The Agent Shift: Building the Foundation for Enterprise AI

The transition from using AI as a tool to adopting AI as an operating model—referred to in this report as the “agent shift”—requires more than better models and faster infrastructure. Organizations need governance and controls that connect people, processes, and data in real time, enabling AI to act reliably on that intelligence.

This shift requires companies to reconsider their architecture and operating model at the same time. That means rebuilding data infrastructure around accessibility rather than volume, replacing fixed technology stacks with composable architectures that can evolve as models and tools change, and addressing questions of AI sovereignty—including where intelligence is performed, who controls it, and how it operates across organizational and jurisdictional boundaries.

Key findings from the report

Enterprise AI scaling is a structural challenge

Process-first companies are moving forward as global AI spending grows exponentially and model capabilities evolve faster than most organizations can integrate them. However, the vast majority of companies still have not grown revenue through AI or fundamentally reimagined how they operate.

Companies that generate sustainable profits share common disciplines. They treat process redesign as a prerequisite for model selection and plan for how the technology will evolve, rather than retooling roles and workflows after deployment. For these organizations, agent migration starts with the operating model.

Data preparation matters more than data richness

Many companies realize too late that having data and having AI-enabled data are two different things. A sovereign, composable foundation can query and prepare data where it lives—without requiring migration or centralization—and transform raw data assets into intelligence that AI agents can act on.

As data residency laws, multi-cloud environments, and structural complexity make centralization increasingly impractical, sovereign control over where models run and where data persists remains adaptable.

Download the report to explore the agent shift, enterprise AI operating models, data preparation, composable architectures, and AI sovereignty.

This content was created by Insights, the custom content division of MIT Technology Review, and not by editorial staff. It was researched and written by humans, and any AI tools that may have been used were limited to production processes under human supervision.