For years, advisory services stood as a bulwark against the relentless march of automation, a place where accountants could seek refuge as software took over more and more of the routine, compliance-based processes that, until recently, defined much of their jobs. Sure, the conventional wisdom holds that computers can now easily crunch the numbers — but it still takes humans to interpret what those numbers mean and to communicate that meaning to the client. However, as artificial intelligence continues to evolve, this conventional wisdom is increasingly challenged.

The truth is that while advisory services still stand as that bulwark, AI has begun nibbling at their foundations as it moves past updating journal entries and into generating the same kinds of data-driven insights and analyses that, before, had only been offered by human advisors. For better or worse, advisory is no longer the safe haven it once was. In the face of these changes, it is no longer enough to simply shift to advisory, as certain areas are already being transformed by AI. Instead, accountants must be smart about which areas of advisory they choose.

A great example of an area where this transformation is nearly complete is financial planning and analysis, according to Joe Woodard, head of accounting and business coaching firm Woodard. While there may once have been a time when a firm could sustain itself on analytics alone, those days are long past, as even public AI models are now capable of analyzing mountains of data in mere minutes, and writing reports on their findings in mere seconds.

“AI can do FP&A, and it does it well. A lot of people were initially disillusioned with ChatGPT because it couldn’t manage attachments and would guess when it didn’t know an answer. But now, it no longer plays the guessing game — it can absorb documents and do so securely, especially in enterprise or team editions,” said Woodard.

As an example, Woodard asked ChatGPT-4o, “If I have a gross profit margin of 60% and annual revenue of $15 million, what’s the ideal headcount for accounting associates, partners and reviewers?” He said the AI broke the problem down in real time — assessing firm dynamics, considering average pay rates, benchmarking professional service models, and calculating a typical partner-to-staff ratio. It even separated associates from other billable staff, factored in administrative and support personnel (“which I didn’t even ask for,” Woodard noted), and ultimately estimated a headcount of 85 to 95. And after doing all that, it provided firm strategy recommendations — niche practice areas, geographic market considerations, outsourcing, automation, and technology investments — all in about 45 seconds.

“So yes, AI is extremely deep and powerful,” Woodard said.

Randy Johnston, executive vice president of accounting training and education firm K2 Enterprises, said that he has seen this as well. For a long time, the kind of work that FP&A advisors did required a great deal of effort and time, but technological advancements mean it’s become much easier and faster to analyze a financial situation.

“The amount of time it takes to get high-value advice for clients in an advisory capacity has been dramatically reduced. If you’re doing pure advisory work that involves strategic analysis, historically, a lot of that was done by manually researching and Googling around. Now, you can use AI to get far better summary results and reference materials. I actually think Microsoft’s Copilot 365 does a great job — it will write the summary and provide the links. Does it find everything? No. But does it generate quick insights? Yes,” he said.

(Read more: “AI in advisory: What work is at risk?”)

Overall, the advisory services that are most at risk of disruption are — like compliance services — those that rely on relatively mechanical, step-by-step processes, which AI excels at. This also includes those that rely primarily on financial modeling, as Johnston says professionals no longer need to spend hours building custom models in Excel when they can now run those same figures through AI “and it does a much better job.”

“You still have to check the results, but it’s far faster than starting from scratch,” he said.

In contrast, Woodard said that operational finance roles, such as controllership, are relatively safe for now. He noted that many areas of corporate finance have been effectively automated away, but actual financial leadership positions, he believes, “will endure for the foreseeable future.”

While AI agents have made stunning advances in just a short time, even these semiautonomous bots are still unable to handle the vast number of responsibilities of a competent finance leader. A controller at a $2 million company, for example, needs to juggle financial oversight, operational problem-solving, team management, compliance enforcement and more, all of which requires dynamic, big-picture thinking that AI, for now, lacks.

“Bookkeepers who report to controllers will see their roles largely automated. Reviewers who check books for controllers will also be replaced by AI-driven analytics. FP&A professionals — who primarily compile financial reports — are at high risk. But controllers and CFOs will remain essential. The controller must operate within the company’s day-to-day reality, and the CFO must engage in high-level strategy, investment negotiations, and executive decision-making. These are operational, relationship-based roles that AI cannot replicate,” said Woodard.

Staying relevant

Woodard cautioned, though, that it’s not so much about the advisory areas themselves but, rather, how accounting firms perform them.

He said the biggest mistake he sees firms make is equating advisory with analytics, saying that if their definition of advisory is just building dashboards and explaining financial reports, then they are ripe for AI disruption. On the other hand, if their definition of advisory remains human-driven and client-centric in a way that allows the professional to understand the full context of their client’s situation, ideally based on a years-long relationship, to guide actionable financial decisions, then even FP&A can be fruitful.

“Financial analytics itself — the science of it — is easily and already being displaced by AI. For a human advisor to stay relevant, they must contextualize knowledge within a relationship with the client, understand operational complexities, and inject wisdom. AI cannot be wise; it can only be analytical … If you’re building a practice around delivering financial insights — essentially just interpreting numbers — AI will disrupt that,” he said.

Establishing such relationships and demonstrating credibility is a process that can take a long time, but Johnston said it can be helped through specializing in particular industries. For example, if a firm specializes in utilities and has its own internal data (perhaps indexed by AI) to give meaning and context to events within that sector, “that’s a real game-changer” as many models rely on public data that is not easily accessible to AIs.

“If you have expertise in any industry — utilities, for example — and you have legacy documents that can be indexed, that data is far more valuable than almost any public data available. Public data is useful, but having private data that has already undergone expert analysis is far more powerful,” said Johnston.

However, there is also the matter of educating clients as to why a human professional is still valuable. While accountants know that AI cannot do their entire job, media hype and technology misconceptions can sometimes lead people to believe that it can.

Even among those who already have an accountant, Woodard said that he has seen clients using AI as their first point of contact and only contacting their human professional for a second opinion. (See sidebar, page 8.)

“This is how AI is undercutting the value proposition of accountants. CPAs are increasingly becoming a second opinion rather than the first source of expertise,” said Woodard.

Relationships are the key to avoiding this. Accountants not only should be working on establishing and maintaining longstanding client relationships, they should be acting proactively to keep them informed of new developments that might affect them — so, rather than waiting for the client to call them, whether as the first or second opinion, the accountant should call the client.

Johnston, though, said that another somewhat unintuitive response might be to lean even further into these routine transactional services by leveraging AI to automate these tasks at a large scale, while still offering high-value advisory. However, one way or another, he said it ultimately comes down to keeping the client at the center.

“The key is staying client-centric. The accountants who remain first points of contact for clients — rather than second opinions after AI — will thrive. AI will become embedded in tools accountants already use, making it more invisible over time. But the accountants who don’t leverage AI will be replaced by those who do — especially offshore professionals using AI more aggressively,” he said.