How Agentic AI Is Taking Your Clients

We are witnessing AI’s massive destruction of the bookkeeping and accounting business. AI isn’t coming for their jobs. No, what’s actually happening is much scarier: a guy in his pajamas, armed with an AI financial stack, is coming for their market.

This is good news for small businesses. Most wish they could do without these financial services. They see them as necessary evils, right up there with colonoscopies and people who bring acoustic guitars to parties.

So, the other day, I was thinking...

“There must be an AI for that.”

Lo and behold, my friend Hooman Radfar has an answer. And it’s called Collective. It serves the roughly 30 million solo ventures that represent over 80% of all small businesses in the US. They generate $1.7 trillion in combined annual revenue. 

Now, I usually don’t trust people who say "market" and "trillion" in presentations. But Hooman has what we call "The Nose for Money." He co-founded AddThis, which Oracle bought for $200 million, and was an early investor in Uber.

So I am thinking… Maybe I should listen. As far as I am concerned, Hooman can drag me anywhere, and I will follow like Mac MacAnally would Kenny Chesney. If you’re from Jersey, that’s your Tom Morello and The Boss.

Hooman gets the truth about the American economy.

Many businesses are often just one person doing all the work. They pretend to have a "Director of Synergy," but it’s more likely just a freelance guy named Phil in Des Moines.

That’s certainly true for most boutique consulting firms I know. One person doing all the work. And most are not well served by traditional bookkeepers and accountants. Keep in mind, these are folks who say they “classify your receipts and reconcile your books until your accounts balance.” They live on a different planet than the 78% of businesses-of-one, which generate less than $50,000 in annual revenue with less than $10,000 in profit. If their bookkeeper gets $2,000 a year and the accountant charges $1,500 just for filing taxes, their income drops by about 35%. 

I don’t know about you, but I would much rather spend that $3,500 on things that actually matter. Things like a vacation to a place where nobody knows what a "1040-ES" is.

Solopreneurs face an additional issue. Human labor doesn’t "scale" well. It can't keep up with the fast "velocity" of microbusiness transactions. In plain English, this means there are too many tiny numbers moving too fast. The average solopreneur now generates hundreds of microtransactions every month. These are scattered across dozens of Software-as-a-Service platforms, which is a tech-industry term for "companies that take $9.99 out of your bank account every month until you die.”

This is the exact mechanism of disruption.

Suddenly, a solo operator has the same operational horsepower as a 10-person agency.

They don’t have to hold "meetings" or "performance reviews" or discussions about “who stole Phil’s yogurt from the breakroom fridge."

They can just move faster. Their robots handle the heavy lifting. This lets humans focus on what really matters, like checking their Substack every 45 seconds for interesting updates.

Some of them are making really good money. I mean crazy good.

This group includes boutique AI consulting, niche digital marketing agencies, high-end professional coaching, digital content creators, and developers of niche software that scales without adding staff. These businesses thrive by using automation tools and platforms like Gumroad, Stripe, and AI tools to manage tasks that previously required a team.

And guess what? Bookkeepers, accountants, and the IRS grow with them. The more you make, the more you pay. That’s especially true when your company is an LLC. You get taxed twice. You pay personal income tax on net earnings plus self-employment tax of 15.3%. Let’s say your LLC makes a profit of $200,000, you pay about $30,000 in self-employment tax on top of your income tax. As a single-member LLC, you are responsible for both the employer and employee parts of Social Security and Medicare taxes.

That’s where a conversion to an S-Corp comes in. It also gives you the fancy Inc-suffix. 

When you hear "S-Corp election," you might wonder if it's just a tricky tax loophole for rich people. And isn’t it requiring compliance with both corporate formalities and strict IRS regulations? Isn’t that why accountants charge more for supporting an S-Corporation? Is it worth it?

Hooman took that structure and just democratized it with Collective. And the metric that matters here is the average savings. By structuring a one-person business as an S-Corp, solopreneurs save an average of $10,000 a year. They do this by shielding a portion of their profits from that brutal 15.3% self-employment tax.

But I have to ask, why couldn't traditional CPAs do this for freelancers five years ago?

Before AI, the compliance overhead was entirely prohibitive. Running an S-corp required managing payroll, filing corporate tax returns, and maintaining meticulous quarterly bookkeeping.

So if a CPA charged a solopreneur like $5,000 in fees just to maintain the compliance, it cannibalized the $10,000 in tax savings. The net benefit to the client was marginal. And the headache for the CPA was massive.

So the math of doing S-corp compliance manually for a single freelancer used to cost more than the tax savings themselves. Collective reduces those compliance costs.

Their secret weapon is an AI accounting suite powered by GPT.

When a receipt comes in, an Optical Character Recognition system reads it, and the AI categorizes it. This reduces bank reconciliation time by 70% and expense categorization by 90%. It leaves the accountants with a lot of extra time.

Now, there is a catch. Large language models are inherently probabilistic. This means that, left to its own devices, an AI might decide that a $700 dinner at a steakhouse is actually a charitable donation to the Texas Beef Council in Austin. That’s no fun in the eyes of the IRS. It could trigger an audit. If you have ever had one, you prefer a colonoscopy.

To fix this, Collective uses Retrieval-Augmented Generation (RAG). It changes the game.

Instead of making ChatGPT guess, they allow it to use resources. Think of regular AI as a student taking a closed-book test while hallucinating on cough syrup. RAG is that student, but forced to take an open-book test using the actual IRS code. It grounds the AI in reality, so it can’t just make things up.

At Collective, human accountants act as editors. The AI writes the initial draft. For example, it might categorize a software subscription as a deductible expense, and then a real human being with a pulse and a fear of audits verifies it.

Bottom line: Robots are doing the math, humans are checking the boxes. And you finally have enough money left over to go on vacation this year.

If you’re a bookkeeper or accountant, you must reposition yourself. Here’s why and how.

Collective made a super-fast categorization machine. But they missed one small detail: humans are fragile. They are also emotional. An AI can categorize a receipt from a hardware store in milliseconds. It knows exactly where that hammer goes on a spreadsheet. But an AI cannot sense the visceral, lung-collapsing panic of a freelance graphic designer who just received an audit notice and is currently hyperventilating into a Starbucks napkin.

An AI cannot hear the anxiety in your voice.

It doesn't know you’re buying groceries at Costco and categorize that expense as supplies. An accountant would call you out on that. AI won’t.

Or let’s say you’re a freelancer in Texas, a state where you do not pay income tax. You hire a contractor in California for two weeks. Suddenly, because you interacted with California, the California government, with the tax-collecting instincts of a hungry Great White Shark, decides you have a "nexus" there. And that means you must register with the state, file California income tax, pay corporate taxes or fees, and collect and remit sales tax on California sales. Failure to comply can result in back taxes, interest, and steep penalties. The AI wouldn’t catch this. And you might be receiving a $35,000 bill from a state you haven't even visited since 2009.

The moment you replace human judgment with an algorithm just to save a buck, you aren't "scaling a business." You are scaling the churn. You are just cycling through traumatized, angry clients at a much higher velocity.

AI can process a tax return at light speed, but it cannot hold your hand during an audit.

AI doesn't have hands. It has code. And code doesn't care if you have to live in a pup tent because the IRS took your condo.

That’s why Collective is more than just a bookkeeper on steroids. It isn’t just a pile of rogue algorithms living on a server in Palo Alto. It’s a back-office platform with a team of actual human advisors. These are people whose job it is to step in when things get "nuanced," which is a professional term for "the client is about to do something incredibly stupid."

The idea is simple: let AI manage the boring tasks.

Meanwhile, human advisors can focus on tricky issues such as state tax nexus, deduction strategy, and compliance. Does it always work without errors? Does your bookkeeper? Mine once categorized a substantial income as an owner’s draw. Mistakes happen.

My point is, if you’re a bookkeeper or CPA and you’re still charging by the hour, I have some bad news for you: You are currently competing with a piece of software that doesn’t sleep, doesn’t take bio breaks, and doesn’t get distracted by watching cat videos on YouTube.

The Great Accountant Purge

The old way of doing business was simple: You did five hours of boring work, and you charged the client for five hours of boring work. But now, the AI can do that same work in five seconds. If you drop your price because the robot is fast, think again.

The new rule is Value-First Pricing. This means if the AI saves the client $10,000, you still charge them for the $10,000 worth of "Financial Clarity." You aren’t selling your time anymore; you get paid for the consequences you create.

The Team of Three (Because Robots are Socially Awkward)

To do this, you have to fire roughly 1,850 of your 2,000 clients. You move from a High-Volume Grind to Extreme Quality. You set up a Team of Three for just 150 lucky families:

• The Client Service Manager: Their job is regular communication and to interpret "nonverbal clues," which is professional-speak for "realizing the client is lying about their nanny maid being an office worker."

• The Client Controller: They handle historical data, compliance, and act as the safety net catching the state tax nexus errors the AI might misinterpret when it’s feeling "probabilistic."

• The Client CFO: A person who focuses on future strategy to reduce tax payments, increase cash flow, generate wealth, and create key performance indicators that guide entrepreneurial decision-making. Check out Eric Ness. He’s your man.

And if a price shopper comes to your firm and just wants cheap automated bookkeeping without the advisory layer, you actively turn them down.

Rise of the Agents (The Machines Are Getting Pushy)

Right now, AI is Passive. It sits there like a well-behaved dog, waiting for you to give it a receipt. But soon, AI will become an Agent. Imagine an autonomous AI agent that monitors your cash flow in real-time. It sees your revenue drop. So it recalculates your salary, adjusts your payroll, and files the necessary paperwork with the state government.

This transition from passive AI to active agents highlights the future of human expertise.

AI first demands lifelong learning.

If you think learning how to prompt a large language model today is sufficient, complacency will erode your expertise within 18 months.

You cannot afford to wait for agentic technology to perfect itself before you adapt. When an AI can move client funds and handle payroll on its own, the human advisor needs to shift from doing the tasks to checking the AI's logic.

You need to become AI-first. You must restructure how your entire firm works. That does sound like a six-month consulting project that most people just procrastinate on. Remember that every journey starts with a first step. Here’s yours:

The 15-Minute Mandate

• Step 1: Find the most mind-numbingly boring task you do. Identify the single most repetitive mechanical task in your back office that is currently draining your margin. This might include manual data entry, onboarding new accounts, expense categorization, or drafting weekly client status emails.

• Step 2: Find one AI tool to automate that specific bottleneck by tomorrow morning.

• Step 3: Stop acting like a data-entry clerk and start acting like a CEO.

That is the line in the sand. You either master the machine or the machine masters you.

And to leave you with one more final provocative thought to mull over as you begin that 15-minute mandate…

Hooman once predicted that we are headed toward the "Billion-Dollar Business of One." This would be a single human being, sitting at home in their bathrobe, orchestrating 500 autonomous AI agents to run a global empire.

This just happened. Matthew Gallagher founded Medvi, a telehealth provider of GLP-1 weight-loss drugs. On day one, Medvi had 300 customers. By month two, another 1,000 people had signed up. By the end of 2025, he raked in $401 million in sales. Naturally, with a massive, multi-million dollar global empire to run, Gallagher did what any sensible CEO would do: He hired his younger brother, Elliot. These two are on track to hit $1.8 billion in sales this year.

Think about that. A billion-dollar enterprise that isn't a massive corporation with 500 employees and a "Human Resources" department that sends out emails about "Office Refrigerator Etiquette." Just one person, his brother, and a fleet of digital minions.

The "Old Way" is dead.

If your primary skill is typing numbers into a box, you are an abacus in a world of Starships. Are you building the agents or are you competing against them? The hard truth for the accounting industry is this: If your firm mainly offers manual data entry, Collective has just made you obsolete. What does it take for you to adopt the AI-First paradigm?

About Andrew Lawless

AI Strategist & Business Transformation Expert

Andrew Lawless is a key figure in the "AI-First" movement. He focuses on changing how professional services and small businesses operate. Andrew spots market disruptions. He connects advanced technology with people-focused business strategies.

He supports the Business-of-One idea. He champions solopreneurs and promotes the progression from manual work to fast, powerful AI tools. Andrew leverages his AI expertise to help advisors transition from hourly billing to Value-First Pricing.

Andrew’s writing and consulting offer a bold vision for the future of work. He believes human skills should be free from dull tasks like data entry. This way, people can focus on key areas such as judgment, emotional intelligence, and wealth creation.