Guardrails That Scale: Building Production-Ready AI Agents in 2026

In this Relevance Live session, Andrew Suryanto (returning guest) revisits one of our most popular topics — building well-behaved agents with guardrails — for a full 2026 refresh, showing how safety patterns have evolved as agents move from prototype to production. You'll learn: - How guardrail best practices have evolved since 2025 - Building compliance frameworks that scale across multi-agent workforces - Designing escalation protocols that hold up under complex use cases - Keeping agents safe without throttling their usefulness - Real examples from agents running in production at Relevance AI This session is ideal for anyone building agents for production use — operations teams, support leaders, or builders who want their agents to behave reliably at scale. We recommend participants have tried the Relevance AI platform and built or experimented with at least one agent to get the most value. Back in September 2025, Andrew ran one of our earliest sessions on agent safety. Since then, the platform has evolved, agent use cases have grown more complex, and what "safe" looks like has changed — this is the updated playbook. --- Get started with Relevance AI: relevanceai.com Join our community: relevanceai.com/community