You need to enable JavaScript to run this app.
Intelligent Data Insight

Intelligent Data Insight

Copy page
Download PDF
Analysis Assistant
Intelligent Dashboard Building
Copy page
Download PDF
Intelligent Dashboard Building

The intelligent data insight analysis assistant helps users build and modify dashboards intelligently, and assists users in interpreting dashboards. This article introduces how to use the analysis assistant's intelligent dashboard features to build and modify dashboards.

Feature overview

The intelligent dashboard building feature of the intelligent data insight analysis assistant includes intelligent building and intelligent modification. Users can describe their dashboard building and modification requirements in natural language, and the intelligent dashboard can quickly generate dashboards, improve building efficiency, optimize dashboard layouts, and enable agile and efficient data insights.

warning

This feature is a sub-feature under the value-added module "Analysis Assistant" and is disabled by default. To use this feature, contact Volcano Engine technical support. After purchasing and enabling this feature, you can use it.

Feature details

  • Intelligent building: Supports generating dashboards using natural language, helping users quickly build dashboards.
  • Intelligent modification: Supports switching dimensions / metrics / chart types, modifying period-over-period comparisons, adjusting layouts, and other operations.

Applicable scenarios

  • Enterprise data visualization analysis: Supports quickly generating multi-dimensional dashboards for core metrics, regional distribution, industry trends, and more using natural language. Suitable for management to monitor business data in real time (such as GMV, redemption rate, and other key metrics) and for operations teams to perform data insights.
  • Intelligent dashboard optimization and iteration: Enables dynamic adjustment of dimensions, metrics, layouts, or chart types (for example, switching from line charts to bar charts) for existing charts. Suitable for data analysts to iteratively optimize dashboards based on business needs and improve data presentation efficiency.

Feature access

Create a new dashboard

  1. Before using intelligent dashboard generation, the corresponding dataset must have the analysis assistant enabled. On the "Dataset" details page, click the "Enable DataWind Analysis Assistant" button.
    Image
  2. After enabling the analysis assistant for the dataset, open the dashboard list and click Create a new dashboard on the left.
    Image
  3. After configuring the dashboard name and path, click "OK" to enter the intelligent building page.

Existing dashboards

Go to the dashboard page, select the dashboard for intelligent building, and click "Edit".
Image
You can click "Intelligent Build" in the upper right corner of the dashboard editing page or the input box at the top of the canvas to enable intelligent building.
Image

Quick start

On the first screen of intelligent building, users can interact using natural language in the dialog box to quickly build a dashboard, or select "Add chart" from the left toolbar to initially build a chart and then use the intelligent building feature to optimize the dashboard. The first screen also provides many demo videos, which users can click to view.
Image
After enabling intelligent building, the top of the dialog box on the right provides input examples for chart modification, chart generation, and layout adjustment. Users can click to quickly experience the intelligent building effect.
Image

Generate charts

Use natural language to generate charts

  1. Click "New Dashboard". When entering the new dashboard editing page, the intelligent building page is displayed first.
    Image
  2. Describe the desired dashboard style in the dialog box using text. Use "@" to specify which dataset to use for building, and use "/" to quickly select a command template.
    Image
  3. During the intelligent dashboard building process, the analysis logic of intelligent building is displayed on the right side of the dashboard editing page, while the building process is simultaneously shown on the left.
    Image
  4. After intelligent building is complete, the required dashboard is generated. If needed, further modifications can be made based on the existing dashboard.
    Image

Generate based on existing charts

  1. Create a duplicate dashboard.
    Image
  2. Edit the duplicate dashboard and enable intelligent building.
    Image
  3. Generate more charts based on existing charts. For example: Enter "Refer to sales revenue over time and sales revenue by region, use profit from the same dataset to create two similar charts, place them at the bottom, and keep the same order." You can continue generating more charts.
    Image

Edit chart

Add metric card

Image

  1. In the dialog box, use "@" to search for the dataset you want to use to build the dashboard. After selecting it, press Enter to confirm.
  2. Describe your requirements for adding metric cards in natural language, for example, "Use the @xxx dataset to generate a metric card for each metric," and press Enter to send it to the intelligent dashboard.
  3. The left side of the dashboard building page will dynamically display the process of creating metric cards. If you are not satisfied with the generated results, you can click "Rollback" or enter natural language to make modifications.

Add dynamic dimension field

Image

  1. Click the chart where you want to add a dynamic dimension. A "Reference" label will appear in the upper right corner. Click this label to reference the chart in the dialog box.
  2. In the dialog box, enter in natural language the specific field of the dynamic dimension you want to add and the final effect you want to achieve, then press Enter to send it to the intelligent dashboard.
  3. The left side of the dashboard building page will dynamically display the process of creating metric cards. If you are not satisfied with the generated results, you can click "Rollback" or enter natural language to make modifications.

Add period-over-period and year-over-year comparison

In the dialog box, enter the chart and the period-over-period or year-over-year information you want to add to the chart.
Image

Add global filter

Based on existing charts, enter in the dialog box the specific field, type, scope of application, display method, and other information for adding a global filter to the dashboard, then press Enter to send it to the intelligent dashboard.
Image

Switch dimension/metric

For a single chart, you can switch dimensions/metrics.

  • Switch dimension: For example, enter in the dialog box "Switch the dimension of the @xxx chart to merchant province."

    Dimension before switching: Business date

    Dimension after switching: Merchant province

    Image

    Image

  • Switch metric: For example, enter in the dialog box "Switch the metric of the @xxx chart to redemption rate."

    Metric before switching: Redeemed GMV

    Metric after switching: Redemption rate

    Image

    Image

Switch chart type

You can switch chart types for a single chart. For example, enter "Switch the chart type of @xxx to bar chart."

Chart type before switching: Line chart

Chart type after switching: Bar chart

Image

Image

Edit period-over-period and year-over-year comparison

  • You can edit the name of period-over-period and year-over-year comparison. For example, enter "Change all period-over-period names on the metric cards to DoD, and week-over-week names to WoW."
    Image
  • You can modify the period-over-period format. For example, enter "For period-over-period decreases on the metric card, display the text in red and add an arrow."
    Image

Click "Edit chart" to go to the "Visual query" page to edit a single chart. When entering the chart name in the input box at the top of the canvas, chart names will be intelligently recommended and can be quickly filled in by pressing the Tab key.
Image
You can also click the chart, then click "Style" in the chart configuration area on the right. When modifying the chart name, you can also use the intelligent chart name recommendation feature.
Image

Adjust layout

Adjust chart layout

The layout can be adjusted so that each chart occupies one row. For example, enter "Adjust the layout so that each chart occupies one row."

Layout before adjustment: Four charts in a 2 * 2 distribution

Layout after adjustment: Four charts, each occupying one row

Image

Image

Equally divided chart layout

The layout can be adjusted so that each row contains two charts, equally divided. For example, enter "Two charts per row, equally divided layout."

Layout before adjustment: Four metric cards in four columns

Layout after adjustment: Four metric cards in a 2 * 2 distribution

Image

Image

Split Tab components

The layout can be adjusted to split charts into two Tab components. For example, enter "Split core metric trends and core metric distribution into two Tab components."

Layout before adjustment: One Tab component

Layout after adjustment: Two Tab components

Image

Image

Feature overview

Select objects

You can use the "@" symbol to select a chart or dataset as the operation object.
Image

  • When selecting a chart: You can modify or adjust the format of the specified chart.
  • When selecting a dataset: You can use the specified dataset to generate more charts on the current dashboard.

Use templates

You can use the "/" symbol to select command templates, including those saved in the "Command template" library and historical commands.
Image
Steps to create and use command templates:

  1. Click "Command template" above the input box to enter the command template library.
    Image

  2. Click the "Create" icon to create a new command template.
    Image

  3. Enter the relevant parameters to create a command template.
    Image

    Parameter

    Description

    Template name

    A brief description of the instruction template's function, which will be displayed on the homepage of the instruction template library to help users intuitively understand the template's purpose.

    Scope of disclosure

    Set the visibility scope for the command template; options include only yourself or the current project.

    • This dashboard command template is visible and available only to the creator.
    • This dashboard command template will be shared within the current project. Users with dashboard editing permissions can use this command template during the intelligent dashboard building process.

    Command content

    The core content of the command template is the stored command template, which can be called directly.

    Add as a quick command

    After setting, you can see the "Quick Command" widget in the left widget area. Click it to use this command template directly for intelligent dashboard building.
    Image
    After enabling, you can select the visible user scope:

    • Only yourself: Only you can see the "Quick Assistant" widget.
    • All visible users of the template: All users who can see the command template can see the "Quick Assistant" widget.
  4. Users can also hover over a command and save it directly as a template.
    Image

  5. Use command templates. You can enter "/" in the dialog box to invoke a command, or go to the "Command Template" library and click the "Add to Dialog" button to add the command to the dialog box. Additionally, the system records how many times each command is used, and all command templates are sorted in descending order by usage count. If the user has added commands to favorites, those will be displayed first.
    Image

  6. Users can also add command templates to favorites and filter them by "My Favorites" or "Created by Me". For templates the user created, they can edit, add to quick commands, or delete them. For templates created by others, users can view template details or add them to quick commands, but cannot edit or delete templates created by others.
    Image

Reference component

When using the smart build feature, in addition to selecting the target component using "@", users can also select a component directly on the dashboard. After selecting a component, a "Reference" button will appear in the upper right corner of the component. Clicking this button will automatically add the component to the list of operation targets for the smart build feature.

Operation rollback

To undo changes made by the smart build feature that do not meet expectations, or to revert the current modification, click the "Rollback" button after the smart build feature completes an operation. This will undo the current step and return to the previous step.
Image

Switch model

The smart build feature provides multiple model configurations, allowing users to switch between different models based on the complexity and stability requirements of the build task, in order to quickly and reliably build dashboards.
Image

Model

Introduction

GPT-4o

Better service stability

GPT-4.1

Better performance for complex tasks, but average service stability. For simple tasks, it is recommended to use GPT-4o.

Doubao-Seed-1.6

Self-hosted model, more stable service but relatively slow. Suitable for both simple and complex tasks.

Quick filter

The quick filter feature is available on both the dashboard editing page and the visual query page.
Users can enter natural language instructions in the filter box (for example, "show data of September only"), then press the Enter key or the send button. The system will automatically parse the instruction and generate the corresponding filter conditions.
Image
The generated result is as follows. Users can click Filter Settings to further adjust this filter or continue to add other filter criteria.
Image
In addition, the system will recommend filter criteria to users based on the current chart configuration and other filter criteria in the dashboard. Users can click filter criteria as needed or press the Tab key to insert the recommended filter criteria.
Image

Special statement

(1) All content output by this service is generated by artificial intelligence models.
(2) This feature is intended only for customers' internal, proprietary business, and personal use. Customers are not allowed to provide generative artificial intelligence services to the public through the Analysis Assistant.
(3) This feature is a value-added module and supports both SaaS and on-premises deployment versions. To purchase or try this feature, please contact sales personnel for consultation.
(4) Currently, the AI capabilities of DataWind Intelligent Data Insight support the following types of datasets:

  • Supports extracted datasets and datasets uploaded as Excel/CSV files.
  • Supports directly connected datasets, currently including only: ClickHouse / ByteHouse, Doris / selectdb.
Last updated: 2026.08.14 16:49:56