From Inbox to ERP: Automating Shipping Instructions in Logistics
TL;DR: Logistics companies receive shipping instructions via email in countless formats: booking requests, airway bills, CMR documents, bills of lading. Documind reads these documents, has three AI models check every value, and writes the result into your ERP or TMS. When the models disagree or a rule fails, it asks a person by email. The result: faster processing, fewer errors, and operations teams focused on exceptions rather than data entry.
Why Shipping Instructions Are So Hard to Process
Whether you operate in ocean freight, air cargo, or road transport, the challenge is the same: shipping instructions arrive in inconsistent formats from multiple sources.
Operations teams face a constant stream of documents:
- Ocean: Bills of lading requests, booking confirmations, dangerous goods declarations
- Air: Airway bill instructions, shipper’s letter of instructions, security declarations
- Road: CMR consignment notes, delivery orders, transport instructions
Each customer sends documents in their own format. Some use structured PDFs, others send scanned images or even photos. Email attachments vary: sometimes a single instruction, sometimes bundled with packing lists, certificates, or DG forms.
Add country-specific requirements (mandatory tax IDs for certain destinations, specific address formats, compliance fields that vary by trade lane), and manual processing becomes a bottleneck that doesn’t scale.
Where Documind Fits In
Documind handles the reading and checking: classifying documents, extracting data and validating it. It can write the result into your ERP or TMS, or hand it to your own automation (RPA, custom integrations or workflow tools) for the next steps.
Document Classification
When documents arrive, Documind identifies what you’re dealing with:
- Is this the main shipping instruction or a supporting document?
- Is there a dangerous goods declaration attached?
- Are there invalid attachment combinations that need to be flagged?
The classification decides where each document goes next, and rejections can be sent back to senders.
Structured Data Extraction
Documind reads each document and extracts the fields you need:
- Reference numbers: Booking number, B/L number, AWB number, CMR number
- Parties: Shipper, Consignee, Notify Party (including multiple notify parties)
- Routing: Origin, Destination, Transshipment points, Place of Receipt, Place of Delivery
- Cargo details: Container/unit numbers, package counts, gross weight, commodity codes, dimensions
- Special cargo: Dangerous goods classifications, temperature requirements, oversized cargo dimensions
Three AI models from different model families read each field. A value flows through automatically only when all three agree and the rules pass. Otherwise a person is asked by email.
Business Rule Validation
Before your automation pushes data to the ERP, Documind can validate against your business rules:
- Party identification: Is the shipper a freight forwarder? Does the email domain match the shipper name?
- Document type detection: Original B/L, Sea Waybill, or Express Release? Air Waybill or House AWB?
- Reference format validation: Do booking numbers follow your prefix conventions?
- Country-specific requirements: Are mandatory fields present for the destination country?
- Customer matching: Can the instruction be linked to an existing customer record?
The check results are recorded with each value, so you can see why something passed or was sent to a person.
Handling Complex Scenarios
Logistics documents come with edge cases. Documind handles them and surfaces the relevant data:
Dangerous Goods: UN numbers, proper shipping names, hazard classes, packing groups, and emergency contact information are extracted from DG declarations across all transport modes.
Temperature-Controlled Cargo: Whether it’s reefer containers for ocean freight or temperature-controlled air cargo, Documind captures temperature settings, ventilation requirements, and acceptable ranges.
Oversized Cargo: Out-of-gauge dimensions for ocean, heavy/oversized declarations for air, and exceptional load specifications for road transport are all captured for downstream processing.
Revisions: When Documind processes a document with the same reference number as a previous instruction, it flags this as a revision. Your automation can then update existing records rather than creating duplicates.
Exception Scenarios
Some scenarios should always involve human judgment. Documind identifies these and asks the right person by email:
- Reference numbers that don’t match between email metadata and document content
- Restricted cargo types for specific destinations
- Container or unit specifications that don’t match the booking
- Missing mandatory fields for the destination country
- Values on which the three AI models do not agree
You decide how to handle these: queue for review, send for approval, or escalate to specific teams.
The Outcome
By using Documind in your shipping instruction workflow:
- Processing time drops: Reading and checking take seconds instead of manual typing
- Errors decrease: Far fewer transcription mistakes reach your systems
- Throughput increases: Documents are processed as they arrive, not when staff is available
- Compliance improves: Country-specific rules are validated consistently
- Audit trails exist: Every step is recorded, with field-level source references
For logistics operations processing hundreds of instructions daily across ocean, air, and road shipments, this means significant capacity freed up for customer service and exception handling.
Ready to Map Your Shipping Instruction Process?
Map your process with us. In the demo, we go live with your first workflow, from the inbox to your ERP or TMS.
Muhammet Aksoy
Documind automates enterprise content workflows: it reads what reaches your teams, checks it, and finishes the work in your systems.