversational experience.
- A foundation for what’s next. We’re continuously investing in Fabric visuals, with more out-of-the-box improvements to look and feel, richer tooltips, interactivity, and accessibility to come.
See it in action
To generate visuals, you don’t need to enable any settings. The data agent responds with a visual when it determines that your intent is to generate one. For example, when you explicitly ask for a chart, when you ask it to "show" you something, or when it determines a visual would make the answer clearer.
Try prompts like the following:
- "Generate a bar chart of revenue by region."
- "Create a line chart of monthly orders for 2025."
- "Show me my top 10 customers by sales."
- "Visualize the monthly revenue composition by brand over time."
The data agent chooses the most appropriate chart type based on your question and the data returned. If you have a preference, include it in your question, for example "as a line chart" or "as a bar chart."
Supported chart types
The data agent currently supports the following visual types:
- Line and multi-line charts
- Bar and multi-bar charts
- Stacked bar charts
- Pie charts
- Scatter plots
- Area and stacked area charts
Visuals are supported across all data source types the data agent works with, including lakehouses, warehouses, and Power BI semantic models.
How visual responses help answer questions faster
Visual responses can help users understand trends, comparisons, and outliers more quickly than text alone. By presenting data graphically within the conversation, the data agent can help users explore results without switching to another experience.
Whether you're reviewing sales performance, tracking operational metrics, or analyzing customer activity, visual responses provide another way to understand the data returned from your questions.
Get started
Visuals are enabled by default, so the next time you ask the data agent a question where a chart helps, you'll see one in the response. To shape how and when the data agent uses visuals, you can add agent instructions, for example to prefer a specific chart type for certain questions.
Try visual responses in your Fabric data agent conversations today. For guidance on generating visual responses, see the documentation for Get visual responses from a Fabric data agent. To customize how visuals are used, see Best practices for configuring data agents. Share your feedback through the Fabric community or your existing feedback channels
Microsoft Employee
