PwC is deploying Claude across its entire global workforce

For the past two years, most enterprise AI initiatives amounted to little more than a dozen analysts playing around with chat prompts and delivering a slick slide deck. Anthropic and PwC want to put an end to the pilot purgatory.

PwC announced a massive expansion of its strategic alliance with Anthropic, preparing to roll out Claude Code and Claude Cowork across its global workforce of hundreds of thousands of professionals. To support the rollout, the two firms are launching a joint Center of Excellence aimed at training and certifying 30,000 PwC staff members in the U.S. alone.

A dedicated business unit around Claude

The marquee move here is structural: PwC is launching a dedicated "Office of the CFO" practice, marking the first standalone business unit built around Anthropic’s tech stack.

The new unit focuses on highly regulated verticals—banking, insurance, and healthcare—aiming to overhaul finance workflows like journal entries, variance analysis, and RFP drafting. PwC started by playing "Customer Zero," dogfooding Claude internally before using it to help Anthropic’s own finance team scale up internal controls and cross-border payroll.

Under the hood, Claude is being integrated into PwC's internal ChatPwC assistant. Through Claude Cowork, it will live directly inside spreadsheets, word processors, and slide software, hooking straight into proprietary enterprise data via Anthropic’s Model Context Protocol (MCP).

10-day insurance underwriting and COBOL legacy rescue

Unlike vague promotional announcements, both companies shared concrete operational metrics from early deployments, claiming overall delivery time cuts of up to 70%.

In insurance underwriting, cycle times dropped from ten weeks down to just ten days—a leap PwC says makes previously uneconomic product lines viable. In enterprise infrastructure, PwC tackled a legacy COBOL mainframe codebase four times larger than originally scoped, reportedly keeping it on schedule and under budget using Claude Code.

Security response cycles shrank from hours to minutes via automated vulnerability patching. Meanwhile, Advocate Health, a healthcare system with 167,000 employees, has signed on as a marquee early adopter to scale these agentic workflows into clinical support and rural community operations.

Why it matters

For enterprise operators and consultants, this shifts AI from an optional typing assistant to an operational requirement. Certifying 30,000 staff members signals that agentic orchestration is becoming a core job qualification. If you run finance or engineering at a major firm, PwC’s pitch will soon be that manual underwriting and slow quarterly migrations are simply legacy tax you no longer need to pay.

For the AI ecosystem, this shows Anthropic's enterprise playbook in action. Selling raw API credits only gets you so far; real enterprise spend is locked behind regulatory audits, legacy databases, and compliance paranoia. Anthropic’s $100 million commitment to its Claude Partner Network pays services giants like PwC to do the heavy lifting of enterprise hand-holding.

We think Anthropic’s positioning as the "grown-up in the room" for regulated industries is paying off against rival models. However, handing automated workflows to agents in high-liability fields like healthcare and financial auditing carries real friction. Compressing a ten-week underwriting review into ten days sounds incredible for efficiency, but it will inevitably test who takes the fall when an agent misses a catastrophic edge case.