Lane Compliance SystemsValidator

Accuracy Validator

Lane Compliance Systems, a DBA of Lane Global Holdings LLC · runs entirely in your browser

Measure how the Lane Compliance Systems review engine performs on your own labeled reports — recall, false-alarm rate, precision, accuracy, and a confusion matrix, per workspace. This page runs the same deterministic engine that powers the product.

This page makes no network requests. You can disconnect from the internet and it will still work. Your report text is processed only in this browser tab and is never uploaded or stored anywhere — the same on-device guarantee as the product.

1 · Load your labeled reports

Paste CSV or JSON, or choose a file. Columns/keys (case-insensitive): case_id, expected_outcome (returned | approved), report_text, and optionally expected_clauses (rule refs that should fire, e.g. LE-1.1;LE-8.2) and notes.

Advanced: paste a custom policy config (JSON) — overrides the template above

Validate against the exact configured policy you would deploy. Leave blank to use the selected workspace template.

Illustrative impact model (editable assumptions — affects only the optional $ estimate)
How to read this. “Predicted” is the engine’s triage: a report is predicted flagged when its rejection risk is high or medium, otherwise cleared — a recommendation, not a final human decision. Recall = share of truly-returnable reports the engine flagged. False-alarm rate = share of approvable reports it flagged. Metrics are only as sound as your labels; reconcile reviewer disagreements before trusting them. This measures how the selected policy behaves on your reports — pick the workspace (or paste the config) you would actually deploy. The dollar figure is an editable illustrative model, not a guarantee.
Lane Compliance Systems, a DBA of Lane Global Holdings LLC · Privacy · Terms · Security · security.txt
This validator runs the same deterministic engine that powers the product. Not legal advice.