AI in Production: What Breaks and Why

Moving AI from pilot to production introduces a new set of operational challenges that traditional monitoring was never designed to handle.  

In this session, we'll explore the most common failure modes of AI and agentic systems in production—from silent quality drift and hallucinations to tool failures, model changes, and rising operational costs.  

Through real-world examples and live demonstrations, you'll learn how to identify issues before users are impacted, gain end-to-end visibility across prompts, models, agents, and infrastructure, and build the confidence needed to scale AI workloads reliably and safely.