Operations | Monitoring | ITSM | DevOps | Cloud

Last week, a fleet of autonomous robotaxis in China suddenly stopped working—at scale. Over a hundred vehicles stalled across a city, stranding passengers in traffic and raising immediate concerns about safety, reliability, and trust in autonomous systems. This wasn’t just a bad day for self-driving cars. It was a distributed systems failure, one that happened in the physical world, not just in dashboards.

If you’re like everyone else who works in software development, it’s a good bet that almost every single thing that you thought you knew about your business and engineering has changed as a result of the advent of modern LLMs. How should you respond to these changes? How should you change how you and your team develop software?

OrionIQ is Logz.io’s new agentic observability platform designed to move teams from detecting issues to resolving them automatically. As AI accelerates software development, operations remain manual: engineers still wake up at 2 a.m. to investigate alerts and rebuild context. OrionIQ uses AI agents to analyze real-time telemetry, investigate incidents, identify root causes, and take action across systems.

Most "agent observability best practices" content reads like a compliance checklist from 2019 with "AI" pasted over "microservices." Implement comprehensive logging. Establish evaluation metrics. Create governance frameworks. Not a single line of code. No mention of what happens when your agent silently picks the wrong tool on turn 3 and you need to figure out why.

This guest blog post is by Tohn Furutani, SRE Engineer at NTT DATA. Over the past year, the conversation around generative AI has shifted from single-shot use cases—such as summarization, Q&A, and chat interfaces—to agentic AI systems that can make decisions based on context, plan multistep actions, invoke tools, and adapt as conditions change.

Teams running LLM applications in production face a cost problem that traditional APM tools were never designed to solve. CPU and memory costs are relatively predictable — a web service processing 1,000 requests per second costs roughly the same week over week. LLM API costs are not. A single user session can cost $0.01 or $5 depending on prompt length, model choice, conversation history, and how many retries happen inside your chain.

Continuous monitoring tools track system health, performance, and behavior in real time across production environments. For a deeper understanding of how this fits into modern DevOps practices, see this guide on continuous monitoring and its impact on DevOps. They collect logs, metrics, and distributed traces across the infrastructure and application layers, giving engineering teams visibility into how their systems are running, where anomalies occur, and when something needs immediate attention.