Event-Driven AI

Most production failures in agentic systems stem from forcing asynchronous workloads into synchronous request patterns. This session gives the community a reusable architectural vocabulary for building agent systems that scale gracefully — patterns that apply regardless of the framework, model provider, or orchestration layer in use. Making async-first thinking a first-class part of the agentic engineering conversation helps the ecosystem avoid a class of reliability failures before they become widespread.

Transcript

1. AgentCon Toronto Event-Driven Agentic AI MAR 2026 Building Autonomous Agents That Think Asynchronously Hugo Guerrero DEVELOPER ADVOCATE
2. AgentCon Toronto 2
3. AgentCon Toronto 3
4. AgentCon Toronto
5. AgentCon Toronto
6. AgentCon Toronto 6 Agents are asynchronous system participants
7. AgentCon Toronto AI integrated into event-driven architecture THE EVENT-DRIVEN AGENT 7 Agents become part of the system event loop
8. AgentCon Toronto An event-driven workflow 8 EXAMPLE
9. AgentCon Toronto 9 AGENTS NEED CONTEXT Prompts alone are not enough Agents require dynamic context: ▪ system state ▪ user activity ▪ workflow history ▪ environment changes ▪ operational signals 9
10. AgentCon Toronto 10 AN AGENT IS ONLY AS GOOD AS ITS DATA If a user changes their delivery address, the agent needs to know immediately Examples ▪ new transaction ▪ user preference change ▪ anomaly detected ▪ workflow completed
11. AgentCon Toronto 11 CHANGE CONTEXT CAPTURE Observing state changes across systems. Inspired by Change Data Capture Examples ▪ OrderStatusChanged ▪ FraudRiskDetected ▪ InventoryUpdated
12. AgentCon Toronto 12 When we combine event-driven agents with context streaming, we get a powerful architecture EVENT-DRIVEN CONTEXT ARCHITECTURE
13. AgentCon Toronto Let’s break things DEMO 13 13
14. AgentCon Toronto AI Agent & RAG Event Flow Architecture DEMO 14
15. AgentCon Toronto
16. AgentCon Toronto Better UX Resilience Agents continue working independently 16 Scalability Event streams scale naturally Observability Agent decisions become events No blocking or timeouts 16 Operational Advantages WHY THIS ARCHITECTURE WORKS
17. AgentCon Toronto 17 17 Building for the Agentic Mesh ▪ Agents are asynchronous actors ▪ Context must be continuously updated ▪ Capture context changes as events ▪ Build agents on event-driven architecture Scale: Handle 10,000 agents without 10,000 open HTTP connections. Resilience: Events can be retried; synchronous calls just fail. Call to Action: Start by moving one 'bot' to an async worker.
18. AgentCon Toronto