GitOps for AI Agents - Building Reliable AI Pipelines With Argo

Explore how to build production-grade AI agent pipelines using GitOps principles and Argo's workflow orchestration capabilities in this 22-minute conference talk. Learn to address the complex orchestration challenges that emerge when AI agents transition from experimental prototypes to production workflows, including state management, retry mechanisms, memory handling, and coordination across multiple tools, APIs, and data systems. Discover how Argo Workflows enables runtime execution of sophisticated agent pipelines by chaining tasks, managing dependencies, and scaling dynamically based on resource requirements, while Argo CD delivers GitOps-style control over Kubernetes infrastructure and agent configurations. Master the implementation of full lifecycle management through declarative deployment strategies, dynamic execution patterns, and safe rollback procedures. Gain insights into advanced techniques for layering agents on top of Argo to unlock enhanced capabilities such as safe iteration on prompt-driven pipelines, automatic rollback of failed steps, and integration of observability tools with human-in-the-loop checkpoints for reliable AI agent operations.