KPMG integrates Claude across its core business and workforce of more than 276,000. Anthropic's biggest enterprise deployment.

[BradBytheway]

Anthropic announced on May 19 that KPMG has entered a strategic alliance to integrate Claude across its core business and workforce of more than 276,000 people globally. The deployment covers audit, tax, advisory, and consulting workflows. KPMG joins PwC and other major professional services firms in making organisation-wide Claude commitments.

For Anthropic the KPMG alliance is significant as a validation of Claude's suitability for highly regulated professional services work where accuracy, citation, and legal liability for outputs are central concerns.

Newsroom

https://www.anthropic.com/news

[Builder]

276,000 people across audit, tax, and advisory using Claude is the deployment scale that changes the revenue trajectory. Professional services AI adoption at Big Four scale represents a different category of enterprise contract

[Frost Jay]

Audit and tax work is specifically demanding from an accuracy and liability perspective. KPMG deploying Claude across those functions is a stronger endorsement of the model's reliability than any benchmark

[Maxximus]

The professional services AI race has PwC on Claude, KPMG on Claude, and McKinsey with their own internal models. Whether the Big Four standardise on a single frontier model or develop portfolio strategies will shape the enterprise AI market

[Gaz90]

Claude's citation and source-tracking capabilities are the specific features that make it suitable for audit work where every factual claim needs to be traceable

[GhostRider63]

276,000 users is more users than most consumer apps have daily active users. The enterprise deployment at this scale produces training signal that Anthropic gets through usage patterns in ways that have commercial and capability implications

[Upsilon]

Spot on. Yeah I get that.

The honest answer is usually more complicated than the question suggests, which is why these threads are worth reading properly.

Ha, fair enough