Technology stack
The challenge
❌ Before
Associates manually cross-checked GTINs, vendor info, weights, and harvest data against internal systems. Each label could take up to 15 minutes, with frequent transcription errors and compliance risk.
✅ After
Barcode scan instantly parses GS1 Application Identifiers and validates fields in real time, surfacing warnings/errors and eliminating manual entry.
Goals & success metrics
Hypothesis: Automating validation via GS1 scanning will cut processing time by >90% while improving accuracy and compliance.
Research & discovery
- Interviews with internal devs & seafood QA specialists to map pain points
- Deep dive into GS1 AI spec (01 GTIN, 10 Lot, 310x Net Weight, 11/13 Dates…)
- Time-motion analysis of current workflow to identify bottlenecks
- Field shadowing in warehouse environments (lighting, gloves, device handling)
User persona & job story
“When a seafood shipment arrives, I want to validate vendor labels quickly and accurately so I can approve loads without manual entry errors, ensuring quality and compliance.”
Information architecture
- Home → Vendor Label Validation
- Scan Barcode → Parse GS1 → Display structured fields
- Real-time validation → Clear indicators (valid / warning / error)
- History of scans → Detailed record for audits
- Graceful fallbacks for invalid/incomplete data
Design & prototype
- Telerik RadDataForm for grouped, scannable field layout on mobile
- Key fields: GTIN, Vendor, Stewardship, Harvest Method, Weight, Pack Date, State of Origin
- Color-coded validation states (green/valid, yellow/warn, red/error)
- High contrast + large targets for floor conditions
Development highlights
GS1 parsing engine
Implemented comprehensive AI parsing with domain validation:
- 01 GTIN · product identification
- 10 Lot Number · batch tracking
- 310x Net Weight · unit/precision handling
- 11/13 Production/Pack Dates · freshness rules
Display & data binding
- SKU → product image binding
- Weight conversion (e.g.,
001550→ 15.50 lbs) - Date formatting (YYMMDD → mm/dd/yy)
- Null/invalid enum fallbacks
- Real-time validation feedback as fields populate
Custom validation logic
- Required presence: GTIN, Vendor, Pack Date
- State-of-origin rules (e.g., USA/RTC constraints on certain SKUs)
- Quantity vs Weight (≥1 required; both preferred)
- Date sanity checks (no future pack dates)
- Harvest method compliance
Results & impact
⚡ 60× faster
~15 min → ~15 sec per label
✓ Accuracy
Automated parsing removes transcription errors
📱 Mobile-first
Android enterprise devices for warehouse floors
🔒 Compliance
Real-time checks before approval
📊 Audit trail
Complete scan history & records
💚 Trust
Clear indicators + raw data fallbacks
Key learnings
- Design for imperfect data: Real GS1 labels vary widely; fallbacks and explicit errors build trust.
- Field testing matters: Lighting, gloves, and scanning posture change design constraints.
- Enterprise a11y is mandatory: Contrast + target sizes are non-negotiable.
- Parser robustness: Extensive sample coverage prevents production surprises.
- Stakeholder alignment: Regular QA syncs ensured validation matched compliance rules.