The transformation in auditing has a name: EY AI agents
The transformation in the world of auditing is happening right now, and it has a name: AI agents.
There is a classic paradox in this profession that every auditor knows well: you learn by doing the same thing over and over again. It is through repetition that knowledge solidifies, that trained eyes start to spot what others cannot see. That trainee who spends hours checking spreadsheets, reconciling accounts, and reviewing documents is not just performing mechanical tasks — they are building a foundation of perception that will support their entire career. It is a slow process, sometimes frustrating, but historically necessary.
But what happens when artificial intelligence starts taking over exactly that repetitive work? That is where things get interesting 👀
EY, one of the largest accounting and consulting firms in the world, is facing this dilemma head-on. Instead of ignoring the tension between the traditional learning model and the new reality of AI, the company decided to redesign the way its new professionals are trained, while simultaneously launching a global AI agents framework for its 130,000 auditors around the planet.
The announcement was made on Tuesday and involves a multi-agent system embedded in EY Canvas, the firm’s assurance platform. The goal is ambitious: to have 100% of audit activities supported by agents by 2028, according to Marc Jeschonneck, EY’s global assurance transformation leader, in an interview with Business Insider. But what does this actually mean in practice for someone just starting their career? And will this bet really pay off, or are we looking at yet another tech promise that sounds better on paper than in real life? Let’s dive into all of it 🚀
What are EY’s AI agents and how do they work
Before understanding the impact, it is worth explaining what exactly EY is putting on the field. The AI agents developed by the company are not simple chatbots or autocomplete tools. They are autonomous systems capable of executing complex sequences of tasks within the audit process — such as researching and summarizing documentation, automating administrative tasks, and flagging points of attention for human auditors. All of this in an integrated, continuous manner, embedded directly in the EY Canvas platform that auditors already use every day.
The model works with multiple specialized agents operating together. The initial launch includes a main assistant accompanied by three other agents focused on document search and summarization and administrative task automation. In total, there are about 20 modular capabilities, but that number multiplies depending on the data exposed to the system and how different capabilities are combined, as Jeschonneck explained.
Two additional agents are expected to be released soon: one that will review auditors’ work papers and suggest improvements, and another focused on reconciliation documentation — cross-referencing invoices and other records with audit samples. This distributed architecture allows tasks that previously consumed entire days of manual work to be completed in hours, with a much higher level of consistency than what a human team operating under deadline pressure can achieve.
One point Jeschonneck made sure to highlight is that, unlike tools such as Copilot that require the user to manually upload files, EY’s agentic system works as an integrated end-to-end solution. The agents already have access to relevant data within the platform, eliminating intermediate steps that fragment the experience and reduce efficiency.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
The number of agents is not what matters
A discussion that has taken over the professional services sector in recent months is about agent count. Earlier this year, McKinsey CEO Bob Sternfels stated that his consulting firm already had 25,000 agents deployed. Numbers like that grab attention, but Jeschonneck has a very different take on this numerical race.
According to him, measuring the number of agents is not the right way to evaluate success. In fact, it could even be a sign of a problem. In the executive’s own words, if someone builds thousands of agents, they probably did not understand how the technology works. EY’s philosophy is that a smaller, well-orchestrated set of agents with modular and integrated capabilities delivers more value than an uncoordinated army of isolated bots.
This approach reflects a level of technical maturity that not all organizations have developed yet. By opting for a modular framework instead of creating specific agents for every micro-process, EY can scale the solution far more efficiently and maintain governance over what each agent does, how it does it, and what data it operates with. In large-scale audits, this kind of control is not optional — it is mandatory.
The learning dilemma when AI handles the repetitive work
Here is the central knot of this whole story. Historically, junior professionals at audit firms like EY learned the profession literally through repetition. You showed up as a trainee, received a stack of documents, and spent weeks checking numbers. It was tedious, but it was formative. Every error found, every inconsistency detected, every pattern recognized in a data sequence — all of it accumulated in the professional’s practical memory and built what experienced auditors call audit sense: that almost intuitive ability to notice when something is off even without being able to explain exactly why.
The problem is that if AI agents start executing exactly those tasks, the trainee of 2025 and beyond simply will not have the same exposure that professionals from previous generations had. And that raises an uncomfortable question: how do you develop judgment without accumulated experience? How does someone learn to supervise a smart agent on a task they have never performed manually?
Marc Jeschonneck acknowledged this challenge directly. According to him, auditors will need to have a considerable level of experience to effectively review what reconciliation agents produce. And for younger professionals, this means the entry point into the career may not get easier right away.
This is not a problem EY invented — it is a tension that any industry faces when automation advances into activities that used to be the gateway into the workforce. But the fact that it is a universal tension does not mean it is easy to solve.
EY’s new approach to training
EY’s answer to this challenge involves a complete reengineering of its professional development programs. The company plans to train its new hires in a fundamentally different way from what was done before.
Instead of learning on the job by repeating the same task across multiple engagements, new hires will work with realistic audit scenarios, supported by adaptive learning tools and short videos embedded directly in the platform itself. It is a learning approach based on analysis and supervision, not on direct execution — which represents a pretty radical shift in the profession’s pedagogical model.
Jeschonneck sees this change as something positive in the long run, even while acknowledging that the transition will be tough. Qualified professionals coming out of college do not want to spend their time on administrative tasks, and according to the executive, they do not need to do certain things a thousand or ten thousand times before finally understanding how it works.
The bet is that this new generation of auditors, trained from the start to work in partnership with AI agents, will develop a different kind of competency: less based on mechanical repetition and more oriented toward critical thinking, intelligent supervision, and decision-making in complex environments. Whether this will work at the necessary scale is still an open question, but the direction chosen shows that EY is taking the problem seriously.
What changes for auditors in practice
For those already in the career, the arrival of AI agents within EY represents a significant repositioning of what it means to be a high-performing auditor. The technical skills of manual checking, which were the base of the competency pyramid, are losing ground to capabilities like critical thinking, risk communication, complex data interpretation, and managing the relationship with intelligent agents. This is not a surface-level change — it is a redefinition of what the market will value in the coming years within this profession.
In day-to-day practice, the transformation should translate into workflows where the auditor begins their day not by opening spreadsheets, but by reviewing reports consolidated by agents, prioritizing the highest-risk flags, and dedicating their attention to the points that truly require human judgment. The idea is that the time freed up by automation gets reinvested in deeper analyses, richer conversations with clients, and a more consultative rather than operational posture.
If this works as planned, the result for EY clients would be an audit that is more thorough, faster, and with a level of data coverage that simply was not feasible under the 100% human model.
The bigger picture: the Big Four and the AI race
Integrating AI agents into auditing and consulting is already part of the Big Four business model in 2026. These firms win contracts when they can prove the value of AI at enterprise scale — and what better demonstration than applying the technology to their own workforce?
This movement is not exclusive to EY. This month, KPMG revealed to Business Insider that it is piloting a program where tax professionals use vibe coding to automate tax and compliance processes. These are different approaches toward the same goal: reducing operational friction and freeing up professionals for higher-value activities.
But while industry leaders invest billions in AI, professions tied to business, finance, and consulting frequently appear on lists of roles most exposed to automation. Hiring for certain positions, such as management consultants, is declining. And PwC, another Big Four giant, cut a third of all entry-level hires in the United States over the next three years, as Business Insider reported in August.
Nobody should worry, says EY
Given this scenario, the inevitable question is: will EY cut people? Jeschonneck said the company does not plan to reduce hiring.
He acknowledges that the number of people needed for the traditional audit of historical financial statements will likely decrease. But the firm’s goal is to maintain the same headcount, redirecting capacity to handle the technology side and the increasingly complex regulatory and client demands.
In the executive’s own words: nobody should worry about starting a career in the accounting world right now. The company will need people with the institutional knowledge built over more than a hundred years in the field to make the technology truly relevant.
It is an optimistic statement, and it is worth noting that it comes from someone directly responsible for selling this vision inside and outside the organization. But the argument holds logic: financial sector regulation is becoming more complex, clients are operating in increasingly global and interconnected environments, and the demand for analyses that go beyond the traditional checklist is growing. If AI agents free auditors from operational work, the space for consultative and strategic roles could indeed expand.
Why this initiative matters beyond EY
What EY is doing is not an isolated case of a big company adopting new technology. It is, in fact, a real-world experiment at scale on how highly regulated, trust-based sectors will adapt to the era of AI agents. Auditing is a field where mistakes carry serious consequences — undetected fraud, incorrectly approved financial statements, ignored systemic risks. That is precisely why the way EY balances automation and human accountability will serve as a reference, or a warning, for other sectors watching this journey closely.
Beyond that, the learning model the company is developing for its trainees could spark a broader discussion about how educational institutions and companies need to rethink professional development in areas where AI is already here to stay. This is no longer about preparing for some distant future — it is an adaptation happening right now, in real time, inside one of the largest professional organizations on the planet.
Of course, the success of this transformation depends on a number of variables still being tested. The quality of the data feeding the agents, the ability to integrate with client systems, the internal adoption curve, and regulatory acceptance of AI use in formal audit processes are all factors that will determine whether the 2028 target is realistic or overly optimistic.
What is clear is that EY is not simply automating tasks. It is trying to redefine what it means to be an auditor in the 21st century, using AI agents not as replacements, but as amplifiers of human potential. Whether this vision will materialize as planned is still too early to say. But the bet has been placed, the investment is real, and the world of auditing will never be quite the same 🤖
