Build a PM Skill that Learns from you (Learning Loops) using Claude Code

Everyone is talking about AI agents solving everything—from coding to world peace. But there’s a fundamental problem no one is addressing: AI systems don’t actually learn after deployment. In this hands-on session, Mahesh breaks down what’s really happening in the world of AI agents—from chatbots (2023) to multi-agent systems (2025) to long-running autonomous workflows (2026). More importantly, he dives deep into the biggest unsolved challenge: learning loops. You’ll learn: - Why current models (even the latest ones) don’t improve over time - The gap between “using AI” and becoming a true builder PM - How to design agents that learn from human feedback - A practical system with doers, learners, and evolving checklists - Real limitations of memory, long-running jobs, and agent performance Plus, a step-by-step lab to help you build your own continuously improving AI agent. If you're a PM, builder, or anyone serious about AI systems—this is the missing piece.