Delivery note and proof of delivery capture
Tavnit reads delivery notes and proofs of delivery, including the handwritten annotations drivers and receivers add, and returns delivered quantities as structured rows. Because purchase orders are extracted into the same shape, checking delivered against ordered becomes a query.
Why this is painful
The delivery note is the only record of what physically arrived, and it is usually the worst-quality document in the chain: photographed in a yard, signed over the print, annotated by hand where quantities differed.
It is also the document that matters most for disputes. If the invoice says twelve and the note says ten, the note is the evidence — assuming anyone captured what it said.
What to extract
| Field | Why it needs care |
|---|---|
| Delivery note and PO number | The keys that tie the delivery back to what was ordered. |
| Delivered quantity per line | The number that actually matters, and the one most often amended by hand. |
| Handwritten amendments | Crossed-out quantities and margin notes carry the real figure. Ignoring them means capturing the wrong one. |
| Receiver name and signature present | Presence of a signature is frequently the acceptance test. |
| Delivery date and time | Drives SLA measurement and dispute timelines. |
| Condition or damage notes | Free text, usually handwritten, and the basis of any claim. |
What makes delivery notes hard
The handwriting is the important part
A printed delivery note says what was supposed to arrive. The handwritten amendment says what did. Extraction reads handwriting as well as print, which is the difference between capturing the shipment as planned and capturing the shipment as delivered.
Photographed in a yard, not scanned in an office
PODs arrive as phone photos at an angle, in poor light, sometimes with a thumb in frame. Extraction copes with most of it, but this is a document type where routing low-confidence results to a person is worth doing rather than trusting every capture.
The value is in the comparison, not the capture
A delivered quantity on its own tells you little. Extracting delivery notes into the same field structure as your purchase orders turns three-way matching into a query on the PO number rather than three documents held up to the light.
How the pipeline handles it
- Email Integration — Drivers and carriers email photographed PODs; forwarding them in processes each on arrival.
- Flows — Delivered lines are table fields, matching the structure used for purchase orders.
- Buckets — Holds deliveries and orders in one place, so discrepancies surface as a query rather than an audit.
- Human in the Loop — Reviews the poor captures and the amended quantities, where the cost of being wrong is a dispute.
Common questions
- Can it read handwritten amendments?
- Yes. Handwriting is extracted alongside printed text, which matters here because the handwritten quantity is usually the accurate one.
- Will it work with photos taken on a phone?
- Usually. Photos, angles and poor lighting are the norm for PODs. Conditional review is worth enabling so the marginal captures reach a person instead of passing silently.
- Can deliveries be checked against orders?
- Extract both into the same field structure and the comparison is a query on the PO number in your Bucket, rather than a manual document-to-document check.
Other document types
Try it on one document
Build a flow, send a real delivery note through it, and see what comes back. Free credits to start.
Start free