The AI Speed Trap: Why Adopting Faster Isn’t Always Better

Small accounting and bookkeeping firms are told to adopt AI fast. LiveCA co-founder Chad Davis, AI educator Jan Haugo and leaders at Isaacs & Associates, BKC and VP CPAs explain why deliberate, human-in-the-loop adoption delivers better results, and how to tell if you’re rushing. Part of Two Cents Magazine Issue 02: Artificial Intelligence.

It’s morning in Spain, and Chad Davis is in no hurry. The kids have been dropped at school. He takes a walk with his wife, then makes breakfast, then runs a few errands. By the time North America’s time zones catch up, he’ll have spent three to five hours this way—a stretch of the day he calls a “cheat code.”

It’s a surprisingly unplugged scene for one of the most tech-savvy people in the accounting industry. Davis is the co-founder of LiveCA, a firm that’s been fully remote since 2012, and the founder of AutomationTown, a community where accountants learn to build with AI. If anyone were going to preach about tech-enabled speed, you’d expect it to be him.

Instead, he says, “I don’t use automation to be faster. I use it to strengthen the parts of the job that allow me to enjoy the slow part.”

In other words, Davis isn’t automating to cram more into the day. He’s automating to protect the part of it he’d rather not hurry through.

That’s the paradox at the heart of this story. At a moment when accounting firm leaders are being told, relentlessly, to move faster, the people leveraging this technology most effectively are often the least interested in speed for its own sake. It begs the question: what do they know that the rest of the profession seems to be rushing past?

The Pressure to Move Fast

If you’ve opened an email inbox, a LinkedIn feed, or a conference brochure lately, you already know the pitch: AI is moving fast, and you need to keep pace or risk being left in the proverbial dust.

Shahram Zarshenas, Financial Cents co-founder and CEO, sees this every day. In his most recent founder’s note, he references the billions of dollars pouring into AI and the alarmist predictions about the accounting profession becoming unrecognizable by 2030. He rattles off all the ways accountants and bookkeepers are being bombarded with AI hype:

“Another webinar here. Another certification there. Another course, another conference track, another tool with a prompt box in the corner and a blinking cursor begging for attention.”

Shahram Zarshenas

What Zarshenas hears from firm leaders who are in the midst of that noise is not excitement, but a tone “that straddles the line of guilt and exhaustion.” One sentence he’s heard “dozens of times, from dozens of people who are exceptionally good at their jobs” is, “I know I should be doing more with it.” These are people who are running healthy firms with loyal client bases—and who’ve built their teams “with care and intention.” Yet somehow, the AI conversation has convinced them they’re failing at something.

But the data tells a different story. In Financial Cents’ survey of 486 bookkeeping and accounting professionals, 95% of firms are already using AI in some capacity. They aren’t laggards.

So what’s actually in the way of firm leaders getting more out of AI? The usual suspects are cost, a reluctant team, or doubt that AI will pay off, but the report says otherwise. Cost was cited by just 6% of respondents, team resistance by 3%, and unclear ROI by only 2%. The runaway winner, at 41%, was time: specifically, not enough to properly learn and implement AI.

That’s the central irony. Firms want AI to give them hours back, but they can’t spare the time necessary to adopt it. “Things are evolving SO FAST that it is almost impossible to keep up,” one firm owner told us. “It would be a full-time job.”

The noise, Zarshenas points out, assumes the problem is that you haven’t learned enough. “But the actual problem is that your week is full and your mental bandwidth is tapped out,” he says.

This exposes the flaw in the urgency narrative. Telling a firm to hurry puts unproductive pressure on the exact thing that takes time, attention, and care: adoption.

Speed Isn’t Competence: The Risk of Confident AI Answers.

Davis understands the great advantage that speed gives. A small firm can have an idea on Monday, put it in front of the team on Tuesday, test it on Wednesday, and have results by Friday. That kind of immediate feedback is one reason small firms are so well positioned to adopt AI. But there’s a catch.

People perceive speed with competence,” Davis says.

Chad Davis

In a small firm, he explains, that perception can be especially costly, because you often don’t have the bench depth to back it up. You may not have a transfer pricing expert, for example. If someone gets a transfer pricing answer out of an AI step in the process, that doesn’t make them qualified to advise on it. Seasoned people who’ve seen the consequences know to stop, but someone newer to the workforce may not. The answer arrives quickly and reads well, and the person acting on it doesn’t know enough to question it.

One firm owner who responded to the Financial Cents AI survey shares Davis’s viewpoint: “AI can produce answers that sound extremely confident, but confidence doesn’t always equal accuracy.” And the numbers show the profession is well aware of that fact. Only 19% of firms trust AI output enough to use it with limited review.

The way out for Davis is a discipline he applies to every tool he implements. “We can’t conflate or confuse authority with access,” he says. Just because you can give an AI access to your practice management software, for example, doesn’t mean you should give it the authority to change what’s in there.

Small firms, he says, tend to get this wrong by going further, faster. They try to automate a whole workflow when they should stop at the 20% mark, pull the data, and run the new process in parallel with the old one for a while. That way, they can see where the edge cases pop up and how the AI performs, and only after that “slowly give it authority.”

The reason is that “the first version is always the easiest,” Davis says. The cost comes in maintaining it, and a build that doesn’t fit the way your firm works can damage your reputation and your team’s trust in you. His preferred approach is “calm, slow, human in the loop”—built on proper development principles. The alternative, he says, is a mistake.

It’s not as easy as just asking AI for a script and all of a sudden everything’s automatic,” Davis says.

Chad Davis

Rushing can end the journey before it begins. Davis compares a person’s first experience with AI to getting a driver’s license and being handed a really bad car. “It has bad power steering, and you’re like, ‘No, I don’t want to drive anymore, because that was a horrible experience.’”

Meanwhile, the car with better steering and suspension “could be the most enjoyable thing you’ve ever experienced on the road.” The trouble, he says, is that you don’t always know the other car exists. That’s how a quick, shallow try can turn someone off for good.

The aforementioned survey suggests many people are in this position: 96% of firms using AI rely mainly on general-purpose chatbots, and only 20% use anything built for accounting work.

Moving fast can look like competence, whether it’s the confident answer, the quickly built automation, or the easy-to-use generalist tool. Beneath the surface, though, a hurried approach can harbor real risk.

Why Rushing Backfires

Rushing feels like progress. You’re doing something, trying tools, producing output. But two experts who know this technology well say, for different reasons, that much of that motion is wasted.

Jan Haugo spends more than 40 hours a week studying, teaching, and building with AI, and she has a blunt view of the do-it-all-at-once approach.

You can’t do it all at once, nor do you want to, because I think that’s kind of a train wreck,” she says.

Jan Haugo

Another common mistake, Haugo explains, is chasing one-off tasks. She points to a peer in the field who recently wrote that after a full year of working with AI, he hadn’t “moved the needle. He’d had a great time playing and building, she recalls, but he’d spent the year focusing on quick, individual tasks.

People go in, and they just think, this quick thing, this quick task,” Haugo says. “That’s the wrong way to look at it.”

Jan Haugo

Her prescription is to build systems rather than collect tasks. “Don’t just say, ‘I’m gonna write this letter,’ or, ‘I’m going to analyze these financials,’” she says.

Instead, Haugo advises, “Put your workflows in, create your systems, create your operating centers.”

Jan Haugo

And documentation comes first: “If you don’t have anything documented, you’re not gonna go far.”

This was a costly lesson for Davis. In 2019, LiveCA hired a developer and put four or five employees on a project he’d been thinking about for years: pulling all of the firm’s spreadsheet processes into one app. It was, he says, “very expensive.”

They scrapped it. “It would cost so much money to get incremental gains,” Davis says. “The timing was off. We didn’t have AI; the quality wasn’t there.”

The failure turned out to be useful, though, because it showed him how much the firm couldn’t fully trust its own data. “If you’ve been in an accounting firm for a long time, the whole question of ‘is that right?’ comes up way too often,” he says. What they were missing was centralized data that served as the source of truth. “So we spent years making that source of truth believable.” Only after that were they able to build the tools that now help run the firm.

When AI arrived, and the idea came back around, Davis went about it differently.

You can’t go all-in. You just can’t try to replace 30 things,” he says. “You have to start a little bit smaller.”

Chad Davis

Neither story is about doing too little. Haugo is up to her neck in AI, and Davis was early with automation long before AI made it easy. What they describe is effort aimed at the wrong target, such as a stream of disconnected tasks or a big build on a foundation that wasn’t ready.

A clever prompt might save a few minutes today. A well-designed process can make the same automated action more reliable every time it fires. But getting there requires deciding which problem matters, understanding the existing workflow, and giving a solution enough attention to become repeatable. That takes focus, which is what the pressure to rush could erode.

The Case for Pacing with Intention

If rushing backfires, what does the alternative look like? The three practitioners below approach AI adoption in different ways, but they hold tight to the same underlying principle.

Rebecca Isaacs started the year intending to bring AI into every corner of her firm, Isaacs & Associates. Then she got into the details and realized “you can’t build Rome in a day.” So she changed the plan. Isaacs now picks two or three things a year to change, with the goal of being fully integrated within three years.

It might sound modest, but the logic checks out. Two or three changes a year over three years is six to nine real changes, each chosen, tested, and absorbed before the next one starts. And the three-year horizon isn’t arbitrary. It lines up with her son finishing college, when she plans to hit the road and work from anywhere.

Her advice to other firm owners follows the same chain: do your homework, spend where it counts, and add things “a little bit at a time, because you can’t do it all at once. It’s too much.”

Patti Storms, who supervises BKC’s client accounting team, approaches new technology with a habit she’s had for years. She doesn’t buy the first release of anything, including phones.

I just as soon wait until they work the kinks out,” she says. “And then I’ll get in there when it’s much better.” The reason is simple: “I’m not interested in being first.”

Patti Storms

Storms also sees a role for those who go first. “Some firms will want to use everything early,” she says, and they’ll be able to give “great feedback to the people who have decided to wait.”

Bri Como, manager of operations and systems at VP CPAs, has built plenty with AI. She’s also thought hard about where to stop. Her philosophy is moderation, and she puts it in terms of dessert.

If you eat a whole cake every day, that’s not good for you. If you have a slice of cake a week, whatever.” The goal, she says, is “don’t use it for everything all the time, but have it kind of limited and in moderation.”

Bri Como

She has a more personal reason, too. At a past job, she leaned on AI to help produce blog content, and she came away with a feeling she didn’t like. “I felt like it made me dumber not having to think,” she says.

All three have decided what to change, how much they can take on, and where the technology really helps, instead of letting the pace be set for them.

Slow Is the On-Ramp to Fast

It would be easy to read everything so far as a case for staying slow, but that’s not quite accurate. The people making the case are, for the most part, pursuing speed too. They just don’t treat it as the end all, be all.

Haugo calls the profession a bunch of turtles, but she doesn’t mean it as an insult. “We go real slow when we’re working to adopt things,” she says.

She’s mindful of the mental shift that has to happen before anything speeds up. A lot of people, she says, use AI the way they’d use Google or stop at drafting emails. The shift is to see AI more broadly. “I keep saying AI is a collaborator; it is not a tool,” she says. Getting there takes work. Haugo says it took her a long time to get to see it that way.

When you make that mental shift, you will take off like a rocket,” Haugo says.

Jan Haugo

Haugo is no enemy of speed. In her words, today is the worst AI will ever be, and learning from a community, rather than alone, is how you “get yourself there faster.” But speed is what you get once the foundation, the mindset, and the habits are in place.

Davis describes what that payoff looks like in terms of compounding interest: Learn a skill, apply it to a problem, and the time it saves goes toward learning the next one. A task that once ate 15 hours a month and takes 15 hours to automate has paid for itself in a month, and the person who built it comes out smarter, too.

Davis has seen this compounding effect firsthand. At AutomationTown, he’s watched people with no idea how to build an app show up two or three weeks later with something working.

The results are even more specific at his firm—LiveCA. A bookkeeping file that used to take 10 hours a month now takes two or fewer, and the time it frees up doesn’t necessarily go to more work. It’s enabled them to have better conversations and spend a little extra time with clients.

Davis ties it all back to how he lives. “I get peace from learning new skills so that I can apply them later and have more peace,” he says. “It compounds over time.”

The survey data points in the same direction. Firms reporting the greatest returns went deeper, not faster: 56% of firms running AI across the organization report measurable ROI, compared with 4% of firms where a few people dabble. Those firms also tend to have things like organized training and a documented AI policy.

Going Slow, on Purpose, Is the Edge

With AI adoption, Zarshenas says, “Speed is the easiest thing to sell and the least interesting thing to deliver. Quality and depth, on the other hand, are what truly differentiate your firm from the one down the street.”

Speed is something any firm can buy, and anyone can copy. Judgment, trustworthy data, a team that knows when to question a confident answer, and clients who feel looked after are much harder to come by, and they take time to build.

Which brings us back to Spain, and a man out for a walk. Davis could use everything he knows to run faster. Instead, he has used it to reclaim his mornings. For him, going slow is the product of doing the work in the right order.

If you’ve been carrying the feeling that you should be doing more, Zarshenas’s message for you is simply, “You are not behind.” Going slow, on purpose, may be the edge most firms are missing in AI adoption in 2026.

Be among the first to know

Enter your email address to receive new issues of the magazine straight to your inbox.