In this 45-minute workshop, we’ll dive into practical strategies for designing agentic systems using Large Language Models (LLMs). Rather than focusing on heavy frameworks, we’ll explore how to build powerful, flexible agents with minimal, composable building blocks. Drawing from real-world examples, we’ll cover five key workflow patterns—prompt chaining, routing, parallelization, orchestrator-workers, and evaluator-optimizer—and show how to integrate external tools while keeping implementations straightforward. By the end, you’ll have clear guidance on when to use workflows versus fully autonomous agents, how to balance complexity with performance, and how to test and refine your solutions for robust, real-world deployments.