An air waybill OCR that knows freight: every MAWB field, pulled and typed.
AWBGuru is OCR and data extraction built for the Master Air Waybill. It turns the whole document — identity, routing, parties, cargo, and charges — into a structured record, straight from a PDF, a phone photo, or a scan. House air waybills go through the same way; see house air waybill extraction.
Every field returns with its own confidence score, so clean documents flow straight through and only the genuinely uncertain fields ever reach a person.
30 free scans to start · no card required · then pay per scan or pick a plan
A Master Air Waybill was never built to be read by a machine.
Which is exactly why generic OCR struggles on them — and why AWBGuru is built for this document specifically.
No two look alike
Every carrier and forwarder prints a different layout. There is no fixed grid to key off — the same field sits somewhere new on every waybill.
The inputs are messy
Skewed phone photos, faxed scans, stamps over the print, and handwriting in the margins are the norm, not the exception.
One file is rarely one document
Bulk PDFs bundle dozens of waybills; a single MAWB can consolidate many house shipments. It has to be split and classified before a field is read.
A wrong field is worse than a blank
A confidently-misread weight or AWB number flows silently into your systems. Extraction alone isn't enough — every value has to be scored.
Why OCR alone isn’t enough for air waybills: a misread can come back with high confidence, so the values also have to be checked against freight rules. See validation that catches what OCR confidence misses.
The full waybill, broken into fields.
One document in, one structured record out — grouped, typed, and ready to load. Every field carries its own confidence.
Identity
- AWB number + prefix
- Carrier
- Flight & date
- Document type
Routing
- Origin (IATA)
- Destination (IATA)
- Routing / transit points
Parties
- Shipper name & address
- Consignee name & address
- Agent / account refs
Cargo
- Pieces
- Gross & chargeable weight
- Volume / dimensions
- Nature of goods
- HS code
Charges
- Prepaid / collect
- Rate & rate class
- Weight / valuation charges
- Totals
Handling
- Special handling codes
- Consolidation flag
- Nature/quantity notes
Send them the way you already work.
There's nothing to install and no format to prep — pick whichever intake fits your operation.
Drag & drop
Drop a PDF, image, or scan straight into the web app.
Email intake
Forward waybills to a dedicated address; they extract on arrival.
Mobile capture
Snap a photo on the ramp — deskew and cleanup happen automatically.
API
POST documents and poll for the structured result — no middleware.
You read the exceptions, not the documents.
Extraction doesn't hand you a wall of text to double-check. Each field is scored on its own, you set one cutoff, and anything above it clears without a click — so the only fields a person touches are the genuinely uncertain ones. There's no model to train on your documents first.
Key values — AWB check digit, airline prefix vs carrier, airports and routing, rate class, chargeable vs gross weight — are also checked against air-freight rules before they clear. See air waybill validation for the checks.
MAWB extraction, answered.
Does it read handwriting and phone photos?
Yes. Handwritten margins, stamps, skewed phone photos, and faxed scans all run through the same pipeline — deskew and cleanup are automatic, so you don't pre-process anything.
How does it handle a consolidation or several waybills in one PDF?
A classify-and-split stage runs first: a bulk PDF is separated into individual documents, and a MAWB that consolidates house shipments is detected as a consolidation before any single-commodity read is attempted — so a consol manifest is never force-fit into one line. House air waybills in the same file are recognized as houses and read against their own field set, not forced into the master schema.
Do I have to map fields to my own system?
No per-carrier templates and no field mapping to configure. Results come back as one consistent record and you consume the format your systems already speak — JSON, FWB 16/17 EDI, or CSV.
How fast is it, and how accurate?
A typical waybill extracts in tens of seconds. Accuracy is reported per field as a confidence score rather than one headline number, so you decide the cutoff at which a field clears automatically versus routes to a quick review.
Extract a real waybill in the next five minutes.
Start with 30 free scans — no card. Drop in one of your own MAWBs and see the record come back.