From a two-route pilot to thousands of customer calls a week.

The company, a large US based e-commerce company in home goods, operates a large-item delivery network across the United States, handling thousands of scheduled deliveries daily. For years, drivers called customers before delivery to confirm appointments and collect special delivery instructions. Manual calls didn't scale with volume. Automated emails weren't driving engagement. When customers weren't reachable, deliveries failed. Analysis of thousands of delivery records showed the root causes weren't primarily customer behavior — they were process gaps that could be fixed with proactive outreach.

A customer coordination agent — running on Atlas (Pedestal’s knowledge layer every agent inherits from) the analyst built from the company’s routing, scheduling, customer, and delivery history systems — called customers ahead of scheduled deliveries, confirmed appointments, collected special delivery instructions, and passed the information to drivers via text. Two routes became a single home-delivery operation, then the whole network. Scaling required more than working automation: evaluation frameworks versioned in Atlas against the company’s standard operating procedures, and monitoring that caught issues before they reached customers. Leadership could see exactly how the system performed before approving each expansion.

Today, Pedestal handles day-of coordination across multiple delivery sites, running thousands of customer calls a week through the platform. Customers engage through voice or SMS — whichever they prefer. What began as a request to distribute driver insights became core infrastructure for customer coordination, because the data showed a bigger problem, and the agent runs on Atlas the analyst built to prove it.