Why AI Needs CPA Verification — And Why the CPA's Moment Is Now

The accounting profession is being told, in many places, that AI is its replacement. The framing is wrong. AI is not replacing the profession. AI is clarifying it.

What AI is actually replacing is labor. Specifically, the repetitive and organizational labor that historically consumed enormous amounts of time inside accounting, diligence, and financial preparation work. AI can now normalize three years of financials in minutes. It can identify likely add-backs, summarize customer concentration, draft reconciliations, organize workpapers, and produce analyses that would previously have required teams of junior accountants and analysts. The productivity improvement is real and irreversible.

But the emergence of AI is simultaneously exposing something that was always true, though less visible before:

Generating financial representations and verifying financial representations are fundamentally different activities.

AI can generate representations extremely well. It cannot stand behind them.

The Division of Labor AI Is Forcing Into the Open

The pattern is already showing up in its smallest form. A CPA opens a draft prepared by an AI tool: an AI-assisted bookkeeping reconciliation, an AI-summarized cash flow analysis, an AI-generated normalization schedule. The work is competent. The CPA reads it, applies professional judgment, and then checks it against the underlying records. That check is what makes the document mean anything to the bank, the buyer, the lender.

The AI did the labor. The credentialed human carried the accountability.

A CPA license and credibility can be lost. AI gets a pass.

Markets, courts, lenders, regulators, and counterparties cannot accept output from a system that has no professional standing to lose. A CPA can lose a license. A CPA carries professional liability coverage. A CPA is bound by AICPA ethics rules and state board oversight. A finding reported by a CPA is backed by a real person whose standing is at risk if the work is wrong. That is what makes it carry weight.

AI has none of those properties. It cannot be sanctioned. It cannot be sued. It cannot be peer-reviewed. The acceleration of AI sharpens those facts rather than dissolving them.

The more capable AI becomes, the more valuable professionally accountable work becomes alongside it. The two are not in competition. They are a system, and the system only works when both are present.

The Information Gap AI Is Exposing in SMB Transactions

The same pattern that plays out in a routine reconciliation plays out at much higher stakes in an SMB transaction.

Buyers today encounter businesses presented with a level of polish that would have been impossible only a few years ago. Financial summaries are cleaner. Narratives are more coherent. Confidential Information Memorandums are more persuasive. AI dramatically lowers the cost of producing convincing representations about a business. But polished information is not the same thing as checked information. It can be a distraction from what actually agrees to the records.

A buyer evaluating an operating business is being asked to commit capital against representations no credentialed party has checked. The information environment has grown louder and more confident. The actual trust floor has not moved.

This is the structural gap AI is exposing. The solution is not less AI. The solution is combining AI, data, and credentialed human accountability into a coherent system that other humans can rely on. AI becomes the productivity engine underneath the work, while the CPA remains the load-bearing accountability layer above it.

The CPA's Work Is Already Everywhere in the Book

Here is the thing CPAs reading this already know: the clients who need this work are already in the existing book of clients.

Not new clients arriving from somewhere. Existing clients whom the CPA already knows, whose situations the CPA has been observing for years, who will need verification work in the next one to three years.

The sixty-something founder approaching retirement with no clear succession plan. Built the business over decades. Kids aren't taking it over. They've been talking vaguely about "selling someday" for years without doing anything operational about it.

The mid-fifties owner with the unsolicited offer. Business doing well. A competitor, strategic acquirer, or search fund just made an approach. The owner is intrigued by the interest but doesn't know how to evaluate it. They call the CPA: "Should I take this seriously?" The financials aren't structured to present the business at its actual value. The financials weren't built to answer that question.

Every CPA reading this can name specific clients who fit these and other familiar archetypes within seconds. Twelve million baby-boomer-owned SMBs are approaching exit over the next decade. Nearly all of those owners already have a long-standing CPA relationship. Most buyers already want financials checked against evidence before committing capital. The work that connects all those realities has existed in theory but was never structured with the CPA at the center.

This is not occasional one-off engagement work when a client finally decides to sell. This is a recurring verification line for CPAs, running across an entire book of clients who will face transition in the next few years. Each of them can bring checked disclosures to the market, so their stated figures carry findings a buyer can rely on rather than representations no one has stood behind.

Why This Work Belongs to a CPA Who Knows the Business

There is a reason this work fits a CPA close to the business rather than a cold third party or an AI system, with one structural requirement that protects its integrity.

The binding constraint is not labor. AI can perform the labor of analysis efficiently regardless of who runs it. The binding constraint is context.

A CPA who has spent years close to a business knows which customer concentration figures are real and which reflect a single accounting quirk. They know which owner add-backs will defend themselves and which won't survive a buyer's question. They understand the inventory practices, the contract patterns, the personal expenses that have been running through the books, the relationships that drive revenue, and the operational realities that no document fully captures.

That contextual knowledge is what makes the procedures efficient and the findings accurate. No buy-side QoE provider arriving cold can replicate it. No AI tool can generate it.

That context, multiplied by AI's labor capacity, multiplied by the next decade of SMB transitions, is what makes this a generational opportunity for the CPA profession. Now.

The Framework to Do It Already Exists

The CPA's professional liability community has correctly trained a defensive reflex: decline improperly structured verification requests. The training is right. But the same reflex has obscured something important. Verification work, properly structured and engaged early, is some of the most valuable work CPAs can perform. And the framework to do it correctly has existed since 2021.

SSAE 19, governing Agreed-Upon Procedures, is exactly the right structural fit for pre-transaction verification work in SMB exits. The scope is bounded. The procedures are defined in advance. The CPA is not opining on financial statements as a whole, not providing assurance of any kind, not absorbing the open-ended liability of a comfort letter. The agreed-upon procedures produce findings, not an opinion. The work is billed at the firm's standard rates as an attestation engagement, with no contingency on whether the underlying transaction closes. That fee structure is consistent with AICPA Rule 1.510 (Contingent Fees) and Rule 1.520 (Commissions and Referral Fees); the CPA's attest independence is governed separately by ET 1.200, ET 1.295, and the AUP interpretation ET 1.297.020.

In practice, the CPA runs agreed-upon procedures on the evidence behind the specific disclosures buyers most often ask to see checked. Stated revenue and customer concentration figures agreed to source records. Owner add-backs and add-back schedules agreed to support. Stated balance sheet items agreed to documentation. Financing-related figures agreed to the underlying detail. Each disclosure has defined procedures, defined evidence requirements, and defined reporting language, and each produces a finding: the stated figure agreed to source, or an exception. Where a buyer's question is legal rather than financial, such as contract transferability, that belongs to counsel, not the CPA. These financial disclosures are exactly the ones a CPA close to the business is positioned to check better than anyone else in the transaction ecosystem. The professional framework already exists. The clients already exist. The moment to bring them together is now.

What Changes When Verification Exists Upfront

When a seller brings checked disclosures to the market when they are ready, the entire shape of the transaction changes.

The buy-side QoE process becomes confirmatory rather than investigative. Months of discovery collapses to weeks. The buyer's question shifts from "is what I've been told true?" to "given which figures already agreed to evidence, what is this worth to me?" Closing timelines compress. Deal failure rates drop. Valuation gaps narrow.

The CPA doesn't get the Friday afternoon call asking for magic. The seller doesn't arrive hoping a broker can keep the deal alive through diligence. The buyer doesn't absorb the cost and delay of discovering what should have been surfaced before the LOI.

The seller negotiates from disclosures already checked against the records, which gives the conversation about value and structure a documented baseline rather than a set of unverified claims.

AI Does the Work. The CPA Reports the Findings. The Platform Holds Both.

The practical architecture that makes this work at scale is a specific division of labor.

The CPA uses whatever tools serve the engagement with their client under SSAE 19: AI-assisted analysis, automated reconciliation, intelligent data organization. The professional judgment, the interpretation, and the standing that make a finding mean something belong to the CPA.

SoForma is the platform where sellers engage an independent CPA for this work, and where they share what was checked, when they are ready, with buyers who want the truth. CPA-Verified status, earned disclosure by disclosure, is what lets a buyer rely on a stated figure instead of taking it on faith.

The Profession's Moment

Every category of commerce that depends on checked information is going to face the same division of labor in the next few years. The categories that get it right will compound. The categories that try to remove the credentialed human, or pretend AI alone is sufficient, will produce volume that no serious counterparty can act on.

Information is cheap now. Findings a credentialed human stands behind are the thing that scales. And that requires a credentialed human standing behind the work.

The CPA's role is not being displaced by AI.

Now is the moment for CPAs to claim what they've earned.

SoForma is the verification platform building the CPA-Verified SMB^™ category. The CPA performs work that didn't exist as a productized category before, in a structure that respects their professional standing, with AI accelerating the labor underneath and each finding recorded permanently in the platform.

See also: Tell the Truth: "CPA-Verified SMB^™" Is What Buyers Always Wanted: the foundational thesis. The CPA's Moment: From Risky Comfort Letter to CPA-Verified SMB^™: how the profession claims this category. What CPA Verification Is — and What It Isn't: distinguishing verification from adjacent work.

Michael d'Amato is the founder of SoForma, trust infrastructure for CPA-verified SMB exit preparation. Based in Miami.