> ## Documentation Index
> Fetch the complete documentation index at: https://docs.replit.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Create charts and explore data

> Visualize uploaded or connected data in Chat, ask follow-up questions, and check your findings.

Ask Replit a question about your data to create interactive charts directly in a chat. You can start with an uploaded dataset or a connected data source.

Interactive charts in Chat are powered by Atta. You do not need to choose a chart type before asking a question.

## Create your first chart

<Steps>
  <Step title="Add your data">
    [Upload a file](/chat/upload-and-use-files) or [connect a tool](/chat/integrate-your-tools) that contains the data you want to explore.

    Identify the file, sheet, or records to use. Include the date range and any categories you want to compare.
  </Step>

  <Step title="Ask a question">
    Describe what you want to understand. For example:

    > Using only the attached sales spreadsheet, chart total revenue by month. Explain which months had the largest changes. If no file is attached, ask me to upload it instead of generating sample data.

    Replit selects a visualization suited to your question. You can also ask for a specific chart, such as a line chart.
  </Step>

  <Step title="Refine the result">
    Continue in the same chat to change the comparison or investigate a finding:

    > Break this down by region, using the same date range.

    > Show monthly revenue rather than cumulative revenue.

    > Which region contributed most to the decline?
  </Step>
</Steps>

## Choose a chart for your question

See [Charts](/features/agent/charts) for a visual guide to bar, line, stacked area, donut, Sankey, and heatmap charts.

Start with the comparison you need. These common chart types suit different questions:

| Question | Chart type | Example request |
| - | - | - |
| How did a measure change over time? | Line chart | “Show weekly signups over the last three months.” |
| How do categories compare? | Bar chart | “Compare total revenue by region for the last quarter.” |
| What share does each category contribute? | Donut chart | “Show orders by acquisition channel as a share of all orders.” |
| When does activity peak? | Heatmap | “Show sessions by day of week and hour.” |

Specify units, time periods, and grouping when they matter. For example, distinguish total revenue from average revenue per customer.

## Try a small example

Save this sample as `monthly-sales.csv`, then upload it to a chat:

```csv theme={null}
month,region,revenue_usd
2026-01,North,12000
2026-01,South,8000
2026-02,North,15000
2026-02,South,9000
2026-03,North,13000
2026-03,South,11000
```

Ask:

> Create a line chart of total monthly revenue in USD from this file. Sum revenue across regions for each month.

Check the result against the file: January totals $20,000, February totals $24,000, and March totals \$24,000.

Then ask:

> Show a separate line for each region. Explain why the total stayed the same from February to March.

North decreased by $2,000 while South increased by $2,000. The regional view explains a change that the total alone does not show.

## Check your findings

Before sharing or relying on a chart:

* Confirm the source data and date range.
* Check whether values are totals, averages, percentages, or cumulative amounts.
* Verify units, axis labels, and category names.
* Ask how missing values and duplicate records were handled.
* Compare a few chart values with the original records.

A relationship between two measures does not establish that one caused the other.

## Charts in Chat and charts in apps

Use charts in Chat to investigate a question and discuss the results. To build an app with charts, filters, and a persistent interface, follow [Create a dashboard from data](/build/dashboard).

The dashboard guide describes a separate app-building workflow. Creating a chart in a chat does not publish a dashboard.

## Next steps

* [Upload and use files](/chat/upload-and-use-files) to add a dataset.
* [Integrate your tools](/chat/integrate-your-tools) to work with connected data.
* [Create a dashboard from data](/build/dashboard) to build an app around your findings.