On-device tools
Analyze CSV Files with AI on iPhone and Mac
Attach a CSV, TSV, JSON, or text file and ask Cumbersome to analyze it. The selected model writes an analysis plan, then Python runs on your device and returns computed results, tables, and charts to the conversation.
Let the model plan while your device does the math
Language models are good at choosing an analysis and explaining a result. They are bad substitutes for deterministic calculation. Cumbersome separates those jobs. The model prepares Python, the app executes it locally, and the model uses the actual output instead of inventing totals from a long table.
You can ask for missing-value rules, calculated columns, grouped summaries, best and worst performers, and charts. The local tool reports when analysis ran and when it completed.
- 1Enable Local Data Analysis in Settings under On-Device Processing.
- 2Attach a CSV, TSV, JSON, or text file to a conversation.
- 3State the formulas and cleanup assumptions that matter.
- 4Ask for tables and charts, then inspect the reported calculations.
Useful for small, practical datasets
This is built for the files people actually receive in email or export from another app: sales records, survey results, logs, inventories, and similar working data. It avoids uploading a file to a separate analysis service just to answer a few questions.
What to know before you use it
- The selected AI provider still receives the prompt and the context needed to prepare and explain the analysis.
- Generated code can contain mistakes. Verify important formulas and totals against the source file.
- Very large datasets can exceed device memory, model context, or a provider request limit.