Skip to main content
Utilnivo

AI & Writing

Utilnivo

AI CSV Column Matcher

Map columns between CSV schemas with explainable, privacy-first suggestions.

No registration
  • AI-powered
  • Stays on your device
  • Beta

Browse more in AI & Writing or all tools.

Private on your device

Your information stays on your device and is not uploaded.

Upload or paste source and target column schemas, inspect inferred types and patterns, review intelligent mapping suggestions with evidence, resolve conflicts, and export a verified mapping definition. Small CSV files are profiled in your browser—raw rows are not uploaded in local mode. Suggestions are not guaranteed; always review before transforming data.

How to use this tool

1. Add a source CSV or paste source column names. 2. Add a target CSV, paste target columns, or define them manually. 3. Inspect inferred types and correct header or delimiter settings if needed. 4. Review suggested mappings, confirm or change them, and mark intentional no-matches. 5. Validate and export the mapping JSON or CSV.

Worked example

Source Cust_ID, Cust_Name, Primary_Email maps to Customer Number, Customer Full Name, Email Address with high or medium confidence and readable evidence for each suggestion.

How it works

Upload or paste source and target schemas. The tool profiles CSV columns locally in a Web Worker, then suggests mappings using alias, lexical, datatype, pattern, and optional semantic signals. You review evidence, resolve conflicts, and export a verified mapping. Raw CSV rows stay in the browser in local mode.

Limitations

Suggestions are assistive and require human review. Local mode supports CSV only (5 MB, 10k rows, 200 columns). Cloud Excel processing is feature-flagged and disabled by default. Do not rely on high confidence as certainty.

Privacy and file handling

Your data stays on your device and is not uploaded.

FAQ

Frequently asked questions

Does my CSV leave the browser?

In local mode, raw CSV content stays in browser memory. Only privacy-safe column descriptors (names, types, pattern labels, ratios) are sent to the matching API for suggestions.

Are mapping suggestions guaranteed correct?

No. Every suggestion includes evidence and confidence, but you must review and confirm mappings before transforming critical data.

Can a source column intentionally have no match?

Yes. The matcher supports unmatched sources and unused targets. It does not force every column to map.

Page last reviewed: