The Gap Between Small and Big Business Was Always Money

A fifteen-person company can now do what a fifteen-thousand-person company used to do. The press release for it came out last Tuesday.

On May 13, Anthropic launched Claude for Small Business. It is a bundle of fifteen pre-built agentic workflows (payroll, monthly close, invoice chasing, marketing campaigns) wired into the software a small shop already pays for: QuickBooks, PayPal, HubSpot, Canva, Docusign, Google Workspace, Microsoft 365. The price is the part that made me sit up. There isn’t one. The bundle rides on an existing Claude plan with no per-seat upcharge.

OpenAI’s ChatGPT Business runs roughly twenty-five to thirty dollars a seat. Microsoft’s Copilot Business sits around twenty. Google bundles Gemini into Workspace. Anthropic charged zero for the small-business version of the same thing. (When a frontier lab gives away what every competitor is charging for, the gift is not really the product. The gift is how they get to your data.)

The capability gap had a name, and the name was money

For my entire career, the gap between a small business and a big business was budget. A Fortune 500 could afford the analysts, the consultants, the contract attorneys, and the planning software. The corner store and the fifty-person manufacturer could not. (I ran a finance consulting firm for twenty-five years. I helped build that gap. My sons would call that a villain origin story.)

That budget gap is also what kept private-equity rollups working. You bought twenty small companies that could not afford the back office of one big one, you added the back office once, and you sold the bundle. The arbitrage lived in the math of corporate overhead.

Daniela Amodei put the case plainly when she said AI was the first technology that could finally close that gap between small and large companies. Whether she is right is the easy question. The harder one is how fast the closing actually happens.

The launch partners are also the test case

Intuit, which owns QuickBooks, and HubSpot both gave quotes for the announcement. Both are launch partners. Both are also down more than thirty percent year-to-date, because Wall Street has decided that seat-priced software is the thing AI eats first. (My D&D players would never walk this cheerfully into the obvious trap. They have learned to read the room better than the public markets.)

The dynamic is interesting to watch in slow motion. Intuit needs to be where its customers’ AI conversations happen, and right now those conversations are happening inside Claude. Sitting out the partnership would be worse than joining it. So Intuit holds the door open. Claude reaches through QuickBooks for the bookkeeping data, returns answers in plain English, and over the next year or two QuickBooks becomes less of a destination and more of a connector. The destination becomes the model. The accounting software is still there, just smaller in revenue and in role than it was a year ago.

Intuit is doing exactly what the situation requires of them, which is a description of what happens when an interface that costs nothing absorbs the work that used to require an interface that cost something.

A historical rhyme that is not quite a repeat

In 1992, QuickBooks democratized the mechanics of bookkeeping. It took roughly eighty-five percent of the small-business accounting market in its first year. (For context, that is not a market share. That is a market.) Before QuickBooks, a corner store either kept ledgers by hand or paid a bookkeeper to do it. After QuickBooks, the owner kept ledgers in software at the kitchen table on a Sunday.

What is new this week is that the democratization went up a level, from mechanics into judgment. The corner store now also gets the work that used to require a fractional CFO: cash-flow scenarios, vendor-contract review, campaign attribution, hiring math. It will not always reach the level a real CFO would produce, and it will not always be right. It will be better than what most fifteen-person companies get today, which is “none of the above, because there is no budget for any of it.”

Anthropic recently overtook OpenAI in US business adoption, roughly thirty-four percent versus thirty-two. The small-business move is how a frontier lab earns durable share. There are around thirty-six million small businesses in the United States, producing roughly forty-four percent of GDP and employing about half the private-sector workforce. The market here is bigger than most people remember.

What it means for the Office of the CFO and professional services

This is the part I think about most, because it is the world I have spent my career in. The pyramid that runs a CFO’s office (and a consulting firm, and a law firm) was built on leverage. A partner sits on top of senior managers who sit on top of managers who sit on top of analysts who do the work. The economics work because each layer is cheaper than the one above it, and because the analyst layer is doing tasks that require effort more than judgment.

The analyst layer is now optional. (I do not say “gone.” I say optional, because nothing about leading-practice services delivery has gone away in a single quarter, ever.) The work that survives moves up the value chain to judgment, context, and the “so what.” The work that does not survive is the work where the answer was already known by the model the moment it had the data.

The implication for finance and professional services leaders is the same one Anthropic’s product team built for the corner store: competent middle-tier work is approaching zero in cost. Competitive advantage shifts to whoever can ask the better question and interpret the answer in the customer’s actual context.

When every business can afford the same intelligence

The honest close is that nobody knows yet what people compete on when the capability moat finishes draining. Brand, taste, distribution, and a salesperson who can read a room are the obvious candidates. (Also, in my experience, the willingness to make a decision on incomplete information, which is the rarest skill in any company.) “Obvious candidates” is the kind of answer the analyst layer used to write, though, and the real answer probably looks different from what the AI consultants are pitching this quarter.

I do not know what the corner store will spend its newly free judgment-tier capacity on. I do know that the economic premise of the consulting pyramid I built is being repriced in front of us, and that the right response is to keep asking what each business actually needs, instead of what we used to sell them.