How a Thai durian exporter cut AWB processing time from 3 hours to 12 minutes

Siam Premium Fruits (name changed for confidentiality) is a mid-sized Thai fruit exporter based in Chanthaburi province, the heart of Thailand's durian-growing region. During peak season from April to July, they ship approximately 300 consignments of fresh durian by air each month, primarily to China, Hong Kong, Japan, and Singapore. Each shipment requires an Air Waybill, export customs declaration, phytosanitary certificate cross-referencing, and temperature-controlled logistics coordination.

Before deploying KabyTech, their 3-person operations team spent an average of 3 hours per day — every day during peak season — just processing AWB data. After one month with KabyTech's AWB Intelligence API integrated into their workflow, that dropped to 12 minutes. This is their story.

The challenge: perishable cargo waits for no one

Fresh durian has a commercial shelf life of about 5-7 days from harvest. By the time the fruit is sorted, graded, packed, and trucked to Suvarnabhumi Airport, the export window is already tight. A delay of even 6-8 hours at the airport — due to a documentation error, a customs hold, or a missed flight booking — can mean the difference between premium-grade fruit arriving in Shanghai and a rejected shipment rotting on the tarmac.

Khun Somchai (name changed), the operations manager at Siam Premium Fruits, described the problem: "During peak season we process 12-15 shipments per day. Each AWB has to be entered into our system, checked against the phytosanitary certificate, matched to the booking confirmation, and then filed with customs. Before KabyTech, one person could handle about 4-5 AWBs per hour. We needed all three ops staff just to keep up, and errors crept in because everyone was rushing."

The errors were not trivial. In the 2025 peak season, Siam Premium Fruits had 11 customs rejections due to data mismatches between AWB data and export declarations. Each rejection took 2-4 hours to resolve, during which the cargo sat in the airport warehouse without temperature control. Three shipments had to be re-graded from Grade AA to Grade B due to quality degradation during the delay, costing an estimated 180,000 THB in lost revenue.

The workflow before KabyTech

Understanding the full workflow makes the inefficiency clear:

1. Receive AWB documents. The airline or freight forwarder sends AWB documents via email as PDF attachments or, in some cases, as photographed documents via LINE messenger. Format varies by carrier.
2. Manual data entry. An operator opens each document, reads the fields, and types the data into the company's ERP system. Fields include AWB number, shipper, consignee, routing, pieces, gross weight, chargeable weight, rate, total charges, special handling codes (PER for perishable, always), and nature of goods description.
3. Cross-reference with booking. The operator compares the AWB data against the original booking confirmation to verify flight number, date, weight, and piece count match.
4. Cross-reference with phytosanitary certificate. Thai agricultural exports require a phytosanitary certificate issued by the Department of Agriculture. The species name, grade, packaging, and net weight on the certificate must match the AWB's nature of goods and weight data.
5. Prepare export declaration. Using the AWB data and supporting documents, the operator prepares the e-Customs export entry through the Thai Customs NSW system.
6. Quality check. A second operator reviews the declaration before submission. In theory. In practice, during peak season, this step was frequently skipped due to time pressure.

Steps 2 through 5 took approximately 12-15 minutes per AWB. Multiply by 12-15 AWBs per day, and the team was spending 2.5-3.5 hours daily on what was essentially manual data transcription.

The implementation

Siam Premium Fruits deployed KabyTech in March 2026, just before the start of peak durian season. The integration was straightforward because they were already using a cloud-based ERP system with API capability.

Step 1: Email forwarding rule (Day 1)

They set up an email forwarding rule that automatically sends all incoming AWB emails to KabyTech's processing inbox. When a PDF or image attachment is detected, KabyTech's API parses it and pushes the structured data to their ERP system via webhook.

Step 2: Field mapping (Day 1-2)

KabyTech's JSON output maps to their ERP fields. The mapping covered all 29 FWB sections, though for their workflow the critical fields were: AWB number, shipper/consignee, routing (origin/destination), flight details, pieces, gross weight, chargeable weight, rate description lines, special handling codes, and nature of goods. The mapping was configured in KabyTech's Operations Portal using a visual field mapper — no code required.

Step 3: Validation rules (Day 2-3)

They configured three custom validation rules in KabyTech:

• SPH code check: Flag any AWB that does not include the PER (perishable) handling code. All their shipments should have it; absence indicates a carrier error that needs immediate correction.
• Weight variance check: Flag any AWB where the gross weight deviates more than 5% from the booking confirmation weight. This catches booking-AWB mismatches before they become customs rejections.
• Routing validation: Flag any AWB where the origin is not BKK (Suvarnabhumi) or DMK (Don Mueang). Their shipments only originate from these two airports; a different origin would indicate a document mix-up.

Step 4: Go live (Day 3)

On day three, they processed their first batch of live AWBs through KabyTech. The team ran parallel processing (manual + automated) for the first week to verify accuracy.

The results after one month

After 30 days of production use, processing 287 shipments:

• Average AWB processing time: 2.4 seconds (API parse time). The human review step — an operator glancing at the parsed data in the ERP and clicking "approve" — adds about 20-30 seconds per AWB. Total time per AWB: under 35 seconds.
• Daily AWB processing time: 12 minutes. Down from 3 hours. One operator handles all AWBs in a single batch each morning. The other two operators have been reassigned to customer relationship management and logistics coordination.
• Customs rejections: zero. In the first month, not a single export declaration was rejected for data mismatches. The cross-validation between parsed AWB data and booking confirmations caught 8 discrepancies before they were filed.
• Extraction accuracy: 98.2%. Of the 287 AWBs processed, 5 required manual correction of at least one field. Four of those were due to genuinely illegible source documents (photographed AWBs taken in poor lighting). One was a parser error on a rate description field that has since been corrected in a model update.

The financial impact

Khun Somchai estimated the following cost savings in the first month:

• Labor reallocation: Two staff members (approximately 160 hours/month combined) redirected from data entry to higher-value work. At a loaded cost of 350 THB/hour, that represents 56,000 THB/month in labor efficiency.
• Avoided customs delays: Based on the previous year's 11 rejections and an average cost of 16,000 THB per rejection (resolution time + cargo degradation), the annualized avoidance is approximately 176,000 THB.
• KabyTech subscription cost: 19,900 THB/month (Professional plan, 500 AWBs/month).
• Net monthly benefit: Approximately 36,100 THB/month in direct savings, plus the avoided rejection costs and the strategic value of having two team members focused on business growth rather than data entry.

What Khun Somchai says now

"The biggest change is not the time saved — it is the stress reduction. During peak season last year, my team was constantly anxious about making errors. Now they review data that is already structured and validated. They catch problems earlier, they have time to communicate with carriers about discrepancies, and they go home at a reasonable hour. For a perishable cargo business where timing is everything, that reliability is worth more than the subscription cost."

Related Articles