From unstructured documents to data you can trust.
Real demo recording · English narration and captions · Fictional data
THE PROBLEM
Manually copying invoice fields and checking amounts takes time and makes discrepancies easy to miss.
THE WORKFLOW
Upload a fictional invoice, extract its fields, review a discrepancy, correct the amount and export the result as JSON, CSV or PDF.
01Document intake
02Structured extraction
03Review and validation
04Result export
RECORDED EVIDENCE
Results, with context.
Results on synthetic data
Check
Cases
All fields correct
32/32
Amount validation
32/32
Date validation
32/32
Dataset
synthetic-regression-v1
Completed cases
32/32
Median / p95 per case
2.25 s / 6.329 s
Evaluation record (UTC)
2026-10-08
Source: the project’s evaluation record, using local inference. Recorded hardware: NVIDIA GeForce RTX 5060 Ti, 595.84, 16311 MiB. Latency includes the full HTTP case and may include initial model loading. These measurements have not been rerun on the machine serving this website.
Evaluated demo