EY gives auditors AI agents

EY just moved artificial intelligence from the side desk to the center of the audit. On April 7, 2026, the firm said it had rolled out “agentic” artificial intelligence across its assurance practice, putting it into the daily workflow of 130,000 assurance professionals working on 160,000 audit engagements in more than 150 countries and territories. (ey.com) An audit is the process where accountants test whether a company’s financial statements match the underlying records. In big companies, that means checking huge volumes of journal entries, invoices, contracts, and disclosures without reading every line by hand. (pcaobus.org) For years, most firms used artificial intelligence more like a search box than a coworker. The Public Company Accounting Oversight Board said in July 2024 that generative artificial intelligence in audits was still focused mainly on administrative and research tasks, even though firms were exploring broader use in planning and fieldwork. (pcaobus.org) EY started laying the track a year earlier. On April 9, 2025, it said a four-year, $1 billion assurance technology program was adding tools that could search and summarize accounting and auditing guidance inside live audit workflows for more than 160,000 engagements. (ey.com) The new step is different because “agentic” artificial intelligence does not just answer questions. EY says the system can orchestrate multi-step tasks inside EY Canvas, the firm’s global audit platform, which already processes more than 1.4 trillion lines of journal entry data each year. (ey.com) Think of the old tool as a smart library clerk and the new one as a junior coordinator that can line up the books, flag the missing pages, and route the file to the next person. EY says the agents are integrated with Microsoft Azure, Microsoft Foundry, and Microsoft Fabric, so the system sits inside the firm’s existing technology stack rather than as a separate chatbot. (ey.com) That changes the shape of entry-level audit work first. Marc Jeschonneck, EY’s global assurance digital leader, told Business Insider that the learning curve for junior staff will be steep even if the tools make parts of the job much easier over time. (businessinsider.com) The pressure point is simple: junior auditors traditionally learned by doing repetitive testing, tying numbers across documents, and digging through guidance one step at a time. If software takes more of that first-pass work, firms have to decide which tasks still train people to spot errors, ask skeptical questions, and know when the machine is wrong. (pcaobus.org) Regulators are already looking at exactly those risks. The Public Company Accounting Oversight Board said firms using generative artificial intelligence still need strong supervision and controls around limits such as data privacy, security, and reliability. (pcaobus.org) EY is trying to answer that by keeping the official line human-first. Its April 2026 announcement says the rollout is meant to reduce administrative burden, improve risk assessments, and still preserve human judgment, professional skepticism, and insight, with full end-to-end audit support expected by 2028. (ey.com) So the real story is not that auditors got a new assistant. It is that one of the Big Four accounting networks is now treating artificial intelligence as part of the production system for audits at global scale, which means every rival firm, regulator, and first-year associate now has to rethink what audit work looks like from here. (ey.com)