Drop in a report. Get the dates back.
Everything your register needs is buried in the PDF. Paco pulls out the date, the contractor, the reference and the interval, and checks every value before writing it down.
The PDF picks the route, not you
A report that is already digital carries its own text: Paco reads it directly, with no OCR and no cost. A scanned document goes through character recognition. An unreadable file goes to human review rather than producing an invented line.
- Native text is read as-is, never sent through OCR
- Scans go through dedicated character recognition
- A file that is neither readable nor scannable is flagged, never guessed
Every value has to exist in the document
An extracted date is compared against the source text. If the string does not appear there literally, it is rejected. That rule is what separates reading from guessing, and it applies to the date, the contractor, the reference and the interval.
- A value absent from the source text never reaches the register
- A letter's own date cannot pass itself off as an intervention date
- Every date found is kept for the reviewer to see
Common questions
What if the report is badly scanned?
It goes to the review queue with the exact reason. You see the document, what could be read, and you decide. No obligation is created from a doubtful reading.
Are my documents used to train a model?
No. Documents are processed to extract your data and nothing else. The processors involved are listed in the privacy policy.
See it on your own reports
Bring three inspection PDFs. We will show you what Paco pulls out of them, and what it refuses to.
Book a demo