A look at PwC’s ‘next generation’ audit

Want to feel old? ChatGPT was released this month…three years ago. Yes, only three. And, yes, we know: It feels like a lifetime.

In those three short years, finance teams have rushed to invest in AI and pilot tools that, they hope, streamline workflows. And while ROI on many AI investments still proves…let’s be nice and call it elusive…we’re finally getting a much better sense of exactly how and where AI can assist with key finance functions, including the audit process.

On October 27, PwC offered reporters a look at what the firm is calling “the next generation audit,” which currently consists of a slew of AI assurance tools either currently in use or in the pilot phase.

Expanding the toolkit. “It’s a transformation of our entire audit practice,” Kyle Maryanski, US assurance partner and leader of the Next Generation Audit for PwC, told reporters during an Oct. 27 roundtable. “Yes, it’s rooted in technology. There’s a big base technology to that. But it also is talking about our people, process, and methodology and how we’re going to do things.”

The “north star,” in Maryanski’s telling, is an “AI-integrated, end-to-end audit platform, which is where we’re going to get at the end of the day.” He said the global scale of the effort will eventually give the firm “consistency in how we collaborate and how we share information across countries, across accounting standards.” The upside of the gradual rollout is that as the firm debuts tools “in pieces and phases, we’re getting benefits immediately along the way.”

And there are lots of tools to roll out: Among the eight PwC introduced at its roundtable, there’s ChatNational, a GenAI chatbot and research tool that’s been in use for a year. It connects to PwC’s global research platform, giving audit and attest professionals answers to questions relating to accounting, auditing, and reporting guidance.

There’s also Evidence Match, which automatically matches information between two records, to allow PwC professionals to respond to and resolve open items faster.

“If you think of an audit, we spend a lot of time gathering information, getting it from the client, matching it to systems, tracing and tying it in, and it goes through a lot of manual work and effort, and then ultimately you get to analyzing the anomalies or the differences, which is really where the real value is added,” Maryanski noted. “This is in accounts payable, accounts receivable, fixed assets, cash. Pretty much all the areas where you do substantive testing, you’re going to use something like Evidence Match.”

All in the timing. On the topic of timelines, an end-to-end, AI-integrated audit process will be “in place for calendar 2026 audits,” Shawn Panson, PwC’s US assurance transformation leader, told reporters. “That’s our hope.”

PwC has invested $1 billion in its Next Generation Audit platform, as well as $1 billion into an initial three-year relationship with OpenAI, first announced in 2023. Not exactly chump change—but ROI is largely tool-specific, according to Jennifer Kosar, PwC’s US assurance AI leader.

“We are looking at different measures of ROI or measures of success on these technologies,” Kosar said. “Accuracy, of course, is important when that is the important measure,” citing tools like Evidence Match, where there’s a “binary yes or no.”

For many tools, however, the bigger measure of success is: “How helpful was it? Did it accelerate the process? Did it make it go faster? These are different measures that are equally, if sometimes not more important, depending on the particular solution you’re trying to build,” she explained.

And if things stop working, the firm is ready to pivot away from a given tool. “AI technologies in particular may solve a particular pain point, and then two to three years later, that’s not the pain point anymore,” she noted. “We absolutely have that baked into the process to measure both usefulness and performance over time.”

When asked if these tools would shrink the minimum viable number of staff on an audit, PwC stressed the need for human oversight. “AI systems are not traditional automation,” Kosar said. “They require a significant amount of expertise to continue to run over time, so to continue to build, to test, to monitor, the subject matter expertise is significant, and so some of these people are deployed to those roles as well.”

But while employees have received extensive training on these new tools, expectations for entry level auditors may be changing.

“We’re definitely looking for different skills, as well as engineers and stuff that we’re now hiring that we would have never hired into the assurance practice before,” Panson said. “Not all those people will definitely be dual track CPAs [with an AI concentration] but we’re certainly looking for more of the dual track so people can serve both those purposes. It’s early days, but we’re certainly starting to see a shift.”