Choose a source
In the trusted-local owner workflow, begin with a bounded CSV, XLSX, or supported Parquet file. No file can be uploaded on this public site. External connections are part of the intended future experience.
Public showcase · Registration and public uploads unavailable · AI demo not enabled.
Five transparent steps take you from a dataset and configured rules to traceable findings and explanations you can use.
Explore the process ↓Each step explains what the platform is intended to do without hiding assumptions behind an opaque score.
In the trusted-local owner workflow, begin with a bounded CSV, XLSX, or supported Parquet file. No file can be uploaded on this public site. External connections are part of the intended future experience.
Confirm the table structure and define the conditions you actually want checked. Users should know what a rule means before authorizing analysis.
The local processing foundation can produce deterministic findings and source-linked evidence. The interface should report progress honestly and distinguish completed work from assessed coverage.
Select a candidate, jump to its supported location, and review nearby records. On-demand data windows are a design objective for large datasets and require safe, authorized retrieval.
Start with an understandable summary of what was observed and not assessed. Access the provenance and technical evidence when you need them.
Explore a fictional finding in the interactive Dataset Explorer preview.