A real AI use, explained simply

Turn a CSV into one chart and a decision-check note

Answer one narrow data question with one chart, explicit assumptions, and rows to verify.

Start here

Use a de-identified copy, not production data, and write one question.

Done when: The chart has a defined metric and unit, assumptions are listed, and source rows or calculations are checked.

Before you start

  • A de-identified CSV copy
  • One metric and time range
Visible evidence boundary

What the original documents

OpenAI officially documents file analysis and chart creation; PandasAI publishes natural-language exploration of CSVs and databases.

Claim boundary: The sources document analysis capabilities, not the correctness of calculations, assumptions, or decisions.

This page explains a source-documented method. It is not an independent Studio Nani reproduction or performance guarantee.

Follow the workflow and check each result

  1. Inspect column names, units, and date range.Output: A data noteCheckpoint: Identify missing or duplicate rows.
  2. Define one metric and calculation assumptions.Output: A scoped analysis requestCheckpoint: Do not mix units or periods.
  3. Request one chart, three observations, assumptions, anomalies, and rows to check.Output: Chart and decision-check noteCheckpoint: Keep generated calculations visible.
  4. Recalculate key values and inspect cited rows before deciding.Output: Reviewed analysis draftCheckpoint: Do not apply generated transformations to production.

Worked output shape — no made-up values

One chart plus three observations, assumptions, anomalies, and source rows to check; no sample values are shown.

  • chartOne chart
  • observationsObservations
  • assumptionsAssumptions
  • anomaliesAnomalies
  • rowsRows to verify

Human completion checks

  • Units, date range, exclusions
  • Key calculations and source rows
  • Confidentiality policy

Limitations

  • Generated code, assumptions, charts, and units can be wrong.
  • Never apply transformations to production without review.

Use this when

The chart has a defined metric and unit, assumptions are listed, and source rows or calculations are checked.

Copy this and change the brackets

From this CSV, calculate monthly [metric] for [period] and make one line chart. List excluded rows, assumptions, and rows I should verify.
How to make it work for you

Source-documented method

The sources document analysis capabilities, not the correctness of calculations, assumptions, or decisions.

Keep in mind

Generated code, assumptions, charts, and units can be wrong. Never apply transformations to production without review.

Next, make it your own

Source and notes

checked 2026-08-06

Open original source ↗