Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore the ongoing debate between agent-based and workflow-based approaches in AI development through this 16-minute conference talk from the AI Engineer World's Fair. Learn why the dichotomy between ReAct-type agents running in loops versus structured workflow graphs is a false choice, as demonstrated through real-world case studies and practical examples. Discover how teams have successfully implemented workflow graph-based solutions, agent-based approaches, and hybrid methodologies that combine both primitives. Understand the definitions and trade-offs of agents versus workflows, examine composition patterns and design considerations, and see how structured workflow graphs can serve as tools for agents or handoff mechanisms between them. Gain insights into architectural patterns for composing agents and workflows, critique dogmatic approaches to AI development, and explore the case for fluent syntax in API design. The presentation includes practical examples of when each approach makes sense and demonstrates how developers can effectively tinker with multiple approaches to solve complex problems.
