Skip to main content
Create interactive charts directly in Chat from uploaded datasets or connected data sources. Describe what you want to understand, or ask for a specific chart type. Charts in Chat are powered by Atta. For a step-by-step example, follow Create charts and explore data.

Chart types

Choose a chart based on the question you want to answer:

Chart examples

The screenshots below use synthetic, illustrative data. They are examples of chart output, not actual business results or Replit plan metrics. Suggested prompts assume you provide the corresponding data.

Bar charts

Use a bar chart to compare amounts across categories or time periods. Each bar represents a value, such as total revenue for a month.
Using the attached sales data, create a bar chart of total revenue by month in USD. Identify the largest month-over-month increase.
Monthly revenue bar chart from January through December, with December showing the largest increase.

Bar chart of illustrative monthly revenue.

Line charts

Use a line chart to track change over time. Add separate lines to compare products, regions, or other groups using the same units.
Create a line chart of monthly revenue for each product line. Use one line per product and identify when one overtakes the other.
Revenue lines for fictional Studio and Canvas products, with Canvas passing Studio in August.

Line chart comparing two fictional product lines.

Stacked area charts

Use a stacked area chart to show both a total and its composition over time. Each band represents one part of the total.
Show monthly membership by plan as a stacked area chart. Use member counts, not percentages, so I can see the total grow.
Stack only values that can be added meaningfully. For example, use mutually exclusive membership categories rather than overlapping groups.
Stacked areas show monthly membership growing across three fictional plan categories from January to December.

Stacked area chart of fictional membership categories, not Replit plan data.

Donut charts

Use a donut chart to show how a small set of categories contributes to a total. Ask for counts and percentages to make the shares clear.
Create a donut chart showing orders by acquisition channel. Label each channel with its order count and percentage of all orders.
Donut chart showing Direct at 42 percent, Search at 31 percent, Social at 19 percent, and Partners at 8 percent.

Donut chart of 1,000 illustrative orders by channel.

Sankey diagrams

Use a Sankey diagram to show flows between stages. The width of each connection represents a quantity, such as the number of people taking a path.
Create a Sankey diagram from acquisition channel to selected plan to paid or trial outcome. Make each connection’s width represent the signup count.
Provide counts for the connections between stages, not just separate totals for each category.
Flows connect Search, Social, and Referral channels to Creator and Teams plans, then Paid or Trial outcomes; 225 signups become paid members.

Sankey diagram of an illustrative cohort of 350 signups.

Heatmaps

Use a heatmap to find patterns across two dimensions. Color intensity represents a measure, such as the number of sessions in each day-and-hour combination.
Create a heatmap of sessions by day of week and hour. State the timezone and highlight the busiest periods.
Heatmap of sessions by weekday and hour from 08:00 to 19:00, with the highest value of 132 on Wednesday at 13:00.

Heatmap of synthetic sessions during an illustrative week.

Provide data and refine the chart

Upload a dataset or connect a data source, then identify the records and date range to use. To avoid confusing sample data with your own records, include this instruction:
Use only the data I provided. If no dataset is attached or accessible, ask me for it instead of generating sample data.
You can continue in the same chat to change the chart type, grouping, labels, or comparison. For example:
  • “Show a separate line for each region.”
  • “Use weekly totals instead of daily totals.”
  • “Replace the donut chart with a bar chart so I can compare similar values.”
Check the chart’s units, aggregation, missing values, and date range against the source before relying on the result.

Charts in apps

These examples describe charts created in Chat. To build an app with charts, filters, and its own interface, follow Create a dashboard from data.