Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn to build reliable agentic AI systems that operate effectively in production environments with massive, noisy datasets through this DevOpsDays Tel Aviv conference talk. Discover hard-won lessons from developing an AI agent that processes millions of Kubernetes events daily while achieving 95%+ accuracy in autonomous troubleshooting benchmarks. Explore the fundamental challenge of maintaining system reliability when 90% of data consists of noise, moving beyond LLM capability concerns to focus on robust system architecture. Understand why most agentic AI implementations fail in production environments, including issues with hallucinations appearing as legitimate insights, inability to validate reasoning chains, and the fragile nature of RAG systems when handling complex, interconnected failure scenarios. Gain practical insights from real production iterations, including techniques for building validation frameworks that intercept LLM errors before reaching end users, architectural patterns that constrain problem spaces while preserving system effectiveness, and methods for implementing evidence-based reasoning systems that support systematic auditing and continuous improvement.
