Harnessing Event-Driven and Multi-Agent Architectures for Complex Workflows in Generative AI Systems

Explore how event-driven architectures and multi-agent systems can revolutionize generative AI workflows in this 36-minute conference talk from the Data Streaming Summit 2025. Learn to address the limitations of current generative AI systems that struggle with complex business workflows by implementing real-time processing capabilities and decentralized decision-making mechanisms. Discover the theoretical foundations behind combining event-driven systems with multi-agent architectures to create highly adaptive, efficient, and scalable AI solutions. Examine practical applications where these integrated approaches enhance the collaborative intelligence of AI systems while maintaining responsiveness to dynamic data flows. Understand the challenges involved in debugging, tracking data flow, and ensuring consistency in these complex systems, along with strategies for implementing robust observability and error-handling mechanisms. Gain insights into LLM capabilities assessment, multimodal support evaluation, and the current state of frameworks for integrating agentic AI with event-driven architectures, while exploring future research directions for advancing these promising technological combinations.