Analyze a dataset

Upload a spreadsheet to Vibe›Work ↗ and ask questions about it in plain language. Work picks the right tools (file reader, built-in code sandbox, Canvas) to compute summaries, generate charts, and run follow-up analyses. You write no code.

Time to complete: ~10 minutes

• A Mistral account (Free plan is enough to try, paid plans recommended for heavier datasets).
• A data file to analyze (CSV, XLSX, or JSON).

1. Open Vibe›chat.mistral.ai ↗ and switch to the Work tab.
2. In the chat toolbar, click the + icon, then select Upload Files and pick your data file (for example sales-q4-2025.csv).
3. Work shows a preview of the file. Confirm it looks correct.

For best results, use files with clear column headers. Work reads them to understand the schema before running anything.

Type a question in plain language. Work writes and runs code in a sandbox and returns tables, charts, and explanations directly in Canvas, a side panel that opens automatically when Work needs to display a document, a chart, or a long output you can edit. You can also ask for it explicitly with phrases like "open this in Canvas" or "visualize this file".

Try one of these:

Summarize this dataset. How many rows and columns are there? What are the key statistics per column?

Show me monthly revenue trends as a line chart.

What are the top 5 products by total sales? Display as a bar chart.

Calculate the correlation between marketing spend and revenue, and explain the result.

Work shows the code it ran alongside the result, so you can verify the logic.

Build on previous results by asking follow-up questions. Work keeps the data in context across the conversation.

1. Refine a chart: "Make the chart wider and add data labels."
2. Filter data: "Show only rows where region is Europe."
3. Compare periods: "Compare Q3 vs Q4 revenue by product category."

To save your work, use the Export button in Canvas to download tables as CSV/Excel or charts as image files. You can also copy tables directly from the Canvas view, or ask Work to "open this in Canvas" to turn the analysis into an editable document.

Your analysis is working if:

• Work correctly identifies column names and data types.
• Generated charts match the data in your file.
• Follow-up questions reference the same dataset without re-upload.
• Statistical results (mean, median, correlation) match what you expect.

If a column is misinterpreted, clarify it explicitly: "The 'Date' column is in DD/MM/YYYY format."