Architect Series: Navigating the Agentic Crux in the Enterprise

Agentic AI — autonomous agents capable of planning, using tools, and executing multi-step workflows — promises a new era of enterprise productivity. Yet early real-world attempts reveal a hard truth: while agents can be remarkably capable, their behavior is often unpredictable. Without a strong underlying data and control architecture, autonomous workflows can drift, improvise, or fail in ways that are difficult to observe, govern, or recover from.

In this session, Redpanda CTO Tyler Akidau and Principal Software Engineer Tyler Rockwood will examine why enterprises cannot treat agents as “just another LLM,” and why governance becomes the defining challenge in the enterprise. 

The discussion will outline what it takes to bring order to this emerging landscape: a governed access layer that enforces data safety, short-lived credentials, lineage and traceability, deterministic action logging, and full replayability. Attendees will leave with a principled framework for deploying agentic systems responsibly — not as experimental prototypes, but as reliable, governed components of production AI infrastructure.