Outlook 2026: Agentic AI Reaches the Tipping Point in Tax and Accounting Firms

AI-powered firms are closing books faster, reallocating staff time to higher-value work, and widening the competitive gap with slower adopters.mo

By CPA Trendlines Research
Cornerstone Report

As artificial intelligence moves from buzzword to business imperative, CPA firms across the U.S. are quietly beginning to deploy generative AI assistants, machine learning tools, and “agentic” AI platforms to automate audits, prepare taxes, and provide financial insights.

With the astonishing surge in AI adoption, firm leaders say we’ve reached a tipping point where those not investing in AI risk being left behind.

In this Cornerstone Report, accounting firms show how they are leveraging AI to transform their operations, the benefits and challenges they are encountering, and what it all means for the future of the profession, including:

• Why AI adoption in CPA firms has hit a tipping point
• How agentic AI is transforming tax, audit, and advisory work
• The real productivity, ROI, and revenue gains firms are reporting
• What AI means for staffing, skills, and firm economics
• The risks, governance challenges, and regulatory implications ahead
• How firm leaders can deploy AI without falling behind

AI Adoption Surges from Experimentation to Mainstream

Just a year ago, AI in accounting was largely experimental – confined to pilot projects at a few forward-thinking firms. Now, new research indicates a majority of firms are jumping on board. According to the 2025 Wolters Kluwer Future Ready Accountant report, AI adoption in accounting firms leapt from 9% in 2024 to 41% in 2025.

“This signals a shift from cautious testing to confident integration across firm operations,” says Jason Marx, CEO of Wolters Kluwer Tax & Accounting. In the same study, 77% of firms said they plan to increase AI investment, and 35% are already using AI tools daily. In other words, more than one-third of firms have woven AI into everyday workflows for tasks like tax prep, research, and client service – a remarkable change in such a short time.

This rapid rise reflects a broader evolution in mindset. Not long ago, many accountants feared AI as a potential job-killer or an unproven gimmick. Now, the narrative has flipped to opportunity and urgency.

“Investment in technology, particularly AI and automation, is accelerating… these shifts signal a broader redefinition of firm purpose and strategy,” Marx notes of the global trends . Practically, mid-sized CPA firms (typically 50 to 500 staff) are seeing their larger competitors – including the Big Four and top 10 firms – aggressively roll out AI, and they know they must follow suit or risk falling behind.

The Big Four have indeed led the way in AI adoption, often developing proprietary tools. Deloitte, for example, has built generative AI into its audit platform to automatically review documents and highlight issues . EY launched a global AI platform that embeds AI across services from tax to consulting . PwC developers are using custom AI software to automate code writing and data analysis, seeing 20–50% productivity gains in those internal tasks . And KPMG has rolled out a “Trusted AI” framework to help ensure AI is used ethically, both in-house and for clients .

Those are big-budget initiatives, but the innovations don’t stop at the Big Four. Mid-tier and regional firms are also adopting AI at a rapid clip. They may not build their own AI from scratch, but they’re eagerly implementing commercial AI-powered solutions. “Smaller accounting firms…are also embracing AI to stay competitive and offer better services,” notes a Thomson Reuters analysis . In a recent Thomson Reuters survey of tax firms of all sizes, 44% of firms using or planning to use generative AI said they use it daily (or multiple times a day), and another 29% use it at least weekly . This heavy usage underscores that AI has moved beyond novelty; it’s becoming a routine part of the work.

Perhaps the clearest evidence that AI is now mainstream is the bottom-line impact being reported. Firms with deeper tech integration are outperforming others. Industry data show that firms making advanced use of AI are achieving faster closes (monthly financial closing 7.5 days faster, on average) , more billable work (one study found an 8.5% reallocation of time from data entry to higher-value tasks ), and higher value services. For example, early adopters have expanded their offerings in areas like advisory and analytics – one global survey found the average firm’s advisory revenue jumped significantly, with 93% of firms now offering advisory services (up from 83% a year prior) as they deploy AI and data to personalize insights .

“AI adoption has more than quadrupled… Firms leveraging AI are working faster, with greater precision and confidence, transforming how strategic decisions are made and how client value is delivered.”

– Jason Marx, CEO, Wolters Kluwer Tax & Accounting

The competitive stakes are high. Thomson Reuters’ 2025 Future of Professionals Report – which surveyed thousands of tax and legal professionals – warns of a growing “AI adoption divide”. Those with a clear AI strategy are 3-4 times more likely to see benefits like revenue growth and efficiency gains than those without a strategy . Alarmingly, the report suggests that firms failing to develop an AI plan now could fall irreparably behind within three years.  In other words, the late adopters may find that by 2028, they cannot easily catch up to tech-enabled competitors who have been compounding productivity gains and attracting tech-savvy clients and talent.

For CPA firms, AI is no longer a “nice-to-have” experiment – it’s becoming a business-critical capability. As we’ll explore, these firms are not just dabbling in AI; many are reengineering their workflows, retraining staff, and fundamentally changing service delivery models around AI. The next sections delve into how exactly they’re doing that, starting with the nuts and bolts of the new AI tools at their disposal.

The New Agentic AI Toolbox: Platforms and Solutions Leading the Charge

Accounting firms today can choose from a rapidly expanding marketplace of AI-driven software tailored for tax and accounting. The major vendors in the profession – including Intuit, Wolters Kluwer, and Thomson Reuters – have each launched cloud-based “agentic AI” platforms in the past 1-2 years, embedding AI capabilities into tax preparation, audit, and firm management software. These platforms are “agentic” in that they can autonomously perform tasks (like data entry or analysis) on behalf of the user, within set parameters. Below, we compare some of the leading solutions mid-sized firms are deploying:

Leading AI-Enabled Tax & Accounting Platforms

Sources: Vendor publications and CPA Trendlines research. Note: Pricing is indicative; actual costs depend on firm size, number of users, and chosen modules. All these platforms are cloud-based and frequently updated, so firms benefit from continuous improvements (but also must adapt to frequent changes). Onboarding often requires dedicating staff time for configuration, training, and process tweaks to fully leverage the AI features.

Each major vendor’s approach has unique strengths, but there are common themes. All of these platforms are cloud-first – which is essential, because cloud computing provides the scalable processing power that modern AI needs and allows real-time data access for distributed teams. They also emphasize integration: Intuit’s products integrate tax prep with QuickBooks data; Wolters Kluwer’s connect tax, audit, and practice management; Thomson Reuters links research tools with compliance software. This integration is crucial – it’s how AI can “see” the full picture of a client’s information and workflow, rather than operating in a silo.

Key AI features across these platforms include:

It’s worth noting that pricing for these advanced platforms can be a significant investment for firms. For instance, Wolters Kluwer’s CCH Axcess with full AI functionality can run into the five or even six figures annually for a larger firm. Many mid-sized practices, however, find a modular approach – maybe starting with an AI tax prep module or an AI audit assistant – more palatable, building up usage (and ROI) over time. The ROI, as we’ll discuss next, often justifies the cost if the tools are utilized fully.

Onfe more point about implementing these tools: training and change management are critical. Firms that succeed with AI dedicate time to train their staff on the new software. Many vendors offer extensive onboarding – Intuit even bundled free access to their suite during beta testing to encourage firms to try it.

But even the best tool will gather dust if accountants don’t trust it or know how to use it. Successful firms often designate an “AI champion” – a tech-savvy team member to become the in-house expert and cheerleader for the new platform, helping colleagues troubleshoot and discover features. When done right, the cultural adoption of the tool follows, and the efficiency gains begin to flow.

Measuring the Payoff: Efficiency, ROI, and Firm Economics in the AI Era

CPA firms are, at the end of the day, businesses – so partners want to know, does this AI stuff actually make us more money or improve margins? Early evidence says yes. The economics of firm performance are visibly shifting in AI-adopting firms, with improvements in productivity and profitability metrics that would have seemed ambitious just a few years ago.

One striking trend is the increase in revenue per employee at firms that leverage AI deeply. Traditionally, many CPA firms have hovered around $150,000 to $200,000 in annual revenue per professional staff. Now, industry observers note that tech-optimized firms are achieving $250,000 to $350,000 (or more) in revenue per employee, thanks in part to automation and the ability for each professional to handle more work.

Automation means an associate or senior can manage a larger client load without working longer hours – the software takes on a chunk of the workload (data processing, initial analysis, etc.), enabling the human to focus on higher-value activities that clients pay premium fees for.

For example, AI-assisted tax preparation has dramatically reduced the time spent per return in many cases. A task that might have taken 5 hours of staff time can sometimes be done in 2 hours with AI, with the machine handling the rote parts and the human doing final adjustments. Multiply that across thousands of returns and you either free up significant hours for other work or you can take on more returns with the same staff. Firms with advanced AI integration report 21% higher billable hours per staff (because they can reallocate time to other billable projects once routine work is automated).

They also report offering more services per client – e.g., upselling clients to advisory projects – leading to 80% increases in premium service revenue in some cases, according to CPA Trendlines Research. When compliance work takes less time, firms don’t just let staff twiddle their thumbs; they can refocus them on consulting, financial planning, or business advisory for the same clients.

Another key metric: speed of service. In auditing and monthly accounting, “time to close” and “time to deliver” are vital metrics. A study by MIT and Stanford researchers found that accountants using an AI-based software saw a notable quality and speed improvement in financial reporting. They achieved a 7.5-day reduction in monthly close time on average, essentially finalizing monthly books in just over two weeks after month-end, whereas those without AI took over three weeks.

Faster turnaround can be a competitive advantage – clients certainly appreciate getting financial statements or tax estimates earlier. Quicker processes also free up calendar time for staff to do more cycles of work in a year. If you can complete a monthly close in, say, 5 days instead of 10, that’s 5 extra days per month the team can devote elsewhere – which might mean taking on an additional client or focusing on analysis rather than reconciliations.

The return on investment on AI projects often comes up in partner meetings. While every firm’s case is different, some benchmark numbers are emerging. Many firms aim for ROI within 18 months or less, and anecdotal reports show this is achievable. For mid-sized adopters, first-year net benefits (either cost savings or additional revenue) in the range of $500K–$900K have been cited. How?

Consider a firm that automates large portions of its individual tax return process: if they prepare 5,000 returns and save just 1 hour on each due to AI, that’s 5,000 hours saved. At a blended rate of $100/hour, that’s $500,000 of labor value. In reality, some returns (especially complex ones) might save multiple hours. Plus, error reduction from AI (catching mistakes before they go out the door) can save money on rework and prevent costly penalties.

There’s also the cost of not investing – which is harder to quantify but very real. If competitor firms are using AI to offer faster service or new analytic insights, they might lure clients away. In that sense, ROI can also be measured in client retention and growth. One survey found 56% of accounting professionals believe a firm’s value proposition suffers if it doesn’t use AI – clients may start to see non-AI-enabled firms as behind the times. Similarly, as we discuss further below, prospective employees factor tech in – so a firm could “lose” by missing out on top talent if it doesn’t modernize.

From a billing model perspective, increased efficiency is prompting some firms to reconsider hourly billing. If a tax return that used to take 4 hours now takes 1, billing purely by the hour would shrink revenue, which clearly isn’t the goal. Instead, many firms are shifting to value-based or fixed-fee pricing for AI-accelerated services. In tax prep, for instance, only about 3% of firms now bill purely by the hour . Most use fixed fees or monthly retainers that reflect the value of the outcome (an accurate return, delivered quickly with advisory insights), not the hours spent. AI makes outcomes better and faster, but clients still value the service and are willing to pay for it; the firm just realizes higher margin because it costs them less time to produce.

To illustrate, a firm might charge $1,000 for a business tax return that used to take 10 hours ($100/hour effectively). If AI support means it now takes 5 hours, the client still pays $1,000 (since that’s the going rate for that service’s value), but the internal cost is halved. That’s not to say firms should gouge clients – rather, they can redirect the freed time to more value-add activities for the client, justifying the fee while delivering extra advice or analysis. In fact, many firms report that by utilizing AI to handle mundane tasks, they can include an “AI review” or extra schedules at no additional cost, which improves quality and client satisfaction. Over time, this can attract more business.

One concrete economic benefit reported is improved realization rates and fewer write-downs on engagements. Projects are more likely to stay within budgeted hours when AI is chipping in, so partners are writing off fewer overages. Also, with AI ensuring more compliance tasks are done right the first time, there’s less back-and-forth and less non-billable cleanup.

Properly harnessed, AI can supercharge a mid-sized firm’s capacity and profitability. It’s like adding semi-autonomous “digital staff” that work tirelessly in the background. That doesn’t mean everything is rosy or automatic – to get those results, firms must manage the transition carefully, invest in training, and sometimes redesign their processes. That’s why some firms not seeing immediate returns may feel frustration. But the data increasingly shows that those who stick with it and align their business model to their new tech capabilities do reap significant rewards.

It’s also important to remember that AI isn’t just about doing the same work faster – it often opens doors to new services and revenue streams. For example, a firm that never offered data analytics or forecasting might now do so because AI tools make it feasible without hiring a PhD in statistics. Some firms are packaging AI-generated insights (like benchmarking a client’s performance against industry data) as new advisory offerings. These can command premium fees. As one managing partner put it, “AI is helping us monetize our know-how in new ways – we can analyze a client’s data and give strategic advice, which they’ll pay more for than a basic compiled financial.”

The economic impact of AI for CPA firms can be transformative. The key metrics – turnaround time, staff productivity, realization, revenue per employee, service mix – all trend positively when AI is thoughtfully implemented.

Talent in the Age of AI: How Roles and Skills Are Evolving

Walk into a mid-sized accounting firm today, and you might not immediately notice that anything is different. The accountants are still at their dual monitors, coffee in hand during busy season, collaborating on client work. But look closer: that tax associate might be using a chatbot to draft an email to a client, the senior auditor might be running data through an AI tool instead of Excel, and the firm’s training coordinator might be scheduling AI ethics training for the team.

The job of an accountant is changing in subtle and not-so-subtle ways, thanks to AI.

Here’s how:

Fewer Routine Roles, More Strategic Roles: AI’s strength is in automating routine, repetitive tasks – the kind often done by entry-level staff. Indeed, we are already seeing a decline in those types of positions. A Stanford study in 2025 found that hiring for entry-level, AI-impacted jobs (like junior accounting roles) fell by 16% in about two years . Many firms aren’t outright laying off staff, but as people leave, they might not replace some of those junior data-entry heavy roles. Instead, they might invest in a data analyst or an advisory specialist. Forecasts from CPA industry consultants project 15–20% fewer entry-level accounting positions by 2027 at firms that fully embrace AI, whereas roles in consulting, strategy, or data analysis could increase by 25%.

In practice, firms are reorganizing their staffing pyramids. Traditionally, a pyramid might have a broad base of junior staff doing the grunt work, feeding up to a narrower point of senior managers/partners. With AI, the pyramid might become more diamond or pillar-shaped: relatively fewer at the very bottom, a strong middle of tech-savvy accountants, and a top that expands into new specialties. For instance, for every 4 tax preparers a firm used to need, maybe now it needs 3, because one “AI-augmented” preparer can do the work of what used to be 1.3 or 1.5 people. But that firm might simultaneously hire an AI specialist or data scientist to work alongside the CPAs in analyzing client data or customizing AI models to firm needs – a role that didn’t exist before.

If there’s one refrain in the industry right now, it’s that CPAs need to upskill. In the Karbon “State of AI in Accounting” survey, only 37% of firms said they are actively investing in AI training for employees – a low number considering the need. Yet those that do invest see dividends: the survey found firms providing AI training were unlocking an extra 40 hours of capacity per employee per year (on average) compared to firms that don’t . Training can range from formal courses on data analytics to lunch-and-learns on how to use the firm’s new AI-based tax software.

The AICPA has also been pushing members to attain “data analytics” and “AI literacy.” In fact, a new CPA specialty credential around technology is being discussed. Many states now count AI and analytics courses toward CPAs’ mandatory continuing education. Forward-thinking firms mandate at least 8-16 hours of AI or tech-focused CPE for their staff annually. Some larger firms even have internal AI certifications – e.g., a bronze/silver/gold level indicating what tools you’re proficient in.

What specific skills are in demand? For starters, data analysis and interpretation. With AI tools spitting out results, someone needs to make sense of them. Accountants with stronger quantitative and analytical backgrounds are prized. Advisory and communication skills have also become more crucial – if AI handles the heavy lifting of number-crunching, the accountant’s value is in interpreting those numbers and communicating impact to clients. So being able to translate an AI-identified trend into actionable business advice is gold. On the audit side, IT audit and understanding of AI algorithms is emerging. For example, auditors now need to consider how to audit a client’s AI system or incorporate AI outputs as audit evidence – that requires knowledge of how AI models work, at least at a conceptual level, to assess reliability.

Recruitment and Retention: The war for talent in accounting has been intense in recent years (even pre-AI). Now, tech adoption is a differentiator for potential hires. Young accountants coming out of college are often digital natives who’ve used AI (like ChatGPT) in school or at home. They expect employers to have modern tools. A striking 76% of accounting graduates say they are more likely to join firms that actively use AI and advanced technologies. This aligns with what firm leaders are observing in interviews – candidates ask about what technology the firm uses, whether they’ll get to learn new software, etc.

Mid-sized firms can actually use this to their advantage against larger competitors: a regional firm that’s heavily invested in tech can attract a star candidate by offering a forward-thinking environment, whereas a Big Four firm that might be perceived (rightly or wrongly) as using armies of staff to grind through work could seem less appealing.

On retention, it’s a similar story. Burnout and overwork have been major issues in accounting. AI has the potential to alleviate some of that by eliminating drudgery and overtime caused by manual tasks. Indeed, Intuit’s marketing manager Langley Barth noted that AI tools in ProConnect are about “reducing burnout from tedious tasks” and allowing staff to focus on more meaningful work . If firms can show staff that AI makes their jobs saner and more interesting, they have a better shot at retaining them in this competitive labor market.

However, there’s nuance. Some employees, especially older ones or those in routine roles, fear AI or feel threatened by it. Change management is needed to reassure them. The firms that succeed create a culture where human workers understand they are “AI managers” or “AI augmenters” – not cogs to be replaced. Accountants are encouraged to see AI as a tool that can make them more effective and free them from tasks they never liked anyway (like endless data entry or cross-footing spreadsheets).

Human-AI Collaboration – The New Norm: We are entering an era of blended teams. A Gartner prediction, shared at an AICPA conference, was that by 2029, one-third of finance staff will be in roles where they work interdependently with AI – essentially sharing their job with an AI counterpart. That idea is no longer far-fetched. Already, we see roles like “AI-Augmented Auditor” or “Tax Reviewer & AI Specialist” popping up in job descriptions. These aren’t separate roles but expectations within roles: e.g., a tax senior who knows how to run the AI tools and will manage a portion of the workflow through them.

The day-to-day is changing accordingly. For example, an auditor might use an AI tool to examine 100% of journal entries for anomalies (something that would have been impossible manually). The AI flags, say, 17 transactions as high-risk. The auditor then uses her professional judgment to investigate those. She might even ask the AI follow-up questions (if it has that capability) like “Explain why these were flagged” and get some reasoning (maybe they were posted at odd hours or just under approval thresholds). The auditor then concludes whether it’s an issue or false positive. In essence, the AI did the heavy lifting of combing through data, and the human applied the reasoning and decision-making – a partnership.

Another example is in tax planning: a CPA might ask a generative AI tool to draft a memo on the tax implications of a client’s potential move from New York to Florida. The AI drafts a comprehensive memo in seconds, citing relevant statutes about domicile and residency rules. The CPA then reviews it, corrects a nuance about New York City tax, and adds a personalized recommendation for the client. The final product is a team effort between AI and human. The human’s role shifts to editor, fact-checker, and relationship manager – ensuring the output is correct and tailored to the client.

In terms of attitude, AICPA CEO Mark Koziel encapsulates the sentiment among forward-looking CPAs: “AI is not going to disrupt the accountant – it’ll change what the accountant does, but it will not replace the accountant. As long as we keep up with the skills we need, we’ll continue to be a profession that prospers.”

That was Koziel speaking at a symposium in 2025, and it resonated widely. It captures the optimism that accountants, armed with AI, can actually become more valuable. They’ll spend more time on advisory conversations, strategy, and judgment calls – the things clients truly value – and less on rote processing. But it also has a warning: “as long as we keep up with the skills we need.” That’s the onus on every professional now, to learn and adapt continuously.

The talent landscape in accounting is tilting toward a more technologically fluent, advisory-focused workforce. Firms that manage this transition – by reskilling their people and perhaps hiring new kinds of expertise – are finding that AI can elevate their human capital rather than diminish it.

AI in Action: Use Cases and Success Stories from Mid-Sized Firms

It’s easy to talk about AI in abstract terms, but what does it actually look like in an accounting firm’s daily work? Let’s explore several real-world use cases – drawn from common practice and specific firm case studies – that illustrate how AI is being applied, and the tangible outcomes.

1. Tax Compliance & Preparation

Use Case: Automating tax return preparation and review.

How AI Helps: AI can auto-extract data from client documents (like W-2s, 1099s, trial balance reports) and populate tax returns. It also runs algorithmic checks on returns to catch errors or missing info.

Benefits: Huge time savings – many firms report 50-70% reduction in preparation time for individual returns when using AI-driven import and validation . For example, instead of a staffer spending 3 hours entering data and another hour proofreading, the AI might do the entry in minutes and flag a few items for review, with the staffer spending one hour total. Some firms have been able to have one preparer handle what used to be the workload of two, during peak season. Also, accuracy improves: AI doesn’t get tired at 11pm in March and transpose digits. A mid-sized firm that adopted an AI tax prep tool saw a dramatic drop in review notes from partners because returns were cleaner to begin with – they even boasted “90% fewer compliance errors” year-over-year after implementation .

Case in Point: Consider how a firm might integrate an AI module into their tax software. They use it primarily for their hundreds of individual 1040 clients. The AI auto-maps client QuickBooks data to tax form line items and pulled in scanned documents. The firm might prepare 55% more returns per preparer than the prior year with similar staffing. Instead of laying people off, they are able to take on more clients and reassign some staff in April to advisory projects, like tax planning scenarios, because compliance was under control.

2. Tax Research & Advisory

Use Case: Rapid tax law research and proactive client advisory.

How AI Helps: Natural language processing (NLP) allows accountants to query vast tax law databases by asking plain-English questions. AI can also analyze a client’s financial scenario and suggest tax strategies or flag issues (e.g., detection of a large increase in income that might merit estimated tax adjustments or an S-corp reasonable compensation review).

Benefits: Speed and depth of insight. Traditionally, researching a nuanced tax question (say, on multi-state income sourcing or a new provision in a regulation) could take many hours of reading textbooks or searching tax libraries. Now, AI-powered research tools like Thomson Reuters Checkpoint Edge with generative AI or Bloomberg’s similar AI assistant can produce an answer in seconds, often with citations to the code and regs. Accountants still must verify and interpret, but it compresses the time dramatically. Also, AI can surface planning opportunities: for instance, if a client’s data is in the system, an AI can scan it and say, “Client’s charitable contributions went up 50% – consider advising on a donor-advised fund.” Firms report that thanks to AI analytics, they have increased the volume of value-added client touches.

Case in Point: At RSM US, a top-5 accounting firm, they developed an internal generative AI called Guidepath to assist with writing complex tax position memos. As RSM’s digital leader Sergio de la Fe says, when dealing with thorny client issues (like interpreting new tax law changes for a specific scenario), their AI can search through the firm’s internal knowledge (prior memos, court rulings, IRS releases) and draft a detailed position paper . “We’ve leveraged language models to help write those memos… accessing all our databases,” de la Fe said . This doesn’t eliminate the tax experts – on the contrary, it augments them, allowing senior tax experts to produce high-quality analyses faster, and involve juniors in higher-level work by letting them explore what the AI found and whether they agree. RSM also uses AI as a “resource of resources,” meaning staff can ask the AI, “What tool can I use to do X for this audit?” and it will suggest the appropriate software or approach – essentially internal consulting at one’s fingertips.

3. Audit & Assurance

Use Case: Automating audit data analysis and routine testing.

Benefits: The audit process becomes faster and more thorough. Instead of sampling 30 transactions out of thousands, AI can check all thousand for certain criteria. This increases assurance (better chance of catching fraud or errors) and can cut down the time auditors spend on manual ticking and tying. Deloitte’s AI, for instance, can do an initial read of all lease agreements to spot unusual terms, which a human might miss until late in the audit if at all. Firms have reported audit fieldwork time reductions of 20–30% with AI – meaning less disruption for clients and lower cost of service (or ability to perform more audits with the same staff). One published stat from case studies is 69% faster audit document review using AI. And because AI can maintain an immutable log of what it checked, documentation is often improved – the AI-generated workpapers can detail what was done, which auditors can then annotate with their conclusions.

Case in Point: PKF O’Connor Davies, a large regional CPA firm, applied AI in their audit data prep. Their tech advisory leader Suma Chandar said they look for cross-functional use cases where “doing one or two things can bring in 30-40% efficiency in their day, whether tax or audit or ESG” . One success was using AI and RPA (robotic process automation) to handle routine SOC 2 compliance testing and investment account reconciliations for audits. They built a “digital worker” (RPA bot with AI) that logs into 100+ investment accounts, pulls data, consolidates it, reconciles to the general ledger, flags anomalies, and produces an audit trail report . According to Carmel Wynkoop, the partner in charge of AI at the firm, this saved about 500 hours a year of manual work – time that their audit staff can now spend on risk assessment and investigating issues rather than gathering data.

Another mid-sized firm, GarbelmanWinslow (GWCPA), used an AI tool to analyze client financial statements: they upload a PDF of the statements (after scrubbing client-identifying info) and ask, “What key factors should we alert the client to?” The AI’s quick analysis often surfaces insights or inconsistencies that the firm then discusses with the client. Samantha Bowling, GWCPA’s managing partner, says that within seconds, the AI provides a helpful rundown, effectively acting as a junior analyst in the process.

4. Client Accounting & Bookkeeping

Use Case: Automating write-up work and bookkeeping tasks.

How AI Helps: AI and machine learning algorithms categorize transactions, reconcile accounts, and even generate draft financial statements with minimal human input. For firms offering Client Accounting Services (CAS), AI tools can handle bill pay, invoice coding, and basic financial analysis alerts.

Benefits: A lot of CPA firms have been expanding into outsourced bookkeeping for clients (CAS practices). AI is turbocharging those operations. For instance, AI bank-feed rules can learn a client’s transaction patterns and post items accurately 90%+ of the time, whereas a human bookkeeper might achieve similar accuracy but at a slower pace. AI also provides continuous accounting – books can be kept up-to-date in near real-time, rather than waiting for month-end. This enables the firm to give clients more timely advice. Also, because these tasks are highly automatable, firms can scale CAS services without linear increases in headcount. There’s a claim that current automation tech (including AI) can handle up to 70-80% of basic accounting transactions automatically , which allows existing staff to manage more clients.

Case in Point: A case study often cited is Botkeeper, an AI-powered bookkeeping platform used by many CPA firms. One mid-sized firm implementing Botkeeper for their small-business clients found they could reassign three staff who were formerly doing manual data entry into more analytical roles (like preparing KPI dashboards for clients) because the routine entries were handled by Botkeeper’s AI. Another example is an internal tool at PwC (a Big Four, but instructive): they developed AI to automatically draft portions of the financial statements and footnotes by pulling data from the trial balance and prior reports. While mid-sized firms might not build that themselves, they benefit from vendor software that does similar things on a smaller scale (like Xero and QuickBooks Online increasingly have AI categorization and even draft reporting features). The result is that the accountant’s role shifts to reviewing what the AI did, correcting it if needed (and in doing so, training the AI further), and then focusing on advising the client on what the numbers mean.

5. Administrative and Other Uses

Beyond core technical work, firms are deploying AI in supporting roles as well:

Common AI Use Cases in CPA Firms and Reported Benefits:

Looking at these, a pattern emerges: AI excels at tasks that are data-heavy, rules-based, and repetitive, and by taking those on, it boosts efficiency and often quality (fewer human slip-ups).

Meanwhile, human professionals are refocusing on tasks that require judgment, context, and interpersonal skills – like consulting with clients on implications of that data, making final decisions on uncertain transactions, or simply spending more time talking to clients rather than poring over paperwork.

One more real-world story to illustrate impact: A mid-sized firm in the Midwest implemented an AI-based audit confirmation service (which uses AI and RPA to send confirmations to banks, lawyers, etc., and chase them up). The managing partner reported that their audit seniors – who used to dread managing confirmations and often had to put in extra hours to get responses – were “absolutely thrilled” that the AI bot now handled 90% of the chasing. Morale went up, and seniors had more time to focus on analyzing the confirmation responses. Little improvements like that in multiple areas added up to them cutting one week off their average audit timeline and giving staff a less chaotic experience.

It’s worth noting that not every AI use case yields dramatic results; some might even fail or need iteration. Some firms tried AI for complex analytical reviews and found it wasn’t accurate enough without significant tweaking. The smart approach is usually to start with “low-hanging fruit” use cases (like data import, which is relatively straightforward) and then progress to more sophisticated ones (like predictive analytics for clients) as confidence and competency grow.

With the enticing success stories, however, also come concerns and new risks, which we will explore next. For every digital “worker” that saves 500 hours, a partner is likely asking: are we sure it’s doing it right and not missing something? Are we exposing ourselves or our clients to any risk by using AI? Those are valid questions, and addressing them is crucial for sustained success.

Navigating Risks, Regulation, and Ethical Challenges

Every new technology brings a set of risks and challenges, and AI is no exception – in fact, AI might bring more than most, given its black-box nature and potential for misuse. Mid-sized CPA firms, like all others, must grapple with ensuring that AI is used responsibly, securely, and in compliance with both professional standards and laws. Let’s break down the key risk areas and how firms are handling them:

1. Data Security & Privacy: Accounting firms are custodians of highly sensitive client data – financials, social security numbers, earnings, etc. Plugging AI into the mix raises questions: Where is the data going? Who can see it? Many AI tools, especially early on, were cloud-based with data possibly passing through third-party servers. Firms have to ensure any client data sent to an AI (for example, to a cloud AI service for analysis) is protected. A major worry is staff unintentionally feeding confidential info into public AI services (like ChatGPT’s free version), which could lead to data leaks. Alarming stats from a 2025 KPMG survey show 46% of U.S. workers admitted to uploading sensitive company info to public AI platforms, and 53% even presented AI-generated output as their own work without disclosure – essentially bypassing controls . For CPA firms, that’s a huge no-no if it’s client data.

Mitigation: Firms are instituting strict AI usage policies. Many have outright bans on using free public AI with client data. Instead, if they want those capabilities, they either use a paid enterprise version with privacy guarantees (OpenAI offers a business tier where data isn’t used to train the model, for example) or leverage AI solutions that run in secure clouds specifically for their data (some larger firms even host AI models on-premises). Also critical is training staff: making sure everyone understands that popping client details into a random chatbot is akin to shouting them in public. Some firms go further and use DLP (Data Loss Prevention) software that flags or blocks if someone tries to use certain web AI tools at work.

2. Accuracy and “Hallucinations:” AI tools, especially generative ones, can produce inaccurate or completely made-up outputs. The term “hallucination” is used when an AI confidently generates an answer that is just false. In accounting, even a small error can have big consequences (e.g., a wrong number on a tax form, or an incorrect interpretation of a law). According to a Karbon survey, 70% of accountants were concerned about data security in AI, and nearly half (47%) worried that AI might decrease human touch or judgement . Accuracy is a top concern: one study noted 62% of accountants worry about errors in AI-generated reports .

Mitigation: Human review remains paramount. The golden rule many firms adopt is: AI can draft or analyze, but a human must validate any final output that goes to clients or official filings. In a Thomson Reuters survey, professionals said AI outputs would need to be “100% accurate” before they’d be comfortable using them without human review – a bar that is arguably impossible, hence human review stays in the loop.

Firms are creating checklists for reviewing AI work (e.g., “If AI drafted a memo, verify each factual assertion and citation”). Additionally, some AI tools themselves provide a confidence score – and accountants are trained to interpret those. MIT/Stanford research found experienced accountants were good at using AI confidence scores to decide when to trust or double-check an output. That kind of discernment needs to be taught.

Another mitigation is starting AI usage in low-risk areas first. For instance, use AI to draft an internal report or do a preliminary analysis, not to finalize an SEC filing footnote without extensive checking. Over time as the firm builds trust in a tool (and possibly customizes it to their needs), they can expand its autonomy.

3. Explainability and Auditability: Clients, regulators, even the firm’s own partners will ask “How did the AI get this result?” If an AI flags a strange journal entry as fraud risk, you need to be able to explain why (maybe it spotted an outlier pattern). Black-box models that just spit out answers without rationale can be problematic. And if regulators come knocking (say, the PCAOB examining an audit), they will want documentation of what the auditors did – that now includes what the AI did.

Mitigation: Many AI software providers in this space are adding explainability features. For example, an AI tax research tool might show the sources it used (cite IRC sections or rulings) to derive an answer, which the CPA can document in their workpapers. In audits, if AI is used to select samples or flag entries, auditors include those AI parameters and outputs in their audit documentation. Some firms say they treat the AI like a junior staff – everything it does is reviewed and documented as if a person did it. Also, there’s the concept of an AI audit trail: some systems maintain a log (date/time, action, result) for all AI-driven processes. For instance, if an AI auto-posts an entry, it logs why (rule applied, etc.).

From a governance perspective, boards and firm leadership are taking note. The Wolters Kluwer report highlighted that best-in-class firms are doing quarterly AI oversight meetings, and annual disclosures to clients about AI use are becoming more common (e.g., telling audit committees that “we used AI tool X in conducting the audit of accounts receivable”). This transparency helps build trust and also forces the firm to reflect on how AI is affecting its risk profile.

4. Ethical and Professional Responsibilities: CPAs are bound by ethical standards like confidentiality, integrity, due care, etc. The use of AI touches these areas. For example, confidentiality is at risk if AI is not secure. Due care is relevant if a CPA overly relies on AI without sufficient competence or skepticism – e.g., if they accept an AI’s tax answer without understanding it, are they exercising due professional care? Also, if AI is used to assist in making judgments (like valuation or going concern), how do biases get identified and mitigated?

There’s also an ethical dimension in terms of disclosure: Should clients be informed when a significant portion of work was done by AI? Some might argue it’s only fair the client knows if a letter was AI-drafted (though it’s edited by a human). Most clients probably don’t mind as long as quality is high, but transparency is generally a good policy.

Mitigation: The AICPA Code says a CPA can use a third-party service provider (which could include an AI service) but must ensure confidentiality and should disclose to the client if it’s a significant portion of the engagement (with some exceptions). Many firms have started updating their engagement letters to mention the possible use of AI or outsourced services in performing the work, while assuring confidentiality and quality controls.

Firms are also establishing internal “AI ethics committees” or task forces. These groups review new AI use cases for ethical implications, set boundaries (e.g., “We will not use AI to make final judgments on high-risk areas; we will use it only as input”), and monitor the landscape for new guidance.

5. Regulatory Uncertainty: While we have accounting-specific standards to think about, there’s broader AI regulation looming. The EU’s AI Act, for instance, would impose requirements on AI systems, especially for high-risk use cases. In the U.S., there isn’t federal AI law yet, but sectoral regs (like SEC or FTC guidance) could emerge if AI causes issues.

Accounting firms also have to consider things like whether using AI might violate any privacy laws (say, feeding personal data into a model could engage laws like GDPR or California’s privacy law, if not handled properly).

Mitigation: Keeping an eye on regulatory developments is key. Larger firms have policy teams; mid-sized firms rely on industry associations or legal counsel for updates. Some firms opt to err on the side of caution – implementing higher standards voluntarily. For example, treating their AI systems as if they were under EU AI Act already (ensuring explainability, risk assessments) could be a strategic move to be ahead of the curve. The Journal of Accountancy reported how one firm started aligning with European guidelines because U.S. ones lagged. Avani Desai, CEO of Schellman (a Top 100 firm focusing on tech audits), says, “Work with regulators early. Don’t wait for the mandate.” That might mean participating in industry groups that liaise with the SEC or IRS on AI issues, so the profession can shape sensible rules.

6. Model Bias and Fairness: Though not as front-and-center for tax and audit as it is in, say, lending decisions, bias is something to watch. If AI were used in hiring or evaluating staff, for example, firms would need to ensure it’s not discriminatory. Or if AI summarized client data, does it inadvertently reflect bias (e.g., prioritizing certain data over others)?

Mitigation: Use reputable models and data sets, and test them. The “trusted AI” frameworks (like KPMG’s) emphasize bias testing. In accounting use cases, bias might show up as systematic errors (like always misinterpreting certain legal provisions in one direction). Continuous monitoring and feedback loops help catch these. In audit, reliance on AI might inadvertently introduce bias if, say, the AI overlooks certain anomalies because they weren’t in its training data – so auditors still need to apply their skeptical mindset.

One emerging practice is “parallel testing” – for a period, run AI tools in parallel with traditional methods to compare results. For instance, if AI picks audit samples, also have a human pick samples the old way and see if AI missed something. If it did, tune the system. Over time, confidence builds that the AI’s choices are sound.

Finally, an often-overlooked risk: how will clients feel about the firm using AI on their work? If a client learns that a report was AI-generated, will they think “am I paying high hourly rates for a computer to do the job?” This is more of a reputational and communications issue.

Many firms choose to be upfront about AI as a positive – “We use advanced tools, including AI, to provide you the best service possible.” They focus on outcomes: faster delivery, more insights, fewer errors. Most clients are fine with that. Some high-touch clients might say, “I want John to do this, not a machine,” in which case the firm can explain John is still doing it, just with smart software help. Also, firms can emphasize their review and quality control over the AI.

Interestingly, a Fortune magazine analysis in 2025 pointed out that 72% of S&P 500 companies flagged AI as a material risk in their own disclosures, often citing reputation and compliance concerns. So, business clients are aware of AI issues themselves. This can actually be a conversation point that elevates CPAs to a consultative role: “We use AI responsibly in our work, and we can also advise you on how to govern AI risk in your company.” CPAs and consultants are seizing that as a niche – helping clients with AI governance, which is a new advisory service some firms now offer, leveraging their own learning.

CPA firms must implement strong governance around AI use.

The formula often cited is People + Process + Technology controls:

Leading firms view this as an ongoing effort, not one-and-done. They might start with a conservative approach, then gradually allow more AI autonomy as controls prove effective. For example, at first maybe AI suggestions are never automatically enacted – always a human clicks “approve.” Later, for low-risk tasks, they might let AI auto-complete things, with periodic audits of its work.

The consensus thus far is that the benefits of AI can far outweigh the risks, but only if the risks are managed proactively. In the next section, we’ll discuss how firms are practically implementing AI and overcoming these hurdles, as well as recommendations for those just starting out.

Overcoming Implementation Hurdles: Change Management and Best Practices

Adopting AI in a mid-sized accounting firm is not like flipping a switch – it’s a journey that involves technology, but perhaps even more so people and process changes. Many firm leaders have shared that the tech was ready before their organization was.

Let’s look at the common challenges firms face when rolling out AI initiatives and how savvy firms are addressing them:

Challenge: Resistance to Change – “We’ve always done it this way.”

Partners and staff who have decades of experience doing tasks manually may be skeptical or fearful of AI. They trust their tried-and-true processes more than a new algorithm. Some may worry that embracing AI reduces the perceived value of their own skills or even threatens their job security.

Response: Top-down vision, bottom-up engagement. Successful firms often start with leadership clearly communicating why the AI adoption is happening – focusing on positive outcomes like “this will free us to do more value-added work” or “this will help us grow without burning out.” At the same time, they engage the rank-and-file early. One accounting firm formed an “AI working group” that included not just IT folks but also a senior partner, a couple of junior accountants, and a tax manager. This group piloted the AI tool and worked out kinks, so when it rolled out firm-wide, these insiders could champion it and train others.

Challenge: Training and Skill Gaps.

Not everyone is tech-savvy, and AI tools can have a learning curve. If not trained properly, staff might misuse the tool or simply not use it at all (leading to shelfware). A survey by Corporate Compliance Insights found only 22% of organizations have a defined AI strategy and many plunge in without full training, which can cause problems.

Response: Invest in comprehensive training and continuous learning. Firms that budget time for training (even if it means non-billable hours) reap the rewards. Some arrange vendor-led sessions followed by internal peer-to-peer training. Many create quick reference guides or “AI tool FAQs” tailored to their workflows. And training isn’t one-off; as AI tools update, they offer refresher courses.

A smart strategy is to use real client work in training. For example, during training on an AI tax software, have staff actually prepare a sample return through it, rather than abstract demos. This builds muscle memory. Also, pairing up less tech-confident staff with a tech-savvy “buddy” can help during the initial phases. The buddy can be on-call to assist until the person gets comfortable.

Challenge: Data Preparation and Integration.

A saying in AI is “garbage in, garbage out.” If a firm’s data (like client records, prior working papers, etc.) are disorganized or locked in silos, AI tools won’t perform well. Many firms find they first need to clean up data or migrate to cloud systems before layering AI on top. Integration with legacy systems can also be a headache – e.g., making an AI tool talk to an old practice management system.

Response: Phase the project and build a data foundation. Rather than tackling everything at once, firms often start with a subset of data or a particular service line. For instance, digitize and clean the tax workpapers for the last two years to feed the tax AI (ensuring consistency in how data is labeled), rather than trying to do tax and audit simultaneously. Some firms create new data warehouses or “lakes” – essentially central repositories where data from different sources is aggregated in a clean, AI-ready format . This might involve hiring a data engineer or using a consultant, but it’s an investment that pays off beyond AI too (it improves reporting, etc.).

On integration, lots of modern AI solutions have APIs and can connect to common software, but a firm might still need to build some custom connectors. The workaround for smaller firms is often manual import/export in the interim (e.g., exporting a QuickBooks file for the AI to analyze, until a direct integration is set up). The key is not to let integration challenges halt the whole project – find a way to pilot even if it’s a bit kludgy at first, then refine.

Challenge: Measuring Success (or lack thereof).

features.AI isn’t cheap or easy, so management wants to see results. But if they don’t establish clear metrics, they might not realize value or might pull the plug prematurely. On the other hand, a poorly managed project might actually not deliver value if, say, people only use 10% of the tool’s features.

Response: Define Key Performance Indicators (KPIs) and track them. Firms set targets like “reduce tax prep hours by 30%” or “each auditor to automate 50 confirmations.” By tracking baseline stats before AI and then after, they can quantify gains. It’s important to be realistic – maybe in year 1 the goal is 15% time savings, ramping up as everyone gains proficiency. If metrics show underperformance, firms dig in to find out why. Often it’s a usage issue – maybe people reverted to old methods under stress. Additional training or tweaking the process can solve that. It’s an iterative approach: measure, adjust, measure again.

Some firms also celebrate wins to keep momentum. For example, if the first AI-prepared tax returns go out with zero review notes from partners, that’s announced firm-wide as a success. Recognition (even small things like an email shoutout or a gift card to the “AI champion of the month”) can motivate staff to engage positively.

Challenge: Cost and ROI Pressure.

Especially for mid-sized firms with tight budgets, AI investments (software subscriptions, possibly new hires like data analysts, training time that isn’t billed) can cause partners to worry about short-term impact on profits. If results aren’t immediate, there could be pressure to cut losses.

Response: Start small, prove value, then scale – aka “quick wins.” Many firms begin with pilot projects: e.g., try the AI tool on 10 clients or for one quarter. This limits the spend and risk. If the pilot shows clear time savings or quality improvement, it’s much easier to get buy-in for a wider rollout. Quick wins also provide internal case studies you can use to convince skeptics.

Additionally, some firms negotiate with vendors for phased or performance-based pricing – maybe a lower cost in year 1 pilot with the understanding of expansion later. Vendors competing in this space often accommodate pilots because they know landing a whole firm is a big win.

ROI analysis should also factor in intangible or long-term benefits, not just immediate labor hour reduction. For instance, if AI leads to winning a new client because of enhanced capabilities, that revenue might not be obviously attributed to the AI in a simple ROI calc, but it’s part of the picture. One approach: create a “benefits register” where all positive outcomes (time saved, new engagements, client compliments, reduced overtime costs, etc.) are logged to build the case.

Challenge: Workflow Disruption and Teething Troubles.

Implementing AI can temporarily disrupt workflows. Maybe the AI outputs require new review steps, or initial results have errors that have to be ironed out. Staff can get frustrated if the new process seems harder at first.

Response: Refine processes and don’t abandon after the first hurdle. It’s common that the first few weeks of using a new AI tool feel slower than the old way – learning curve effect. Firms need to message this expectation and encourage perseverance. Many create new standard operating procedures (SOPs) that incorporate the AI: e.g., “Step 1: Import data via AI tool; Step 2: run AI diagnostics; Step 3: human review and finalize.” Embedding it into checklists ensures it’s not seen as optional or extra, but part of the job.

During the initial period, frequent feedback loops are useful. Daily or weekly quick meetings to discuss: what’s working, what’s not? Perhaps the AI is misreading some documents – feed that back to vendor or adjust the doc template. Perhaps staff discovered a feature that saves more time – share it across the team. In one firm, juniors discovered the AI tax software could also auto-generate a nice client summary letter from the return – something the firm hadn’t been doing due to time, but now can. They incorporated that into the process, adding client value.

Challenge: Keeping up with Continuous Change.

AI tools evolve rapidly. New features roll out, models get updated. The firm’s processes and controls might need updates too. It’s a moving target, not a one-time implementation like, say, switching from one tax software to another which then stays stable for a year.

Response: Stay agile and foster a culture of adaptability. Firms sometimes establish an ongoing “AI committee” (beyond initial rollout) to continuously monitor tech developments. They subscribe to industry newsletters, attend conferences, network with peers about what’s new. If a vendor releases a great new feature (e.g., AI in auditing can now test entire populations for something that wasn’t possible before), these firms pounce and pilot it. Essentially, they treat AI adoption as a continuous improvement program, not a project with an end date.

Some adopt an experimental mindset: allocate, say, 5-10% of staff time or IT budget to experiment with new AI ideas every quarter. Not all will pan out, but this way they don’t stagnate. As one partner quipped, “We’ll never be ‘done’ implementing AI – it’s like saying you’re done improving.” They have embraced that notion.

Challenge: Vendor Selection and Reliability.

With so many AI solutions, picking the right one is daunting. And relying on an external tool means if it has downtime or issues, your work could be stuck.

Response: Careful due diligence and backup plans. Firms are increasingly doing proofs-of-concept with multiple vendors before committing. They involve the actual end-users in trials to gauge which tool fits best. They also review vendor security (due to the earlier point on data protection) – many send vendors security questionnaires or even do on-site audits for bigger deals. Once a tool is chosen, firms negotiate SLAs (service-level agreements) if possible, to ensure support if something goes wrong.

Having a backup process is wise: e.g., “if the AI system is down, we revert to manual for that period.” Or maintain capability to do things the old way in parallel for a while until trust in the new way is absolute. Some firms keep legacy processes documented just in case, but over time as confidence grows, those become less needed.

Having discussed these implementation aspects, it’s evident that adopting AI is as much about change management as it is about technology. Firms that succeed often say the key was getting their people on board and iterating on processes, rather than any magical tech trick.

The Road Ahead: Future Outlook and Strategies for Firm Leaders

As we cast forward into the later 2020s, it becomes clear that  AI will become even more deeply ingrained in accounting workflows, and the firms that leverage it effectively will set the pace for the industry.

Nearly Autonomous Processes: We can expect many routine accounting processes to become “near-autonomous.” That doesn’t mean no humans involved, but minimal intervention. For example, consider the tax compliance cycle – within a few years, it’s plausible that a client’s transactions flow continuously into an AI-driven tax engine, which updates an ongoing draft tax return in real time. By year-end, the return is 80-90% complete without much human touch, aside from confirming unusual items. The role of the CPA at that point is to review and strategize on the final 10%, rather than manual data entry through January and February. Workflows will approach a point where they “run themselves,” with professionals overseeing multiple automated streams of work rather than preparing each item by hand . Audit processes too could reach near real-time status – some firms talk about “continuous audit” aided by AI agents that monitor client systems year-round, flagging issues immediately.

Proliferation of Vertical AI Agents: Just as spreadsheets were a general tool that spawned specialized versions (for finance, stats, etc.), AI will likely spawn specialized agents for different domains of accounting. We might see an “AI Audit Assistant” that is distinct from a “Tax Planning Genie” or a “CAS Bot.” In fact, Thomson Reuters already branded “CoCounsel” as a tax-and-accounting-specific AI assistant, separate from generic AI . We’ll see more niche AI solutions for things like sales tax compliance (imagine an AI that stays on top of every sales tax rule change and automatically configures a client’s systems), or for forensic accounting (AI that can sniff out fraud patterns across millions of transactions and maybe even generate a draft of the investigative report). These vertical agents will multiply market impact, because they’ll be tuned to understand accounting concepts at a deeper level than broad models. For firm leaders, staying abreast of these developments and being willing to adopt the right tool for the right task (instead of one-size-fits-all) will be key.

Heightened Emphasis on Explainable and Ethical AI: As AI is used for more critical decisions, regulators and clients will demand more explainability and accountability. We touched on how 72% of big companies now disclose AI risks ; that trend will likely spur formal regulations or at least industry standards. We might see audit standards updated to address AI – e.g., the PCAOB could issue guidance on using AI evidence in audits (perhaps requiring documentation of how AI tools were validated). The IRS or state boards might set rules on AI-prepared returns (for instance, maybe requiring disclosure if a return was largely prepared by AI, or guidelines on taxpayer responsibility). Smart firms will engage with bodies like the AICPA to help shape these standards. The AICPA, in fact, launched an “AI Task Force” (hypothetically, given the direction) to issue recommendations on AI governance for firms. We can anticipate frameworks akin to today’s cybersecurity frameworks – perhaps an “AI control framework” – becoming part of firm peer reviews or practice inspections. One could imagine future peer review checklists: Do you have an AI use policy? How do you validate outputs? How do you secure data in AI systems? Being ahead on those governance fronts will become a competitive differentiator (clients might favor firms with an “AI Seal of Quality” if such a thing emerges).

New Services and Business Models: AI will also open new business opportunities for CPA firms. Many mid-sized firms are already advising clients on implementing AI in finance or doing risk assessments of client AI systems. This can bloom into a service line – akin to how firms developed IT consulting or cybersecurity practices. Given the trust in CPAs, there’s talk of CPAs serving as independent AI verifiers for clients, somewhat like how they lend credibility to financial statements. The Journal of Accountancy even floated the concept of “CPAs as AI system evaluators” . For instance, a client might ask their CPA to evaluate the controls and bias testing of an AI they want to use in finance, providing a level of assurance to stakeholders. So, firm leaders should consider investing in internal expertise not just to use AI, but to consult on it. This might mean hiring a data scientist or training some auditors in AI risk, positioning the firm to do AI-related advisory projects.

Talent Strategy Shift – From Recruitment to Reskilling: The talent crunch isn’t going away, but AI might change its nature. If fewer entry-level folks are needed (or if they need a different skill profile), firms could partner with universities to shape accounting curriculum towards more tech. Some firms already are hiring non-traditional talent – for example, economics or computer science grads – into accounting roles and training them on accounting, figuring it might be easier to teach accounting to a techie than tech to an accountant. The next generation of CPAs will likely all have some coding or AI familiarity – the pipeline will adjust (in fact, the CPA Exam is undergoing changes to incorporate more technology knowledge).

For existing staff, expect a continuous upskilling mandate. Those unwilling to learn might find their career prospects limited. But those who do may find accelerated paths: perhaps the traditional 10-15 year path to partner shortens if technology leverage means younger CPAs can take on more responsibility sooner. Firm leaders should create clear pathways for tech-savvy staff – maybe a role like “Director of AI Implementation” or making such skills criteria for promotion to manager.

Competitive Landscape and Consolidation: Historically, technology shifts in accounting (like the move to PC software, or to cloud) tended to first be leveraged by larger firms, but eventually leveled the field or at least reconfigured it. AI could in some ways level things (a 50-person firm with great AI tools can potentially outserve a 200-person firm that doesn’t use them well). We might see some consolidation as laggards decide it’s easier to merge into a tech-forward firm than catch up. On the flip side, nimble mid-sized firms that excel in AI might capture more market share, possibly expanding beyond their traditional regions or sectors because technology removes some barriers. If your AI-enhanced team can handle twice the clients, you might expand services to smaller companies or not-for-profits that you couldn’t profitably serve before.

Client Expectations: In the near future, clients – especially younger entrepreneurs or CFOs – will expect their CPA to have AI-powered capabilities. They’ll ask questions like, “Do you use AI to ensure nothing is missed in my audit?” or “What analytics can you provide us from our data?” Clients might not know the term “agentic AI,” but they’ll want the speed, insight, and proactive advice that AI enables. Keeping client communication about AI positive is key: some firms, for example, highlight in proposals that “we use advanced AI tools that allow us to complete the engagement faster and with deeper analysis, at no extra cost to you.” That signals efficiency and innovation, which can help win business.

Recommendations for Firm Leaders:

• Collaborate and Learn: Join industry groups focusing on AI (the AICPA, state CPA societies, and global accounting networks are all hosting AI roundtables these days). Sharing experiences with peer firms can prevent reinventing the wheel and keep you updated. Also, involve stakeholders like clients – some firms have beta-tested AI-generated analysis with friendly clients to get feedback on whether it’s useful.

• Mind the Human Touch: As processes automate, double down on human relationships. Ironically, the more we automate, the more critical the distinctly human aspects become – trust, empathy, judgment. Train staff in “soft skills” to complement the tech. Ensure partners spend time with clients discussing the meaning of all the fancy new reports AI produces. In essence, use AI as an enabler to enhance client service, not as a substitute for it.

The tools are here, and adoption is accelerating.

CPA firms have a chance to leap forward by embracing these technologies thoughtfully. Those who act with clarity and intent – training their people, managing risks, and focusing on delivering greater value to clients – are likely to thrive and set themselves apart. Those who drag their feet may find the gap widening to a point where catching up is extremely difficult.

The accounting profession has undergone many evolutions – from paper ledgers to spreadsheets to cloud platforms – and each time, firms that adapted emerged stronger. AI is arguably a bigger leap than any of those, but it carries the same promise to free smart people from menial tasks so they can do more impactful, meaningful work.

For firms, this is a moment of opportunity to punch above their weight by leveraging AI. With strategic adoption, continuous learning, and a commitment to ethical, high-quality service, these firms can not only improve their own operations but also elevate the value they provide to clients in a rapidly changing business world.

The future of accounting won’t eliminate accountants – it will elevate them. And the journey to that future is happening now, one AI pilot and one success story at a time.

References:

1. Bramwell, Jason. “Accounting Firms Are Choosing Transformation Over Tactics, Wolters Kluwer Report Says.” CPA Practice Advisor, Oct. 9, 2025 . (Highlights AI adoption surge from 9% to 41%, with 77% planning more investment and 35% daily use)

2. Thomson Reuters Tax & Accounting Blog. “How do different accounting firms use AI?” Nov. 4, 2025 . (Examples of Big 4 and smaller firms’ AI use cases; 44% using AI daily among those using/planning to use GenAI)

3. Karbon. “State of AI in Accounting Report 2025.” (Survey data: 85% of accountants positive on AI, but only 37% of firms invest in training; firms investing in AI training save 22% more time ; 76% believe graduates prefer AI-using firms ; 56% say firm value drops if not using AI )

4. MIT Sloan. “How generative AI can make accountants more productive,” Aug. 5, 2025 . (Study by Xie & Choi: AI use led to 7.5-day faster monthly close ; 8.5% of time reallocated from data entry to higher tasks ; improved reporting detail +12%; need human judgment for low-confidence outputs)

5. Strickland, Bryan. “Reputation, security, compliance: Why AI risk disclosures are surging.” Journal of Accountancy, Oct. 29, 2025 . (72% of S&P 500 companies disclosed AI risk in 2025, up from 12% two years prior, citing reputational, security, compliance risks)

6. NYSSCPA. “Six Accounting Firms Report on How They’re Using AI,” May 7, 2024 . (Real firm examples: PKF using AI for cross-functional efficiency, Armanino’sdigital audit worker saving ~500 hours , GWCPA using AI to analyze financial statements in seconds , CLA automating info extraction saving hundreds of hours , RSM using generative AI for tax memos )

7. Hood, Daniel. “AI in accounting: Weighing the pros and cons.” Accounting Today, Apr. 2, 2025 . (Quote from AICPA CEO Mark Koziel: “AI will change what accountants do but will not replace them” ; Gartner prediction: by 2029, one-third of finance staff will have AI-human shared roles )

8. KPMG. “The American Trust in AI Paradox: Adoption Outpaces Governance,” Apr. 29, 2025 . (Study findings: 70% of U.S. workers eager for AI benefits but 75% wary of downsides; 44% use AI without authorization, 46% uploaded sensitive data to AI ; quote emphasizing need for strong “Trusted AI” governance )

9. Thomson Reuters (Joan Feldman via Attorney at Work). “AI Adoption Divide Dominates the 2025 Future of Professionals Report.” Oct. 2025 . (TR survey: firms with AI strategy 3.5x more likely to get benefits; 81% with strategy seeing ROI vs 23% with none ; 91% pros say AI must be more accurate than humans, 41% say AI outputs need 100% accuracy before no human review ; barriers include accuracy, security, lack of time/resources)

10. Axios (Ina Fried). “AI is already taking jobs away from entry-level workers,” Aug. 26, 2025 . (Stanford study using payroll data: 22–25 y/o employment in AI-impacted jobs down 16% since late 2022 ; Brynjolfsson notes speed of change and disparate impact on entry-level; caution that companies using AI to augment hire more, those using to replace hire less )