Costa Tsaousis: Agentic Observability — The Path to AI Co-SRE

In this talk, Costa Tsaousis explores the evolution of observability, emphasizing its shift from passive dashboards to autonomous AI agents capable of diagnosing, recommending, and remediating issues. He outlines key architectural patterns and machine learning foundations that facilitate this transformation, drawing on his experience in building production AI observability systems. Costa discusses the advantages of consensus-based machine learning over traditional threshold alerting, explains multi-agent orchestration patterns for complex troubleshooting, and provides insights on the future of AI Co-SRE. Attendees will gain a framework for evaluating AI observability tools and an understanding of the current state of the technology.