Dashboard

Create data visualizations and reports using natural language. DB-GPT converts your questions into SQL queries and renders the results as interactive charts.

How it works

1. You ask a question about your data in natural language
2. DB-GPT generates the appropriate SQL query
3. The query runs against your connected database
4. Results are rendered as charts, tables, or reports

Getting started

Prerequisites

• A database connected to DB-GPT (see Data Sources)
• Test data loaded (optional — use the built-in examples)

Using the Dashboard

1. Navigate to Chat in the sidebar
2. Select Chat Dashboard mode (or start a Dashboard conversation)
3. Choose your target database from the dropdown
4. Ask a question about your data

Example questions:

Show me monthly sales trends as a line chart
What are the top 5 products by revenue? Show as a bar chart
Create a pie chart of customer distribution by region

Chart types

DB-GPT's visualization engine (GPT-Vis) supports:

Chart Type	Best For
Bar Chart	Comparing categories
Line Chart	Trends over time
Pie Chart	Proportions and distributions
Table	Detailed data display
Scatter Plot	Correlations between variables
Area Chart	Cumulative trends

Guiding visualization

Include the desired chart type in your question for more precise results: "Show monthly revenue as a line chart".

Loading test data

DB-GPT includes sample data for testing:

# Linux / macOS
bash ./scripts/examples/load_examples.sh

# Windows
.\scripts\examples\load_examples.bat

This loads sample datasets into SQLite that you can query immediately.

Tips for better results

• Be specific — "Show the total order amount per month for 2024" works better than "Show me some data"
• Name the chart type — Mention "bar chart", "line chart", etc. for targeted visualizations
• Reference column names — If you know the schema, use actual column names for precision
• Iterate — Refine your question based on the initial results

Next steps

Topic	Link
Connect more databases	Data Sources
Chat modes overview	Chat
Text2SQL fine-tuning	Fine-Tuning