Question 1

What is the KPMG-Anthropic alliance and what does it actually include?

Accepted Answer

The KPMG-Anthropic alliance, announced on May 19, 2026, is a global strategic partnership in which KPMG will integrate Claude across its core business and workforce of more than 276,000 employees. Claude will be embedded within KPMG's Digital Gateway — the firm's primary client-delivery platform — starting with the Tax and Legal practice and expanding to other advisory services, including Audit, Advisory, and Consulting. The deployment runs on Microsoft Azure. Full implementation is targeted for the end of September 2026. The practical implication is that Claude will be embedded in the work product that KPMG delivers to its clients: tax analysis, audit documentation, advisory reports, and legal review. This is not a productivity tool deployed for internal KPMG staff. It is an AI layer baked into the deliverables that KPMG's clients receive. KPMG's client base includes a significant fraction of the Fortune 500, global financial institutions, and sovereign entities — which means Anthropic's reach extends far beyond the 276,000 KPMG employees to the organizations those employees serve.

Question 2

How many employees is PwC training on Claude, and what does the PwC-Anthropic partnership involve?

Accepted Answer

PwC's expanded partnership with Anthropic involves training and certifying 30,000 PwC professionals on Claude, with deployment starting in U.S. teams and expanding globally. The partnership goes beyond simple tooling: PwC will use Claude Code and Cowork — Anthropic's AI-native development and collaboration tools — and will establish a joint Center of Excellence with Anthropic to develop enterprise AI practices. PwC is also using Claude to build technology, execute deals, and reinvent enterprise functions for clients, which means Claude is embedded in the consulting work PwC delivers. The joint Center of Excellence will develop training programs, best practices, and methodology for deploying Claude across enterprise environments — effectively making PwC a co-developer and distributor of Anthropic's enterprise playbook. This is a materially different arrangement than a typical enterprise software license: PwC is not just consuming Anthropic's product, it is helping build and certify the methodology for deploying it at scale inside other organizations.

Question 3

What does the KPMG-Anthropic alliance mean for Salesforce, SAP, and other enterprise software vendors?

Accepted Answer

The Big Four becoming AI delivery channels is a significant distribution threat to traditional enterprise software vendors, though the dynamic is more complex than simple displacement. The risk for vendors like Salesforce and SAP is not that Claude replaces their software — it is that Claude becomes the interface through which users interact with their software, while Anthropic's platform captures the value of that interaction. A KPMG auditor using Claude to analyze financial data is likely pulling that data from a system of record that remains Salesforce CRM or SAP ERP. The underlying data assets don't move. What moves is the locus of user value — from the system of record's native interface to the AI layer on top of it. For enterprise software vendors, the strategic question is whether to compete with Claude's position as the interface layer or to embrace it and ensure their data is optimally accessible to LLM workflows. The vendors who invest in LLM-native APIs, deep integrations with Anthropic's platform, and agent-accessible data structures will benefit from the Big Four deployment wave. The vendors who treat AI as a feature to bolt onto their existing interface will face the slow erosion of the workflows that define their retention.

Question 4

Why are the Big Four professional services firms leading enterprise AI deployment in 2026?

Accepted Answer

The Big Four occupy a structurally unique position in enterprise AI adoption for three reasons. First, they already have trusted relationships inside every major enterprise organization — they are the advisors who approve financial statements, design transformation programs, and advise on regulatory compliance. That existing trust dramatically reduces the adoption friction that AI products typically face in enterprise environments. When KPMG recommends and deploys a tool as part of its client engagement, that tool bypasses the standard procurement cycle, IT review, and change management process that slows direct AI adoption. Second, the Big Four have the professional services model that naturally monetizes AI-enhanced output: they bill for advice and analysis, not for software. Embedding Claude into their workflow increases the quality and speed of their output without changing their revenue model, making adoption incentives extremely high. Third, they have the scale and institutional credibility to invest in training, methodology, and change management at levels that most enterprises cannot afford to do internally for experimental AI tools.

Question 5

What is Claude's dreaming feature and what results have enterprise deployments shown?

Accepted Answer

Claude's dreaming feature, shipped by Anthropic at the Code with Claude 2026 conference on May 6, extends Claude Managed Agents' memory capabilities by reviewing past sessions to find patterns and helping agents self-improve over time. Dreaming is a scheduled process that reviews agent sessions and memory stores, extracts patterns, and curates memories so that agents improve incrementally with each deployment. The feature addresses the single most common enterprise complaint about AI agents: they don't learn from their work. Early production results are significant. Legal AI company Harvey reported that task completion rates increased approximately 6x after implementing dreaming — meaning agents that previously required human intervention to complete complex tasks now complete them autonomously at six times the previous rate. Medical document review company Wisedocs cut document review time by 50%. Netflix is using multi-agent orchestration, another feature released simultaneously with dreaming, to process logs from hundreds of simultaneous builds. These are not lab results — they are production metrics from companies using Claude Managed Agents in real workflows at commercial scale.