Enterprise Architecture in the Age of Agentic AI
Cristi Nastase, ExxonMobil
Cristi Nastase, Sr. Principal Enterprise Architect – GenAI at ExxonMobil, discusses how enterprise architecture must evolve as agentic AI moves from experimentation into operational reality.
The presentation explores how the centre of gravity in AI is rapidly shifting from traditional machine learning toward GenAI and agentic AI, while governance, controls, and operational discipline struggle to keep pace. As organisations push beyond experimentation, ExxonMobil highlights the growing need for enterprise ecosystems that can support AI at scale without creating fragmentation, uncontrolled sprawl, or “pilot purgatory”.
A major focus of the session is trust and explainability. Nastase argues that organisations must move beyond “black box” AI by designing systems with verifiable lineage, auditability, and policy-driven controls built in from the start. In practice, this means enabling operators to understand not only what an AI agent recommends, but also the data, models, and logic behind every decision.
The session also challenges organisations to rethink where real competitive advantage will come from in the age of agentic AI. Rather than relying solely on foundation models, ExxonMobil emphasises the importance of intelligent workflows, proprietary operational data, strong governance, and human judgement capable of asking the right contextual questions. The message is clear: enterprise AI success depends as much on architecture and organisational readiness as it does on the technology itself.
