100 cartons → 120 cartons
The reviewer receives the old value, proposed value and source page together. In this example, the package-count change holds the revision until approval.
AI Operators · Use cases
Automate logistics document processing and TMS data entry from emails and PDFs. Extract shipment fields, validate documents and write approved data to your TMS.
Discuss document automationYour SOPs. Your systems. Your team in control.
The operational opportunity
Logistics document processing connects incoming emails and PDFs to accurate operational records. It includes document classification, shipment-data extraction, validation and automated TMS data entry—not just OCR or comparing two files.
Shipflow’s Document Operator classifies files, extracts shipment data, checks it against customer rules and compares related documents when needed. It enters validated data into your TMS and sends unclear or conflicting information to your team with the source documents attached.
Use the same approach across Ocean, Air and Road: identify the shipment, validate the source data and complete the system entry. Document types, field mappings and review rules are tailored to each operation.
Your team defines the source of truth, destination fields and approval rules once, before launch. The operator applies them to every document.
Document to TMS · Worked example
A revised Shipping Instruction arrives by email. The AI Operator matches the existing shipment, checks each field and prepares the TMS update. A changed package count needs review before this revision can proceed.
Illustrative workflow using fictional shipment data.
“Please use the attached revision: 120 cartons, gross weight 2.4 metric tonnes. The route remains Shanghai to Rotterdam.”
Scroll across to compare all columns →
| TMS destination | PDF · page 1 | Current TMS value | Proposed action |
|---|---|---|---|
| Shipment reference | SF-DEMO-1042 | SF-DEMO-1042 | Match existing shipment |
| Port of loading | Shanghai | Shanghai | Keep existing value |
| Port of discharge | Rotterdam | Rotterdam | Keep existing value |
| Package count | 120 cartons | 100 cartons | Review 100 → 120 |
| Gross weight | 2.4 metric tonnes | Not entered | Normalize to 2,400 kg |
The destination labels are illustrative. Field names, units and source-priority rules are mapped to your TMS during setup.
The reviewer receives the old value, proposed value and source page together. In this example, the package-count change holds the revision until approval.
Recheck the record, apply the approved revision and verify the saved values. Retain the source version and review decision with the task.
Complete when: the correct shipment contains the approved values and the system update is verified. Carrier submission is a separate workflow.
See how Shipflow reads and writes CargoWise data through eAdaptorWorkflow coverage
Configure the Document Operator around the job each document needs—not a single extraction template applied to every file.
Collect documents from agreed channels, identify the document type and shipment, read the relevant content and manage revised versions.
Check required fields against related documents, TMS records and customer rules. Surface only the differences or omissions that require action.
Enter approved data into the shipment record and move the workflow to its next step.
Representative document types
How it works
Bring files from the agreed inbox, portal or workflow into one process. Classify the document type, match it to the shipment and distinguish the current version from superseded or duplicate copies.
Read the fields the operational workflow needs, such as parties, references, routing, cargo, equipment, dates and document-specific values. Normalize the information to the formats expected by your TMS and SOP.
Check fields, formats and business rules against shipment records. Treat a document as ready only after it passes validation.
Apply relationship-aware checks across documents such as MBLs and HBLs, Shipping Orders and Shipping Instructions, or commercial invoices and packing lists. Define which values should correspond, which can legitimately differ and which discrepancies require action.
Write validated values to the TMS. If information is missing, unclear or conflicting, send the source file and exact fields needing review to the responsible person.
Human control
Your team sets the SOPs, system access and approval rules before launch. AI Operators act within those limits and escalate cases that need a decision.
Customer result
Asia Shipping uses Shipflow AI Operators to automate document comparison, including MBL and HBL checks. The deployment reduces time spent on this task by 97%+. This is evidence for the comparison task, not a measured result for TMS data entry or all document processing.
This result relates to the customer workflow described. Results depend on workflow scope, input quality and deployment configuration.
“Prior to Shipflow, our operators had to manually compare data fields between MBL and HBL three to four times per shipment, which was extremely time-consuming. With Shipflow, this process is now fully automated by AI.”
Getting started
Bring representative documents, including scans, attachments, revisions and difficult examples. Define whether the first workflow is processing, comparison, TMS data entry or a connected sequence, and agree what completed work means.
Specify the data required from each document, how documents connect to shipments and one another, and which source takes precedence when values conflict. Identify the exact destination fields in the TMS.
Define the checks, exception routes and approvals that apply before information is entered or an external action is taken. Confirm the supported input channels, file formats, system access and write-back permissions for the deployment.
Compare the same document types and validation rules before and after deployment. Report the sample size, time period, exclusions and review rate alongside the result.
These are measurement definitions, not promised results.
Common questions
Optical character recognition (OCR) converts an image into readable text. Extraction identifies useful fields. Shipflow connects those steps with validation, cross-document checks and actions in the operational workflow, subject to your rules and system access.
Yes. MBL/HBL comparison is part of the document-comparison deployment at Asia Shipping. The fields, document relationships and exception rules for another implementation are configured around that operation’s requirements.
Yes. Shipflow reads documents, checks extracted fields against shipment context and writes approved values into the connected TMS. For CargoWise, the workflow can read and write through eAdaptor. Your team defines the document types, field mappings, access and approval rules before deployment.
Yes. An email and PDF can start a document-processing workflow that matches the shipment, extracts the required fields, validates them and updates the TMS. Test the actual file formats and layouts used by your team, including scans, revisions and incomplete documents. Unreadable or conflicting information goes to review instead of being treated as valid data.
Workflows can be configured around documents such as MBLs, HBLs, Shipping Orders, Shipping Instructions, pre-alerts, commercial invoices, packing lists and arrival notices. Supported file formats, fields and actions are confirmed for the selected workflow and systems.
No. Different document roles can lead to legitimate differences. During setup, your team defines which fields should correspond, which differences are expected and which require investigation. The operator applies those rules rather than treating every difference as an error.
The workflow matches the file to the existing shipment, checks its version and compares it with the current record. A revised value follows your review rules. Duplicate files should not create a second shipment or repeat an update; an uncertain system response is reconciled before retrying.
No. Start with one repeatable document workflow and a clear validation standard. Expand to additional types after the team has reviewed the results and agreed the relevant rules, exceptions and system connections.
Start with one workflow
Start with a free 15-minute AI consultation. Describe the document your team rekeys most often, the destination system and the checks it needs. We will map a first workflow around those requirements.
Discuss document automation