Reading documents
How attachments become text the rest of the product can work from.
What this process achieves
Turns attached datasheets, specification sheets, bills of quantity and images into text — keeping table structure where it matters — so line items and parameters can be read from them.
Who is involved
- The system — converts, including character recognition for scanned pages.
- Sales person — retries a document that failed.
What starts it
Attachments arriving on a conversation. Reading runs before line-item split on a processing run, and can be retried per document.
The steps
- Each attachment is queued when the message is stored. Nested files inside a forward are lifted out as attachments in their own right.
- Tables are read as tables, so a multi-instrument annexure is not flattened into prose.
- Scanned or image-only pages go through character recognition. The lowest page confidence is recorded (the lowest, not the average).
- A conversion that fails is marked failed with the reason. It is skipped from then on unless someone retries that one document.
- Beyond a total text cap, whole documents are dropped from the end and the count is recorded. Documents are never cut mid-way.
Rules that apply
R22, R49, R50.
What happens if it stalls
| Situation | What the business sees |
|---|---|
| Recognition skipped | The document reports that distinctly, rather than as blank |
| One page is noise | Low confidence is flagged; the text is kept |
| Retry | Failed → waiting, then read or failed again |
Related: Line items · Processing a case.