Fleet managers face the challenge of turning massive data volumes into actionable insights. Verizon Connect, a global fleet management provider, addressed this by implementing agentic AI to analyze data from over 1.2 million active vehicles generating 500 million data points daily. The system handles data at scale while maintaining cost-efficiency through a serverless architecture.
The solution begins with anomaly detection, where structured data is processed to identify specific issues before triggering AI agents. These agents, built using Strands Agents SDK and running in AWS Lambda, autonomously investigate patterns and generate insights. The process involves two stages: first, aggregating anomalies into coherent insight candidates, then using detailed investigation to produce data-backed insights.
This approach allows the system to adapt to each user's fleet context rather than relying on static rules. The agentic AI enables real-time pivoting of investigation strategies when unexpected correlations emerge, making it particularly effective for fleet management where variables are unpredictable.
Source: awsml
