.png)
Swisscom
Telecom · Switzerland
1.1B CHF efficiency gain
Swisscom combines predictive network AI with a GenAI customer assistant resolving 70% of first-line queries
Two AI tracks in parallel: predictive maintenance that auto-remediates 60% of network issues without an engineer, and a GenAI assistant that handles 70% of first-line customer questions, together producing the largest cumulative AI efficiency gain in European telecom.
Friction
Swisscom faced pricing erosion and heavy 5G and fibre investment pressure. Customer contact happened through call centres with long wait times and high repeat query rates. Network management was reactive, with issues fixed after customers noticed them.
Breakthrough
Network AI platform combines telemetry from thousands of network elements with anomaly detection and root-cause analysis. Auto-remediation scripts resolve 60% of issues without an engineer. The customer service assistant uses a combination of proprietary and commercial LLMs with retrieval over Swisscom-specific product data and CRM context.
Impact
1.1 billion CHF cumulative efficiency gain over 2020–2025. NPS rose 18 points in customer service segments where the AI assistant is active. 70% of first-line customer queries resolved without human intervention. Network downtime reduced by 30% through predictive maintenance. 80%+ AI adoption among employees in supporting functions.
Unlock the full analysis with breakthrough, impact, what made it smart and its technical approach below!
Problem
Telecom operators face a double pressure: network complexity is growing with 5G, fibre, and IoT, while customer expectations are rising simultaneously. Traditional approaches could not scale to meet both challenges within the cost envelope.
What made it smart
Running network AI and customer AI on a shared platform architecture created compounding returns. Network reliability improvements reduced inbound customer contact; GenAI handled the remaining queries; the combination produced results neither system could have achieved alone.
Technical approach
Network AI ingests telemetry from thousands of network elements. ML models detect anomalies, identify root causes, and trigger auto-remediation scripts. Customer assistant built on combined LLM stack with CRM and product data retrieval. Both systems share underlying data infrastructure and ML ops processes.
Strategic lesson
The most durable AI transformations are cross-domain. When reliability AI and customer AI run on the same platform, improvements in one create spillovers in the other.
Reflection question
Where in your organisation are operational reliability and customer experience treated as separate improvement tracks, and what would happen if they shared an AI infrastructure?