From data overload to actionable insights: How Verizon Connect scaled agentic AI to 100,000 users

2026-05-27 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Data Science & Analytics · Depth: Intermediate, long

Summary

Verizon Connect, a global fleet management provider, developed and scaled an agentic AI solution to transform overwhelming telematics data into actionable insights for 100,000 users daily. Managing over 1.2 million active vehicle subscriptions generating 500 million data points daily across 80,000 indicators, the company faced challenges identifying critical patterns manually. Their solution, rolled out in November 2025, employs a two-stage agentic architecture. First, a serverless statistical model using AWS Step Functions and Lambda performs computationally intensive anomaly detection, identifying what anomalies exist. Second, Strands Agents, running on AWS Lambda, orchestrate LLMs (initially Claude 4.5 Sonnet, then Claude 4.5 Haiku, and now Amazon Nova 2 Lite) to reason about why anomalies occurred and how to address them. This approach reduced input token costs by 70% with Nova 2 Lite, delivering insights within a five-hour window by managing concurrency via Amazon SQS.

Key takeaway

For AI Architects and MLOps Engineers scaling generative AI solutions, consider a hybrid agentic architecture to manage data overload. You should offload computationally intensive numerical analysis to specialized serverless models before engaging LLMs for reasoning and insight generation. This strategy, exemplified by Verizon Connect's 70% cost reduction, allows you to deliver personalized, actionable intelligence efficiently to a large user base while maintaining performance and staying within API quotas.

Key insights

Agentic AI transforms vast fleet data into actionable insights by dynamically investigating patterns and adapting analysis.

Principles

• Offload numerical analysis from LLMs.
• Prioritize insights based on context, not static rules.
• Agentic systems adapt to unpredictable data patterns.

Method

A two-stage process: statistical models detect anomalies, then Strands Agents orchestrate LLMs to reason, aggregate, prioritize, and investigate detailed insights using various data tools.

In practice

• Use serverless statistical models for anomaly detection.
• Deploy agents with open-source SDKs like Strands Agents.
• Optimize LLM costs by transitioning to efficient models.

Topics

• Agentic AI
• Fleet Management
• AWS Serverless
• Large Language Models
• Anomaly Detection
• Strands Agents
• Amazon Bedrock

Best for: AI Engineer, MLOps Engineer, AI Architect

Related on AIssential

• Presentation: What I Learned Building Multi-Agent Systems From Scratch — Specialized, lean AI agents outperform monolithic prompts, drastically reducing task times and improving efficiency.
• Scaling Down Is the New Scaling Up — Future agentic AI prioritizes hardware-aware design for efficient on-device operation over raw model size.
• Orion at Gravity: Trustworthy AI Analysts for the Enterprise — Agentic analytics combines semantic layers and broad context engineering to deliver trustworthy, actionable business insights.
• Position: Agentic AI System Is a Foreseeable Pathway to AGI — Agentic AI, leveraging task decomposition and multi-agent collaboration, offers exponentially superior efficiency for AGI over monolithic scaling.
• Your agentic AI pilot worked. Here’s why production will be harder. — Scaling enterprise agentic AI is an architecture, governance, and organizational challenge, not primarily a technology problem.
• The Sequence Radar #828: NVIDIA’s GTM Releases, Bezos’s $100B Bet, Xiaomi’s Ambush, and the Fracturing of OpenAI — The AI industry is rapidly shifting towards agentic infrastructure and physical world integration, moving beyond conversational paradigms.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.