Status quo
The existing picking system generates a picking summary at the shipment-ID level. This forces pickers — whether ASRS-driven (NXS1) or manual (NXS2) — to fully close one shipment before moving to the next. There is no mechanism to consolidate picks for the same product (PID) across multiple orders within the same picking run.
Problem
For bulk orders, multiple orders frequently contain the same PIDs. Under the current model:
- Pickers travel to the same shelf location multiple times for the same PID across different orders.
- ASRS receives fragmented waves instead of consolidated requests.
- Picker efficiency degrades significantly at scale.
- There is no batching intelligence — every order is treated in isolation.
Goals
- Consolidate picks for the same PID across bulk orders into a single picking event — measured by the reduction in total pick trips per shift.
- Introduce cluster-based physical sorting to map picked items back to individual orders — zero mis-sorts at the pigeonhole level.
- Support both picking paths — ASRS (NXS1) and manual (NXS2) — under one unified cluster model, operational at launch.
- Minimise order dwell time in the picking pool through configurable thresholds.
- Maintain backward compatibility — QC failures and non-bulk orders continue on the existing flow, with no regression in non-bulk SLAs.
My role
I designed this 0→1, with Claude as my design partner throughout. This was an AI-first project: I used Claude to generate working HTML prototypes, converted them into editable Figma screens, and used that as the starting point for building out the system.
Some snippets
