Shipsy’s slot planning engine designs delivery slots that the network can actually hit, prices them for capacity, sells them through a consignee-facing booking flow, and then enforces adherence in execution — with predictive drift detection tied to Clara’s proactive outreach. Enterprises running scheduled last-mile on this engine typically hit 95%+ slot adherence, up from 70–80% on legacy time-window systems.
Why we built this
A slot is a promise, and most networks are offering slots they can’t keep. The root cause is that the slot catalog is designed by the marketing team (who want short windows to maximize conversion) and the operations team is handed the bill. By the time an operator realizes a slot is chronically over-promised, thousands of consignees have already booked into it and missed-slot metrics are embarrassing.
The other failure mode is the opposite — ultra-conservative 4-hour windows that are easy to hit but destroy CX because the consignee has to wait home for half a day. We built slot planning as a closed-loop mechanism: the slot catalog is derived from provable network capacity, priced per slot, and monitored continuously.
How it works
The engine runs four linked components:
Component 1 — Capacity-based slot design. For each geographic slice and service type, the engine computes the network’s provable capacity per candidate slot (e.g., 9-11am on Tuesdays, lane-by-lane). It accounts for historical completion rates, driver availability, vehicle count, and the operational cost of serving that slot (fewer stops fit in a tight urban 2-hour slot than in a suburban 4-hour slot). Slots are published to the booking flow with capacity caps.
Component 2 — Dynamic pricing and availability. As consignees book slots, availability decrements in real time. Peak slots can be priced at a premium (or closed early if at capacity). The engine supports multiple pricing strategies: flat, dynamic-by-capacity, dynamic-by-demand. Sold-out slots are visibly closed to avoid overcommitting.
Component 3 — Plan-aware execution. Bookings flow into Astra’s daily planning. Astra treats slot windows as hard constraints — never reordering stops in a way that breaches the slot. Slots are sequenced intra-slot by micro-cluster routing, so a dense cluster of bookings inside a slot is delivered in an efficient sequence without leaving the slot.
Component 4 — Drift detection and proactive rescue. During execution, the predictive ETA engine watches for slot-drift risk — when the cumulative delay in earlier stops projects a high probability of missing a downstream slot. Risk is scored per pending slot. When risk crosses threshold, three actions are available: dispatcher-triggered replan (pull in a nearby driver), Clara-driven proactive reschedule (offer the consignee an alternative), or escalation to a human CX lead for high-value shipments.
All three actions are prioritized by value-at-risk — a premium-slot breach for a high-value consignee takes precedence over a standard-slot breach for a low-value one, and drivers are replanned accordingly.
Here’s the sequence at a glance:
Early results
Enterprises deploying slot planning plus adherence typically report, within 60 days:
- 95%+ slot adherence on the slots the engine has priced — up from 70–80% on catalogs designed outside the engine.
- CX scores up 5–10 points because the slots the network offers are slots it hits.
- Reduced re-attempt costs because slot-breach-driven failed deliveries are rare — the system catches drift before the miss.
- Planner and dispatcher time on slot incidents drops 50%+.
A global big-and-bulky retailer leading in furniture and home goods uses this engine for its 95% first-attempt delivery rate on scheduled furniture deliveries — tight slot design plus proactive drift detection.
What’s next
Three upgrades: consignee-negotiated slot swaps where Clara autonomously offers and accepts alternative slots with the consignee during drift, calendar-integrated slot booking for B2B consignees with booked dock times, and forecast-driven slot catalog reshaping where the engine proposes catalog changes based on seasonal or structural shifts in demand.