AI agents are fundamentally changing enterprise data management by requiring direct database access rather than relying on traditional batch processing pipelines. These agents frequently create short-lived databases to test various reasoning chains before executing tasks, which necessitates new approaches to data governance and auditing.
Sanjeev Mohan, principal at SanjMo and author of the book 'Designing the AI-Driven Data Foundations,' argues that data observability must now be integrated directly into the system architecture. This shift is essential because agents act as first-class data consumers that build context during runtime.
The industry is moving toward platforms capable of converging structured and unstructured data to support these agentic workflows. Mohan’s book, published by John Wiley & Sons Inc. in June 2026, explores how these developments are redefining long-standing architectural principles.
The topic of AI inference and agentic workflows was featured in a dedicated track at the KubeCon + CloudNativeCon NA conference. Experts emphasized that the success of AI models in production depends heavily on the quality and transparency of the underlying data foundation.
