Many agentic AI demos rely on tightly coupled API calls and step-by-step orchestration. This approach works for prototypes, but it breaks down in production environments where failures, latency, retries, and scale are needed. Agents become brittle, workflows are difficult to evolve, and resilience is hard to achieve.

In this session I'll be discussing an alternative architecture: building an Agent Mesh using Event-Driven Architecture. Instead of wiring agents directly to APIs or to each other, agents communicate by publishing and subscribing to real-time events. Each agent focuses on a specific responsibility, reacts to the events it cares about, and emits new events that move the system forward.

Using our open source Solace Agent Mesh as a reference, I'll explain in this beginner-friendly talk how an Agent Mesh enables loose coupling, scalability, resilience, and continuous autonomy for agentic systems. After listening, you'll see why events are a better coordination mechanism than synchronous API chains, and how Event-Driven Architecture provides the backbone that allows agentic AI to move from demos to production-ready systems.

If you want agentic AI to work beyond prototypes, this session explains the architectural shift required.