Vodafone's AI Field-Ops Rollout Cuts Repeat Engineer Visits

手法

Vodafone (10,000+ field technicians, ~80,000km driven daily combined) built "Field Technician Assist" with system integrator Celfocus on Google Cloud, using GenAI/ML to turn unstructured incident data into recommendations. The tool presents engineers with the three best solutions for an incident at the troubleshooting stage of a 4-stage field-service process (travel, onsite troubleshooting, quality verification, ticket closing), engineers pick/rate the recommendation, feeding continuous improvement.

示唆

Results were presented by Vodafone exec Emilio Varas at TM Forum's DTW Ignite 2025 conference and reported by trade outlet TelcoTitans/VodafoneWatch. The 28% figure is an improvement over an earlier 18% reduction seen at proof-of-concept stage, showing gains compounding as the tool matured from PoC to MVP. Less-experienced engineers gaining access to the same AI-derived recommendations as veterans is framed as "harmonising quality" across a large, skill-variable field workforce — the value comes from standardizing a fleet's worst-case performance up toward its best, not from a single breakthrough fix.

成果・金額

28% reduction in repeated site visits achieved (up from 18% at proof-of-concept), targeting 35%; ~10 minutes average time saved per incident per engineer; deployed context covers 10,000+ field technicians and ~80,000km of daily combined travel.

引用

"Vodafone has seen a 28% reduction in repeated site visits when using the app, and the amount of time spent by engineers on each incident has reduced by an average of ten minutes when using Field Technician Assist... the 28% reduction in repeated visits associated with the minimum viable product release representing a significant uptick on the 18% reduction initially seen with the proof of concept. Through continued feedback and enhancement, the operator is now targeting 35% reduction in repeated visits."