
November 23–27, 2026
Enterprise Architect
IBM
Generative AI is rapidly moving beyond copilots and conversational assistants toward autonomous and semi-autonomous agents that can reason across enterprise data, invoke APIs, coordinate workflows, and make or recommend business decisions. But moving from an impressive prototype to a production-grade enterprise AI system requires much more than connecting an LLM to a set of tools. In this session, I will share a practical architecture for building governed agentic AI systems that combine large language models, retrieval-augmented generation, enterprise APIs, structured data, workflow orchestration, human oversight, security, observability, and policy controls. I will explore how organizations can determine where an agent should simply provide information, where it can recommend an action, and where it can safely execute an action on behalf of a user. Using real-world enterprise patterns and representative examples from software modernization, content operations, digital experience platforms, and workflow automation, attendees will learn how to design multi-step AI systems that are scalable, auditable, secure, and measurable. I will also cover common failure modes—including hallucination, excessive autonomy, weak context, tool-selection errors, uncontrolled costs, and lack of traceability—and show architectural approaches for addressing them. The session is designed for architects, engineering leaders, developers, and technology executives who want to move beyond AI experimentation. Attendees will leave with a practical framework for deciding where agentic AI creates genuine business value and how to take those use cases from proof of concept to trusted enterprise deployment
See every session across the AI Agents Conference November 2026 line-up.
View Agenda