AI agents can write code, design systems, and make decisions. But should they act alone? In high-stakes environments, the answer is no. The most effective AI systems aren't fully autonomous. They're collaborative: agents propose, humans approve. In this workshop, we'll build human-in-the-loop systems from scratch. You'll learn how to design approval workflows, implement strategic checkpoints, and create escalation paths that balance automation speed with human judgment. We'll cover practical patterns for different risk levels: when to let agents run, when to pause for review, and when to require explicit approval. Through hands-on exercises, you'll implement HITL patterns using real tools and frameworks. By the end, you'll have a working system where AI agents handle routine work autonomously while surfacing critical decisions to humans at the right moments. Bring your laptop. We're building, not just talking.

Talk Level:
ADVANCED

Speakers:

Lyubomir Bozhinov: Technology Leader | Bridging Strategy, Innovation & Scalable Execution

Stefan Angelov: Architect Lead | Engineering Manager | Senior Java Developer | Public Speaker