This session is for developers, technical leaders, and product builders looking to move beyond simple chatbots and start building real agentic systems that can reason, plan, and take action.
The session aims to demystify what "agentic AI" actually means in practice and provide a clear roadmap for designing, building, and deploying agents that deliver genuine value rather than impressive demos.
The session will cover:

What makes a system "agentic" - distinguishing agents from workflows, chains, and traditional LLM applications
Core architectural patterns including tool use, memory, planning, multi-agent collaboration, and human-in-the-loop design
A practical walkthrough of building an agent from scratch - from prompt design to tool integration to evaluation
Common failure modes and how to handle them: hallucinations, runaway loops, cost spirals, and unreliable tool calls
Production considerations including observability, guardrails, evaluation frameworks, and security
An honest look at where agents excel today and where they still struggle

The focus will be on giving attendees both the conceptual foundations and the practical patterns needed to start building agentic systems that work reliably outside of a demo environment.