Field-Service Operations at a European Telco — 32% More Jobs Per Engineer Day on Fibre Rollout

MindMap deployed Route Optimizer (Rt) as the per-engineer-day scheduling-and-routing engine, Workforce Scheduler (Ws) as the multi-day workforce-planning layer, AI Voice Agent (Vb) for the customer-confirmation and pre-arrival coordination, and Delivery Predictor (Dv) as the per-job duration-and-complexity prediction layer.

Phase one was the per-engineer-day optimisation build. Route Optimizer takes the day's work-order queue, the engineer-skill-and-tool-loadout availability, the geographic distribution of the jobs and the per-job duration-and-complexity predictions, and produces the optimised per-engineer schedule. The optimisation considers the live constraints (traffic, weather, parts availability at the regional depot) and re-optimises continuously through the day as job durations differ from predictions.

Phase two was the per-job complexity prediction. The previous workflow had treated jobs as homogeneous within their work-order-type category, leading to systematic schedule slippage when individual jobs turned out more complex than the category baseline. Delivery Predictor uses per-job features (the specific customer address, the building type, the historical complexity at that address, the per-customer prior-job history) to predict per-job duration and complexity with materially higher accuracy than the work-order-type baseline.

Phase three was the customer-coordination build. The AI Voice Agent calls each scheduled customer in the customer's preferred language ahead of the scheduled visit to confirm availability, identify any pre-visit blockers (the customer needs to move furniture, the customer needs to confirm building-access arrangements), and reschedule if the customer's availability has changed. The proactive coordination eliminates much of the previous wasted-trip volume.

Phase four was the engineer-experience layer. Each engineer's mobile app now shows their optimised day with the rationale for the schedule, the per-job complexity prediction with prep guidance, the recommended sequencing within the day with re-optimisation as the day progresses, and the customer-coordination status for each upcoming job.

The platform runs on the operator's Azure tenant in the EU region with full GDPR compliance. The engineer-and-customer data — meaningful PII exposure — stays inside the operator's tenant.

Route Optimizer's per-day scheduling-and-routing engine is a constraint-satisfaction model with continuous re-optimisation. The optimisation handles the per-engineer constraints (skill matrix, tool-loadout, certification requirements per job type), the per-job constraints (time windows, duration estimates, geographic locations), and the dynamic constraints (traffic conditions, weather impacts, customer-cancellation events). The engine runs continuously through the day, re-optimising remaining jobs as actual durations and other variables update.

Workforce Scheduler handles the multi-day workforce-planning layer — engineer-availability rosters, leave-and-training allocation, regional-depot-capacity balancing, demand-forecast-driven engineer-recruitment planning. The scheduling considers the operator's overtime-and-allocation rules as well as the union-agreement constraints.

Delivery Predictor's per-job duration-and-complexity model is a gradient-boosted-tree ensemble trained on the operator's historical job-completion data. Features include the per-address historical complexity (some addresses are reliably more complex than others), the building-type characteristics (high-rise installs differ from single-family installs), the customer-prior-history (customers with prior complex jobs are more likely to have complex new jobs) and the work-order-type baseline.

AI Voice Agent handles the customer-coordination calls in the operator's required EU languages, with the conversation logic targeted around the confirmation-and-coordination flow rather than generic conversation. The voice integration uses the operator's existing telephony infrastructure.

Integration with the operator's existing CRM, work-order-management system and field-engineer mobile apps is via each system's standard inbound API, with the platform feeding optimised schedules and recommendations into the existing systems rather than replacing them.