Building Autonomous Dark-Store Spatial Routing

Role Independent Product Strategist

Timeline August 2026 – Present

Overview Zepto is built on the promise of 10-minute quick commerce. While the delivery network is highly optimized, the physical reality inside a "dark store" remains chaotic during peak demand. HyperCart was created to explore a new operational model: using algorithmic batching and spatial routing to minimize picker transit time inside the warehouse, shaving critical seconds off the fulfillment SLA.

Problem During an analysis of Zepto’s fulfillment lifecycle, I observed a recurring bottleneck during evening demand spikes. Whenever orders surge, pickers typically:

1. Receive an order on their handheld device.
2. Look at the screen, locate the aisle, and walk to the item.
3. Scan the item.
4. Look back at the screen for the next item (which might be on the opposite side of the store).
5. Bag the items and hand off to the rider.

The result is:

• High cognitive load on the picker
• Inefficient "zig-zag" walking patterns across the dark store
• SLA breaches during peak times (monsoons, festivals)
• Increased fatigue and error rates (picking the wrong SKU)
• Congestion in popular aisles

The challenge was not rider speed. The challenge was physical navigation latency and screen-checking friction inside the dark store.

Research & Discovery To validate the problem, I modeled dark-store operations and analyzed typical quick-commerce fulfillment telemetry. Several common patterns emerged:

• Key Insight #1: Pickers spend up to 40% of their 2-minute limit looking down at their screens rather than moving.
• Key Insight #2: Orders are traditionally routed chronologically, not spatially. A picker might walk to Aisle 1, then Aisle 6, then back to Aisle 1.