EY Deploys 150 AI Agents for 80,000 Auditors as Big Four Automate Compliance Core
With banking clients collectively cutting over 10,000 jobs in Q2 2026, the world's four largest auditing firms are deploying AI agents in work that has trained generations of accountants.
Scope of Investment
Ernst & Young has deployed 150 artificial intelligence agents for 80,000 tax professionals worldwide. According to EY itself, the agents process over 3 million tax compliance cases per year and were developed with Nvidia on the firm's own server infrastructure, not on public cloud, due to client data confidentiality requirements.
KPMG has committed $2 billion over five years to AI through a partnership with Microsoft, with a stated goal of $12 billion in additional revenue from AI-enabled services. Deloitte operates Zora, a system of agents developed with Nvidia for automating invoice processing and analyzing financial trends. PwC has about 200,000 internal users on the ChatPwC platform and uses GL.ai for automated review of accounting entries.
The foundation of an external audit of large corporations consists of high-volume, repeatable tasks: reconciliation of trial balances, confirmation of balances with counterparts, and testing internal controls on samples of transactions. In the Big Four, this work is performed by professionals with up to three years of experience, allocated in Acceleration Centers concentrated in India (Bangalore, Hyderabad, Pune, Chennai) and regional centers in Eastern Europe and the Philippines. These professionals represent the backbone of an industry that, when combined across the four firms, employs over 1 million people worldwide.
The Signal from Banking Clients
In the Q2 2026 results reported between July 13 and 15, the largest U.S. banks showed record profits combined with workforce reductions. Wells Fargo ended Q2 with 197,000 employees, a decrease of 3,500 from the previous quarter and 15,000 from the same period last year. CFO Mike Santomassimo stated in a conference call with analysts that the company expects to operate with fewer employees than currently. Collectively, Bank of America, Wells Fargo, Citigroup, Goldman Sachs, and Morgan Stanley reduced their headcount by over 10,000 positions in the second quarter.
These banks are audited by the Big Four. When a bank cuts 3,000 positions in the middle office over a quarter, the scope of the audit work on these operations shrinks. Firms charge based on the volume of testing, bank confirmations, and reconciliations reviewed; fewer operations mean less revenue unless the firm can deliver more analysis for fewer professional hours. AI agents resolve this exact equation.
The Regulator and the Employment Argument
The PCAOB (Public Company Accounting Oversight Board), U.S. regulator of audits of listed companies, requires that every audit conclusion be signed by a licensed auditor and announced in 2026 that it is developing specific standards for the use of AI in audit evidence. Until these standards are in place, firms are deploying agents in support tasks, not in the evidence and conclusion stages. An agent that processes 100,000 reconciliations does not replace the partner who signs the report; it replaces the 15 associates who prepared the file that the partner reviewed.
The argument that automation "will create higher-level jobs" for displaced professionals is what each of the four firms repeats in official communications. The question that this argument avoids is arithmetic: if a firm goes from 15 to 3 professionals to audit a client of the same size, the 12 excess professionals need to find demand elsewhere. Audit firms allocate professionals where there is revenue, not based on projected future demand.
India, Brazil, and the Speed of Automation
In India, where the Big Four concentrate their highest-volume Global Delivery Centers in Bangalore, Hyderabad, Pune, and Chennai, the automation of basic audit tasks is advancing faster than in other markets. The argument is straightforward: an agent that executes reconciliations at a marginal cost close to zero directly competes with an entry-level professional who, even in Indian centers, represents a significant fixed cost for firms.
In Brazil, the Big Four operate Shared Services Centers in São Paulo that serve both local clients and support operations for global clients. The regulations of the CFC (Federal Accounting Council) do not explicitly address the use of AI agents in audit evidence, which adds regulatory caution to local deployments. Until the CFC issues a statement, the pace of automation in the Brazilian market is expected to lag behind that in Indian centers, creating a window for local teams to reposition skills before automation arrives at the same speed.