From Chatbots to Multi-Agent Systems: Lessons from Building Agentic RAG in Production
Corey Yang-Smith
Date:
Invited conference talk at YYC DataCon 2026, BMO Centre, Calgary, AB, Canada
Presented an invited technical deep-dive at YYC DataCon 2026 (Stage 3, Room 212; 3:25–4:00 pm) at the BMO Centre in Calgary, AB, Canada on September 11, 2026.
The talk examined why and when multi-agent systems are useful, compared trade-offs across orchestration, decomposition, specialization, and coordination patterns, and explored how retrieval can be integrated into an agent’s reasoning process. Drawing on lessons from developing Oliver, an AI system for public procurement, it closed with practical considerations for production systems, including reliability and fault handling, LLM evaluations, telemetry, and balancing latency, cost, accuracy, and user trust.
Recommended citation: Corey Yang-Smith. "From Chatbots to Multi-Agent Systems: Lessons from Building Agentic RAG in Production." Invited conference talk, YYC DataCon 2026, BMO Centre, Calgary, AB, Canada, September 11, 2026.
Download Slides