Automated document reading in freight means extracting the useful information from an incoming document (a quote, a booking confirmation, an invoice) and linking it to the right shipment. It combines character recognition, which turns an image into text, with interpretation of the content: what kind of document is this, which shipment does it belong to, what amounts and dates does it contain. The aim is not to automate everything but to prepare the work, so the team can focus on the cases that need a decision.
Why it matters in freight
An international shipment generates many documents from different parties: forwarder, ocean or air carrier, trucker, customs broker. They often arrive by email as attachments, each in its issuer’s own layout. Before anyone can use them, someone has to open them, identify the shipment, copy the references and file the document.
A received document is not yet usable data. Automated reading saves time on exactly that sorting, keying and linking work.
How automated document reading works
1. Recognise the document
Depending on the file, extraction uses text already embedded in it (a PDF generated by software) or OCR (Optical Character Recognition) for a scan or a photo. The tool then identifies the document type: quote, booking, invoice, supporting document.
2. Extract the relevant fields
Reading an amount is not enough: its currency, its nature (freight, surcharge, ancillary charge) and the matching shipment reference matter too. Which fields to extract depends on what each document type will be used for.
3. Link it to the right shipment
Information is only useful once attached to the shipment it concerns: booking number, bill of lading, air waybill or internal reference. This is often the hardest step, because one shipment carries different references depending on who issued the document.
4. Handle exceptions
A shipment that cannot be found, an ambiguous document or an unexpected charge has to reach a person. Without a named owner for these cases, automation moves the problem instead of solving it.
What the tool does, what the team decides
| Step | Prepared by the tool | Decided by the team |
|---|---|---|
| Intake | Identify the document and its issuer | Choose which documents and flows to process |
| Extraction | Read supported references, dates, lines and amounts | Set the expected fields and check doubtful cases |
| Linking | Suggest or make the link to the shipment | Resolve documents with no match or an ambiguous one |
| Invoice audit | Compare lines with the quote or rate card | Accept a justified difference or request a correction |
In OneChain: Zach
In OneChain, Zach works on emails and documents within the workflows configured with your team. It recognises documents, finds references and extracts supported information; depending on the workflow, documents are attached and shipment fields are updated automatically. Recognised document types and updated fields depend on that workflow. An unrecognised shipment, an ambiguous document or a charge needing explanation stays with your team, which also sets the rules for future documents. For the broader picture, see artificial intelligence in OneChain.
Example: preparing a freight invoice audit
Freight invoice auditing means checking that what is billed matches what was agreed. An invoice can add up correctly and still contain a line that needs explaining; conversely, a difference from the quote can reflect a service that was genuinely added.
- Identify the invoice and the shipment references.
- Extract the items covered by the workflow: charge, amount, currency.
- Link the information to the shipment file and its documents.
- Review differences against the quote or rate card in invoice control.
Extraction prepares the review; it does not establish that every field is correct or that the invoice is approved. A flagged difference guides the review, but on its own it does not prove a billing error. The review ends with a documented decision: accept a justified explanation or prepare a correction request. For more, see our invoice audit module.
What about e-invoicing?
In France, all businesses must be able to receive electronic invoices from 1 September 2026. Large and mid-sized companies must also issue them from that date, and small and micro businesses from 1 September 2027. The reform covers purchases and sales between businesses established in France and subject to French VAT; once it applies, a plain PDF sent by email is no longer a compliant invoice within that scope (French tax administration).
A structured e-invoice already carries machine-readable data, which reduces the need for OCR. It still does not say which shipment it belongs to, or whether the amount matches the agreed quote, so linking and auditing remain necessary. Invoices from providers established outside France are not covered by this obligation: those transactions go through a separate data report to the tax authority (e-reporting), without a structured e-invoice.
Measure value on a real flow
Pick a frequent document type. Track processing time, corrections needed, duplicates and correctly linked documents. For invoice audits, separate what was reviewed, the confirmed differences and the corrections actually obtained. Use the observed results to decide which documents to tackle next.
Frequently asked questions
What is automated document reading?
It is the automatic extraction of information from a document (PDF, scan, attachment) so it can be used in software. It combines character recognition, document type identification and linking to the right record.
What is the difference between OCR and AI document processing?
OCR turns an image into text. AI document processing goes further: it works out what the document is, which fields matter and which shipment they relate to.
How do you audit a freight invoice?
Find the shipment and the agreed quote or applicable rate card, compare each billed line, then review differences against the shipment documents. Then decide whether to accept the difference or request a correction.
Does automated reading replace human approval?
No. It prepares the data and flags differences; exceptions and business decisions, such as approving or disputing an invoice, stay with the team.
Does e-invoicing make OCR unnecessary?
Not entirely. Structured invoices reduce keying, but they still need linking to the shipment and comparing with the quote. Other freight documents, and invoices outside the scope of e-invoicing rules, can keep arriving in other formats.