A CPA in Bogotá is replacing his staff with AI agents, and he thinks the rest of his profession has about two years to catch up or get caught.
Nick Dawson hasn't hired anyone new this year. He's not planning to. The 30-something CPA runs a US tax firm from a WeWork in Bogotá, Colombia, where he preps returns for small and medium-sized American businesses, does tax planning, handles bookkeeping, and fields panicked calls from clients who haven't filed in four years. It's a real accounting practice with real clients and real deadlines. The April 15th crunch is a month away and his team is heads-down grinding through returns like every other firm in the country.
But when Dawson talks about the future of his business, he doesn't talk about hiring. "I'm not hiring people," he says, matter-of-factly, from his desk at the coworking space. "I'm building AI agents."
Those agents, built on tools like Anthropic's Claude and open-source frameworks, handle the kind of work that used to fill the calendars of junior accountants: data entry, tax code research, document prep, compliance checks. Dawson is automating the machinery of his own firm in real time, and he believes the rest of the profession has a narrow window to do the same. Maybe two years. Maybe less.
This would sound like tech-bro fantasy if the numbers didn't back it up. Over 300,000 accountants and auditors have left the profession since 2020, shrinking the workforce by 17%. CPA exam participation has dropped more than 30% since 2016. Three quarters of practicing CPAs are approaching retirement age within the next 15 years. And in early 2026, both Goldman Sachs and Intuit announced partnerships with Anthropic to bring AI agents directly into accounting workflows.
The profession has survived every technological shift thrown at it for the last century and a half. But this one feels different. Not because the tools are smarter, though they are. Because the humans are disappearing at the same time the machines are showing up.
The Calculator, the Spreadsheet, and the Thing That Comes Next
Accounting is one of those professions that's been quietly reinvented by technology over and over again, and each time the people doing the work said the same thing: this changes everything. And each time, they were right. It just never changed as fast as they expected.
In the 1880s, William Burroughs invented the adding machine and accountants stopped doing arithmetic by hand. In the 1960s, electronic calculators arrived and the adding machine became a doorstop. In 1978, VisiCalc turned personal computers into financial modeling tools, and by 1985 Microsoft Excel had fundamentally altered how every firm on the planet operated. QuickBooks went mainstream in 1998. Cloud computing moved everything online in the 2010s.
Each wave killed a set of skills and created new ones. The accountant who could run a 10-key calculator blindfolded gave way to the one who could build a pivot table. The one who kept perfect paper ledgers got replaced by the one who understood database queries. Nobody mourned the old way for long. The new tools were just too useful.
Dawson sees the pattern clearly, but he also sees how it's accelerating. "You had all these old CPAs that used graph paper and rulers and sticky notes and 10-key calculators, and when the computer came out they're like, 'I like doing it this way,'" he says. "It's like that, but it's times a thousand and it's moving faster. It's exponential."
Here's what that acceleration looks like in practice:
| Era | Technology | What It Replaced | Years Between Shifts |
|---|---|---|---|
| 1880s | Adding machine | Manual arithmetic | Baseline |
| 1960s | Electronic calculator | Mechanical adding machines | ~80 years |
| 1978-1985 | Spreadsheets (VisiCalc, Lotus 1-2-3, Excel) | Calculators + paper ledgers | ~15-20 years |
| 1998-2010s | Cloud software (QuickBooks, Xero) | Desktop spreadsheets + filing cabinets | ~15 years |
| 2024-2026 | AI agents | Junior staff + manual research + data entry | ~10 years |
Sources: INAA, CPA Practice Advisor
The gap between revolutions is compressing. The jump from adding machines to calculators took 80 years. The jump from spreadsheets to cloud took 15. The jump from cloud to AI agents is happening in under a decade, and the agents themselves are improving on a monthly basis. Dawson started with one set of AI tools two months ago, pivoted to a completely different stack last month, and expects to pivot again before tax season ends.
According to a Deloitte survey, 58% of accounting firms had adopted some form of AI by 2024. But adoption is a generous word. ADP Research found that only 19% of accounting professionals use AI tools daily, while 17% have literally never touched the technology at work. The profession is split right down the middle, with one camp sprinting forward and the other pretending the starting gun never fired.
The Talent Crisis AI Walked Into
Here's the thing about AI arriving in accounting right now: it didn't walk into a healthy profession. It walked into one that was already bleeding out.
The numbers are stark. Over 300,000 accountants have left since 2020. The pipeline of replacements is drying up fast, with accounting graduates hitting a 20-year low in 2025. Eighty percent of firms report difficulty hiring skilled professionals, and the problem is getting worse, not better.
| Metric | Figure | Source |
|---|---|---|
| Accountants who left the profession since 2020 | 300,000+ | Bureau of Labor Statistics |
| Workforce decline | 17% | Ramp / BLS analysis |
| CPAs approaching retirement (within 15 years) | 75% | AICPA |
| Drop in CPA exam candidates since 2016 | 30%+ | CPA Trendlines |
| Firms reporting difficulty hiring tech-skilled staff | 80% | CPA.com 2025 AI Report |
| Accounting graduates (2025) | 20-year low | CPA Trendlines |
Sources: Ramp, CPA Trendlines, CPA.com
This isn't a cyclical downturn. It's structural. Young people looked at the profession, looked at the hours, looked at the pay relative to tech and finance, and chose something else. The result is a workforce that's simultaneously shrinking and aging, with no cavalry coming over the hill.
Which is exactly why someone like Dawson sees AI not as a threat but as the only viable path forward. Even during this tax season, with the April deadline bearing down, he feels the tension between the present grind and the future he's building toward. "Right now we have the deadline in a month and four days," he says, "and I've had to put the AI stuff on hold. We're all hands on deck. We got to get these projects out."
That's the reality for most small firm owners right now. You can't rebuild the plane while you're flying it through a thunderstorm. But the moment tax season ends, Dawson plans to go right back to building. He's already launched an AI optimization arm of his company, offering current clients the same efficiency tools he's building for himself. The pitch is straightforward: "We talk about options and use my experience to come up with planning opportunities. And then we finish the call and they say $14,000 and I cost $3,000. Do you want to work together? I mean, it's pretty easy value creation."
The talent shortage makes this math even more compelling. When you can't hire enough humans, machines stop being optional.
Two Years, Then the Floor Drops
Dawson doesn't sugarcoat his timeline. He thinks knowledge workers, himself included, have a roughly two-year window to extract maximum value from AI before the economics of their industries shift permanently.
"I think there's a two-year window where I could make a lot of money, scale and create efficiencies," he says, "and then before the god LLMs take over and the knowledge workers are just not as needed."
He's careful to clarify that he doesn't see this as doomsday. He calls it abundance. "Everything's going to be accessible to everyone at all times for nothing. Energy. Intelligence." The implication is that when the cost of cognitive labor approaches zero, the pricing power that professionals like CPAs, lawyers, and consultants have enjoyed for decades starts to erode. Not because the work goes away, but because it gets cheap enough that the margins collapse.
He's not alone in thinking this. A Fortune report from March 2026 cited Anthropic's own research suggesting a possible "Great Recession for white-collar workers" as AI capabilities outpace the rate at which humans can adapt. Thirty-seven percent of business leaders told researchers they anticipate replacing human workers with AI by the end of this year. Anthropic CEO Dario Amodei has publicly stated, with what he described as 70-80% confidence, that the first billion-dollar single-employee company could emerge in 2026.
The signals from the top of the market are just as loud. Goldman Sachs spent six months co-developing autonomous accounting agents with Anthropic for trade accounting and client onboarding. Intuit is rolling out AI agents across TurboTax and QuickBooks this spring. These aren't experiments. These are production deployments from companies that serve millions of small businesses and individual filers.
But Dawson sees a buffer that most doom-and-gloom forecasts ignore: regulation. "You still need a CPA to sign the return," he says. "And the government moves at a snail's pace." He's right. IRS rules require a licensed CPA or enrolled agent to sign prepared returns, and that requirement isn't going anywhere fast. "And then when you have issues with the IRS," he adds, "like some AI is not going to come in and solve those issues."
"The IRS is the least fun experience in life. It's rubber glove treatment. It's bad."
That regulatory moat buys time. How much time is the question nobody can answer with certainty. The technology is already capable of doing 90% of the mechanical work. The last 10%, the judgment calls, the client relationships, the signature on the return, that's where humans survive. For now.
Dawson makes an interesting distinction between what he calls the "bits" side and the "atoms" side of the economy. Knowledge workers, the people who move information around, have the shorter runway. Physical workers, the plumbers and electricians and construction crews, have longer. "That's going to be here for a long time until you have robots, which will come very soon," he says. "I'd say in four years we'll have robots doing amazing things."
Industry analysts are watching carefully, but they want receipts. A CPA Practice Advisor analysis from late 2025 argued that 2026 would be the year of "proof, not potential," with firms demanding hard, auditable impact from AI investments: faster closes, cleaner forecasts, measurable savings. The honeymoon period for AI hype in professional services is ending. What replaces it will depend entirely on whether the tools actually deliver.
The Emotional Bottleneck
Ask Dawson why more CPAs aren't doing what he's doing, and he doesn't talk about technology gaps or budget constraints. He talks about feelings.
"If you can just strip away the people who are not using this technology, the question is why," he says. "And under that it's most likely going to be emotionally related."
Fear. Ego. Stubbornness. The sense that everything is moving too fast and the safest response is to do nothing. Dawson has seen this pattern before, not in tech adoption, but in his own clients.
"I have a new client who comes to me and says, 'I haven't paid taxes for four years. I'm freaking out,'" he explains. "It's like, okay, well, we gotta deal with this now. We're not going to put it under the rug." He draws a straight line from that emotional paralysis to the way business owners are responding to AI. "You can kick things under the rug and ignore them. It doesn't make them less true or more true."
The parallel is sharp. A client who hasn't filed taxes in four years isn't dealing with a math problem. They're dealing with an anxiety problem that manifests as avoidance, which creates a bigger problem, which creates more anxiety. The cycle feeds itself until someone intervenes or the IRS comes knocking. Dawson sees the same loop playing out across his profession with AI. The longer you avoid it, the more overwhelming it feels, and the more overwhelming it feels, the longer you avoid it.
His prescription is blunt. "We're really not that smart in the big scheme of things," he says. "And so we have to understand that and then use our...if we have the awareness to understand that we are the bottleneck, we are the weakness, it's not the technology and the LLMs, it's us. Why aren't the answers good? It's because of us. Why aren't we building amazing things? It's because of us. If you understand that, then you can solve for that variable."
That kind of radical honesty about human limitations isn't something you hear from most business owners, especially successful ones. Dawson's argument is essentially that the biggest competitive advantage in the AI era isn't technical skill. It's emotional maturity. The willingness to admit what you don't know. The discipline to sit with discomfort instead of running from it.
"Be humble or be humbled."
He points to the generational divide he sees among peers. His dad, also in the accounting world, has embraced ChatGPT and impressed Dawson with how far he's come. But there are plenty of CPAs his father's age who haven't touched any of it. And younger accountants who use AI casually but aren't, as Dawson puts it, "obsessed." They're reactive instead of proactive. "I think that'll work for a period of time," he says. "Until it doesn't."
According to the Journal of Accountancy, the profession is already grappling with how to train the next generation of accountants when the entry-level tasks that used to build foundational knowledge are increasingly handled by software. It's a chicken-and-egg problem: if junior accountants never do the manual work, how do they develop the judgment to oversee the AI that does it for them?
AI as a Language (and a Child)
When Dawson talks about how business owners should approach AI practically, he reaches for two analogies that stick.
The first: AI is a language. "I look at AI like a language. It's just like Spanish," he says. "It's just like a muscle and you train it and you have to learn it, you have to understand it." This isn't accidental. Dawson learned Spanish in two weeks during a vacation in Peru, a trip that changed the trajectory of his life. He moved to South America, kept working remotely for his US firm, and eventually quit to start his own practice with a single client. That was five years ago.
The language analogy works on multiple levels. Like a language, AI rewards daily practice. Like a language, it has grammar and syntax that feel foreign at first and then become second nature. And like a language, you can survive without it, but you'll always be limited in what you can access.
His second analogy: AI is a child that's growing. "It's like a child that's growing and it's gaining...it's maturing over time." The models get smarter, the capabilities expand, and what was impossible six months ago becomes routine. He started building with one platform two months ago. Last month he switched to something better. Next month he'll probably switch again. The child keeps growing and you have to keep up.
His practical advice is almost comically simple, and he knows it: "The funny answer is, ask the AI that question for yourself." Want to know how AI can help your accounting firm? Ask the AI. Want a plan for implementation? Ask the AI. It sounds circular, but there's real wisdom in it. The tool is the teacher.
What separates Dawson from a casual user is intensity. "Some nights I'm up till 2 AM, 4 AM just going, whether it's vibe coding or building," he says. He describes his relationship with AI in terms usually reserved for religious experiences. "God just put my purpose in my lap. My purpose is AI. My purpose is teaching AI to others, helping others. This stuff comes naturally to me and it's exciting to me."
That kind of passion is rare and probably not replicable for most practicing CPAs who just want to get their returns filed on time. But Dawson argues you don't need to match his obsession. You just need to start. "It's one day at a time. Take bite-sized pieces. It doesn't have to be bigger than it sounds."
Workers who do invest in AI skills are already seeing returns. Multiple labor studies from 2025 showed that professionals with demonstrated AI proficiency command significantly higher wages than peers without those skills. The premium isn't small. The gap between AI-literate and AI-illiterate professionals is only widening.
Dawson's firm, Nick Dawson CPA, is planning to launch online classes and virtual seminars in 2026 where thousands of professionals can learn about the tools he's implementing, see live demonstrations, and start building their own AI capabilities. It's a bet that the market for AI education in professional services is about to explode, and that the people already deep in the trenches will be the ones best positioned to teach.
The Fork in the Road
Industry watchers are starting to describe what's happening in accounting as a two-lane split. Lane one is fast, cheap, automated, and focused on execution. Lane two is advisory, relationship-driven, and rooted in trust. The CPA who thrives in the next decade will use lane one as infrastructure and live in lane two.
Dawson is already there. The parts of his job he loves, working with people, talking tax strategy, meeting new clients, those aren't going anywhere. "All that stuff I was doing as a CPA, like that's the least thing I'm passionate about," he says of the mechanical work. "I love working with people. I'll still get to do that."
What's disappearing is the grind. The four hours spent on tax code research. The data entry. The compliance paperwork. Dawson doesn't miss any of it, and he doesn't think his clients will either.
The question hanging over this entire conversation, and over the profession itself, is whether the majority of CPAs will adapt in time. Not in time to avoid discomfort, but in time to shape what comes next instead of being shaped by it. The firms that figure out AI in the next 12 to 24 months will set the pricing, the service models, and the client expectations that everyone else has to live with. The firms that don't will spend the next decade playing catch-up in a market that's already moved on.
Dawson knows fear is part of the equation. He doesn't dismiss it. "Fear is fine. Fear is okay," he says. "But if you live in the fear, you can't move. You're stuck."
His advice for the CPA who's been watching all of this from the sidelines, the one who knows something is shifting but hasn't figured out what to do about it, isn't complicated. Ask the AI a question. Then ask it a better one. Then ask it something you don't know the answer to. Start treating it like a language you're learning, a few words a day, and trust that fluency comes with time.
The graph paper era ended. The calculator era ended. The spreadsheet era is ending. What comes next is already here, sitting in a WeWork in Bogotá, filing tax returns with a team of agents that never sleep and never ask for a raise.
"Fear cannot help anyone forward in this future," Dawson says. "It will just occupy your mind and paralyze you."
Tax season ends April 15th. After that, the real work begins.
Nick Dawson is the founder of Nick Dawson CPA, a full-service US tax firm. Find him on social media at @NickDawsonCPA. Hunter Miranda is the founder of Automarka Digital, an AI integration consultancy for small businesses.
Published by Executive AI Report
Subscribe to the Briefing
Continue accelerating your intelligence with unfiltered ROI tracking, tool benchmarks, and architectural implementation drops.