KPMG integrates Claude across a workforce of more than 276,000

KPMG has announced a global alliance with Anthropic to bring Claude into its core business and give more than 276,000 employees access to the assistant. The professional-services firm says the rollout will cover its audit, tax, legal and advisory work across 138 countries and territories. The announcement was published by Anthropic on May 19, 2026, so it describes a strategic expansion rather than a short product trial.\n\nThe alliance has two connected parts. First, Claude will be embedded in Digital Gateway, KPMG’s platform for client work. Built on Microsoft Azure, the platform brings together KPMG expertise, proprietary tools and client data. Anthropic says Claude Cowork and Managed Agents will let KPMG professionals and clients build AI capabilities inside that environment. The company gives tax work as an example: an agent for adapting to changing regulations, which previously took weeks to build across several tools, can now be assembled in minutes. That is a claim from the companies, not an independent productivity measurement.\n\nSecond, KPMG and Anthropic will develop new Claude-powered products and modernise business functions together. KPMG is also becoming a preferred partner for private-equity firms that want to deploy Claude and Anthropic’s agents in portfolio companies. One named offering, KPMG Blaze, is designed to use Claude Code to help teams update older IT systems. The partners also plan to use Claude to find and fix vulnerabilities in critical systems under KPMG’s Trusted AI framework.\n\nThe scale matters because it moves generative AI further into regulated, knowledge-intensive work. Giving employees access is only the visible part. The harder work is deciding which data an agent may handle, how its output is reviewed, and who is accountable when a recommendation affects a client. KPMG and Anthropic point to security, governance and human judgment as part of the deployment; the source does not provide independent evidence that these controls work in every setting.\n\nFor businesses, the announcement is a signal that AI competition is increasingly about distribution and workflow integration, not just model benchmarks. For makers, it shows why enterprise platforms and implementation partners can determine whether an AI system becomes useful infrastructure. For employees and clients, Claude may reduce the time spent moving between tools, but the value will depend on clear permissions, audit trails and people who can challenge the system’s answers.