The Setup
An online pharmacy customer. Same-day home delivery of prescription and over-the-counter items. Two delivery shapes offered on one platform, an immediate 2-hour window and a scheduled evening window. The customer ran the storefront and pushed orders to Mycelium via a thin CRUD API. Mycelium ran fulfillment.
What Was Deployed
- Autonomous dispatcher, no human in the loop. Combined orders across time slots. Applied the customer’s rules and Gett’s carrier rules.
- Order combining by location and window that packed up to five deliveries per courier trip.
- Carrier export to Gett, with order-level rule sync.
- White-labeled consumer tracking link that expired after delivery.
- Both delivery shapes on one optimization backbone.
Results
- 1 → 5 deliveries per courier trip. Before Mycelium, couriers ran one delivery at a time. After, order combining packed up to five per trip into the same origin/destination shape.
- Same-day delivery promotion. Moved from paid premium to free default consumer offer. The economic result of packing more deliveries per courier-hour.
- Two delivery shapes on one backbone. No separate pipelines for immediate vs scheduled.
- Zero human dispatchers in the loop.
The primitive that came out of this
The order-combining primitive built for this engagement now lives in the constraint library. Combining by location and time window across mixed delivery shapes is no longer a custom build per customer. It became a shape any vertical with mixed-cadence delivery can pick up.
That is how the library compounds. Work modeled once for one customer becomes capability available to the next. The order-combining shape is one of dozens built into the constraint library across nine verticals.
The Economic Lever
The courier cost per run didn’t fall. The number of orders per run rose. Order combining by location and time window was the mechanism. That compression is what let the pharmacy make same-day free without raising prices.