When PwC migrated its global workforce to Microsoft 365, the firm saw an opportunity that went well beyond productivity. With 285,000 people working across more than 100 countries, and with client-serving teams handling sensitive data across every major jurisdiction, the firm needed AI that was not just capable but governable.
The challenge of responsible scale
Most enterprise AI deployments start in one region or one function. PwC's approach was deliberately global from the beginning. The firm began its rollout in 11 countries simultaneously, using a decentralised model that allowed each territory to set its own adoption priorities while maintaining a shared governance framework. Data localisation, regulatory compliance, and security standards varied by territory — the framework had to accommodate all of them.
A hub-and-spoke model
PwC created a global knowledge centre and sharing framework so that adoption practices developed in one territory could transfer to others. Territory teams monitored their own usage, efficiency gains, and adoption patterns across Teams, Outlook, and other Microsoft applications. The firm built custom dashboards — initially using Microsoft-provided data sources, then evolving them as Copilot Studio became available — to track usage at a level of granularity that let leaders identify where AI was generating value and where adoption was stalling.
Results at scale
Within roughly a year of the initial rollout, PwC had completed deployment across the global network. By April 2026, the firm's 285,000+ users were generating more than 20 million Copilot actions per month. The firm estimated that the deployment created more than one million hours of workforce capacity during April alone — time that auditors, tax advisers, and consultants spent on higher-value work rather than on document assembly and information retrieval.
What remains ahead
PwC is now working with Microsoft to develop more granular reporting on agent usage across the network — moving from measuring adoption to measuring business impact per agent. The firm's view is that the next phase of Copilot value will come not from broad deployment but from purpose-built agents trained on specific practice-area knowledge.
“We're embedding AI across our business so our teams can unlock new value, act with confidence, and stay ahead.”