Spot unusual records
Review statistical outliers using interquartile-range fences and see how many values deserve a closer look.
Understand your spreadsheets, uncover outliers, explore trends and create presentation-ready charts. Your data stays on your device.
Import, inspect, visualize and export in one place.
Explain already-calculated statistics in plain language. Your full CSV is never passed to the AI model; only a bounded summary of findings and column profiles.
Load or paste CSV, then optionally load a model. Standard insights and exports never need AI.
AI text is an unverified interpretation, not an additional statistical calculation, fact check, causal conclusion or prediction. A limited selection of report facts is sent to the model on this device; model scripts and weights may download from external CDNs after consent.
Each insight includes its calculation or scope. Statistical patterns are not explanations of cause.
Try a sample or import a CSV to reveal trends, missing values, outliers and correlations.
Choose a dataset to see a visualization.
| Column | Sample |
|---|---|
| Upload a dataset to preview it. | |
| Column | Detected type | Missing | Distinct | Summary |
|---|---|---|---|---|
| Column diagnostics will appear here. | ||||
Numeric and date types are inferred. Mixed or ambiguous columns may be classified differently than you expect. IQR flags indicate candidates to inspect, not definite data errors.
Review statistical outliers using interquartile-range fences and see how many values deserve a closer look.
Explore distributions, dominant categories and correlations without shipping spreadsheets to a server.
Group recognized ISO dates into daily averages, totals or counts and visualize observed patterns.
No. Parsing and statistics run in your browser. Core parsing and statistics use deterministic browser calculations. Optional AI explanations use a separately loaded local language model, and do not upload the full CSV to an inference server.
CSV and TSV text with a header row. Quoted values, embedded commas and newlines, UTF-8, tab/semicolon/pipe delimiters and missing cells are supported. Excel XLSX files are not directly supported.
No. These are exploratory summaries only, without causal or predictive conclusions. Verify data definitions, completeness and sampling before decisions.
Analyzing your dataset locally.
Import, paste or sample a supported table and review the detected column types. Inspect cited findings and flagged data rows, choose chart axes and export the required chart or report. Keep the original data alongside your review.
Core profiling is deterministic; optional local AI explanations are unverified narrative, not causal/predictive analysis. IQR flags are review candidates, not definite errors. Percentages use percentage points; ambiguous dates and mixed units need review. Exported data-row indices are not physical CSV line numbers. Chart reduction/group limits and resource budgets apply. Formula-like CSV text is prefixed for spreadsheet safety; XLSX import is unsupported.
Import, paste or sample a supported delimited table, review detected column types, and inspect deterministic missingness, duplicate, numeric, trend and correlation findings. You can examine flagged rows, choose chart axes and export charts, data or a written report.
Authored example: Profile a table of monthly delivery durations by route. Inspect unusually long rows before flagging them and compare trends only after confirming the date and duration units are consistent.
Limits and model handling: Core parsing and statistics are deterministic; optional AI explanations use a separately loaded local language model and are narrative suggestions, not causal or predictive conclusions. The source table is not sent to an inference server by that local explanation path, though runtime/model assets require external downloads. Percentages use percentage points; exported data-row indices are not physical CSV line numbers. XLSX import is unsupported, and formula-like CSV text is prefixed for spreadsheet safety.
No. They are exploratory findings and review candidates. Check the source, sampling, units and data definitions before making a decision.
No. This workspace accepts supported CSV/TSV text tables. Export a suitable table from your spreadsheet first.