Bringg launched Predictive Time on Site (PToS), an AI capability that replaces static service time estimates with per-stop predictions trained on each merchant's operational data. It creates more-accurate per stop estimates and unlocks fleet capacity hidden in padded schedules. Bringg also launched an interactive Route Capacity Value Calculator.
The problem PToS solves
Most enterprise delivery operations use an average service “time per stop” to define a routing plan. Drivers spend 62% of their route time off-vehicle, which makes service time the largest route plan input [Calderón et al., Transportation Research Part A, 2026]. Dispatchers then pad schedules with safety buffers that lock fleet capacity and suppress delivery slots at checkout. That inefficiency impacts both costs and revenue considering labor represents 50–60% of last-mile delivery expenses, and 55% of consumers churn after a negative delivery experience [Bringg, 2025].
“PToS gives retailers two things from one operation: the route accuracy that controls costs and the fleet capacity that drives revenue," said Guy Bloch, CEO of Bringg. "That’s the real prize of AI in the last mile, and it starts with getting the inputs right."
What PToS delivers
PToS trains a dedicated model for each merchant using their historical order data. The model weights the variables that impact stop duration: SKU weight and combinations, building access constraints, installation type, crew configuration, and service complexity. Each model retrains automatically when seasonal patterns shift or workflows change.
In a proof of concept, PToS ran against six months of a major grocery retailer's delivery data. The model cut average service-time estimation error by 54%, from roughly five minutes to two and improved forecast accuracy by 39%. Bringg's capacity model projects the operation could serve the same daily volume with 17% fewer routes, recover roughly 25 minutes per route, free about 60 driver-shifts a day, lift deliveries per hour by 18%, and save over $500,000 annually in labor (Bringg POC, 2026).
“Static service time averages hide capacity and generate costs that operations teams can’t attribute or fix," said Yishay Schwerd, CPTO at Bringg. "PToS addresses that problem at the source: a dedicated model per merchant, trained on their own data, that gets more accurate over time."
Predictive Time on Site complements Bringg's larger route planning and real-time execution modules and is immediately available to Bringg's 800 customers across 70 countries.
