The End of Traditional Technical Service? How AI Agents can Transform Customer Service
Key Notes from our speaker Abhimanyu Kanwar
-
Technical service engineers face major delays due to the need to navigate complex documentation across diverse systems, leading to extended downtime, strained customer relationships, and increased operational costs.
-
AI solutions were developed using technical manuals and ticketing data, with a focus on tailoring responses to distinct user personas based on role, customer type, and region to improve relevance and effectiveness.
-
Persona modelling was central, with engineers, supervisors, and varied industry users requiring different tones, levels of detail, and content framing to align with their practical needs and expectations.
-
Two models—standard RAG and graph RAG—were tested among experienced field engineers; graph RAG outperformed in accuracy, hallucination rate, and dramatically reduced resolution time from 29.5 minutes to 20 seconds.
-
Engineers preferred direct, efficient responses over polite tone, especially among younger and less formally educated users, who benefited most from AI-supported troubleshooting.
-
AI adoption promises substantial cost savings by reducing the need for expert travel (€2k–5k per visit) and minimising downtime (up to €100k/day), with future focus on domain-specific accuracy, multimodal inputs, and proactive support.
