Deep Dive to AI Agent Observability

This 31-minute conference talk explores how OpenTelemetry can be adapted for AI Agent observability in cloud-native environments. Dive into the challenges and solutions for monitoring multi-agent systems, tracing inference workflows, and correlating AI-specific metrics like model performance and decision latency. Learn practical approaches to extending OpenTelemetry for AI-specific requirements, integrating with popular observability tools, and generating actionable insights for production-grade AI systems running on Kubernetes. Through demonstrations and real-world use cases, discover how to achieve transparency, reliability, and optimization for AI-driven architectures. Presented by Guangya Liu from IBM and Karthik Kalyanaraman from Langtrace AI at a CNCF event.