EY and Microsoft Transform Work with Agentic AI
EY has become one of the 14 founding members of Microsoft’s Frontier Firm AI Initiative, a collaboration with Harvard University focused on redesigning work and decision-making through AI. As a Client Zero for many of Microsoft’s innovative solutions, EY transitioned from experimenting with agentic AI to implementing a hybrid model where AI enhances human capabilities to create enterprise-scale value.
The transformation began with foundational work on cloud and data. EY built EY Fabric on Microsoft Azure, a unified cloud-based platform that powers global innovation. The firm also focused on making data accessible, governed, and trusted, exemplified by EY Helix, which unified audit data and analytics across services. These steps provided the infrastructure needed for AI integration.
As generative AI (GenAI) adoption surged, EY rapidly deployed EY.ai EYQ, a private GenAI ecosystem, to 300,000 employees. The firm integrated Microsoft 365 Copilot and enabled teams to build their own agents using Copilot Building and Microsoft Copilot Studio. These tools were unified into a bespoke agentic AI platform, embedded in core workflows and decision-making processes.
The results have been significant, with a 15% productivity boost across the organization. Over 16,000 EY developers use Microsoft GitHub Copilot, accelerating software development by 70%. PowerPost, an AI-driven finance automation tool, reduced lead times by 95% and operational costs by 37%. EY Canvas, built on Microsoft Azure, has enhanced 160,000 audit engagements annually, processing 1.4 trillion lines of journal entry data.
EY’s approach to AI transformation is guided by four principles: integrating AI into existing ecosystems, redesigning work rather than just automating it, fostering human-agent collaboration, and ensuring governance, trust, and cultural change. The collaboration with Microsoft serves as a blueprint for other organizations aiming to become Frontier Firms, leveraging human-AI collaboration to create value at scale.