All EY assurance professionals will now have access to AI agents

Big Four firm Ernst and Young announced the global embedding of enterprise-scale agentic AI in its assurance engagements, meaning that all audits will now use the technology in its firms worldwide. 

Processing Content

Specifically, the firm is directly embedding a new multi-agent framework, integrated with Microsoft Azure, Microsoft Foundry and Microsoft Fabric, into EY Canvas, its global assurance technology platform which has since been enhanced to support a number of AI use cases for auditors. This overall agentic integration immediately embeds AI in all phases of the audit globally, which is meant to tailor workflows to engagements, streamline processes, provide additional insights, and generally improve the audit experience. 

On a practical level, this includes new capacities like enhanced project management and administrative task automation, such as such as assigning tasks, requesting information or drafting review notes, as well as summarizing audit documentation, such as all conclusions documented related to one specific matter throughout the audit file. This is in addition to search and summarisation of relevant accounting and auditing guidance. Within this year, they also expect the platform to do things like document reconciliations to external evidence and assist teams with drafting standard workpapers. Marc Jeshonneck, global assurance transformation leader with EY, said it is not just about accelerating existing audit steps but fundamentally changing how audits are executed end-to-end. 

“The key design principle was to embed AI directly into the audit platform, so our auditors aren’t having to navigate several separate tools, move around files, repetitively provide context in long prompts or switch between applications. It’s one assistant, built into the audit platform itself, using a multi‑agent framework that orchestrates the underlying AI capabilities and thereby seamlessly takes care of routine, repetitive and administrative steps that sit behind an audit. This allows auditors to focus on risks including areas requiring professional judgement and it elevates their experience. That approach embeds the technology to adapt to the audit workflow, rather than forcing auditors to adapt to the technology,” he said in an email. 

This new approach to the audit process will also require new training on how best to apply it. To this end, EY also announced a global training program to further upskill all of its global audit and technology risk professionals this year. The structured program will include immersive and in-person learning and will be continuously updated in line with developments in regulation, technology and methodology. 

Jeshonneck, though, said the training is not to turn auditors into prompt engineers and data scientists, as the platform is designed to be intuitive, with additional support available when needed (e.g. embedded short videos explaining platform features.) While, yes, it will include training on the technology itself and how it is applied, more of the focus will be on how to use AI responsibly in order to augment the skills and judgement of auditors as capacity shifts from legacy tasks towards higher value work. 

“Extensive work has been carried out to redesign training, including virtual and in-person events, self-service materials and well-equipped coaching and expert networks across the globe. All the training will continuously be updated as the capabilities expand, and we will make use of technology to deliver training such as by using simulations and adaptive learning,” he said. 

He described this new development as the latest in a journey begun four years ago to create what he said would be the next generation assurance technology platform. This release, he said, is the result of several years worth of development, testing and feedback from real world use, with EY acting, effectively, as “client zero” for all these new capacities. 

“We were able to determine which, how and where agents genuinely add value; the level of required training support, and human review; and how to design and operate controls so outputs are transparent, reliable and reviewable. These are just some examples of us following EY’s nine principles of responsible AI,” he said. 

And they are not done. Ultimately, it is expected to support all end-to-end audit activities by 2028. When asked what the human would do at that point, he said their work would evolve but ultimately they would remain intimately involved with the audit process. 

“Ultimately, the role of humans is elevated, as they will still own decisions, build on their experience, form their expectations, review outputs, challenge anomalies and lead client conversations. We know that the work auditors do has to evolve given the changing technological and regulatory landscape, so demand for early career professionals with accounting knowledge remains, as we combine it with the power of technology. Furthermore, the complexities associated with providing assurance on AI are dynamic and still emerging, creating new – not less – demands for audit teams,” he said.