Key results
The company
Fastweb + Vodafone
fastwebvodafone.itIntegrated fiber broadband and 5G mobile telecommunications provider in Italy.
Result highlights
- 90% correctness rate for AI agent
- >86% one-call resolution rate via AI
- 82% resolution rate for customer queries
- 5.2/7 Customer Effort Score achieved
The story
One of Europe's leading telecommunications providers serving millions of customers across Italy with complex connectivity and billing needs.
Existing chatbots struggled to resolve nuanced requests requiring contextual understanding, while human agents relied on manual, time-consuming searches across multiple systems and knowledge bases to find answers.
The organization built an agentic architecture using LangGraph where a supervisor node routes queries to specialized sub-agents that can execute API transactions directly. Operational procedures are converted into a Neo4j knowledge graph via an automated ETL pipeline, powering an internal assistant that guides consultants through step-by-step diagnostics. LangSmith monitors the production environment, using automated daily evaluations to grade response correctness against company guidelines.
Scope & timeline
- Knowledge graph updates deployed in hours
Quotes
“You can’t run agentic systems in production without deep observability. LangSmith gave us end-to-end visibility into how our LangGraph workflows reason, route, and act, turning what would otherwise be a black box into an operational system we can continuously improve.”