Most AI companies sell you a story about being better than the people doing the job today. Rational is not making that claim, and holding back is the interesting part.
It is building an accounting firm with nobody in it. Not a tool for accountants, not an assistant sitting beside one. A firm, staffed entirely by AI employees, handling client intake, the bookkeeping and the filing with the IRS. It claims roughly ninety per cent lower cost and round-the-clock availability.
The bet is not that the machine is smarter than a bookkeeper. It is that the machine is finally as good as one, and that being as good with nobody on the payroll is a completely different kind of business.
What they built
Rational's pitch is unusually literal: the first accounting firm with no humans. The work runs start to finish inside the company, from signing up a client through to filing, and the founders argue that as AI agents become normal and people learn to trust them, they can take over whole industries, including ones built on trust like accounting.
It is based in San Francisco. Jibril Moinuddin, the chief executive, was a founding engineer at Buckeye AI and TrueFoundry and dropped out of the University of Michigan. Christ Xu, the chief technology officer, was the youngest senior software engineer at Tencent and has competed at both the international informatics and mathematics olympiad camps.
Who else is in this field
Automating the books is one of the most attempted businesses in software, and the list of people who have tried is very long. QuickBooks and Xero own the ledger most small businesses actually keep, with FreshBooks, Wave and Sage alongside them. Bench sold software plus humans for a decade and collapsed in late 2024, leaving clients scrambling for their own records. Pilot took the same idea upmarket to funded startups. Botkeeper sold automation to accounting firms instead of competing with them. Puzzle, Zeni and Digits rebuilt the ledger around automation. Ramp and Brex came at the same data from company spending.
Every one of them kept a human somewhere, usually because the client wanted to know whose name was on the work. Rational's claim to being different is not better automation. It is that it took out the person all the others kept.
The point: equal to a person, at nobody
Read the claim carefully and it is a modest one. Rational is not promising accounting a good firm could not do. It is promising accounting that is roughly as good, available at any hour, at a tenth of the price, with no staff.
The aim is not to be better than people. It is to match them and then remove them, which is a far easier target to hit and a far stranger thing to own.
Matching a person while employing nobody produces something with almost no precedent: a working business with revenue, clients and a service to deliver, and no payroll, no hiring, no managers, no office politics, and nobody important who might leave. Every ordinary professional firm is valued at a discount precisely because its assets go home at night and can resign. This one cannot.
What is actually being sold is the setup
Which leads to the thing worth taking from this file, offered as a prediction rather than as reporting.
A firm like this is built, whether the founders put it this way or not, as something to be bought. Not a product to license and not a team to hire, but a complete working machine somebody else can buy and run.
Think about what the buyer gets. A sign-up process that answers at any hour. Email and letters that never fall behind. The document handling, the reconciling, the filing calendar. Someone to answer the phone, someone to book things in, someone to chase. None of that is accounting expertise. All of it is the dull machinery that makes a professional firm actually work, and it is the part that takes years to build and is nearly impossible to staff reliably.
The valuable thing is not the bookkeeping. It is the whole running machinery of a firm, sold as one item, with the staffing problem already solved.
And the buyers are everywhere. Regional accounting practices that cannot recruit. Groups buying up small firms. Established payroll and bookkeeping companies with plenty of customers and ageing operations. Private equity firms already buying accountancies for the steady revenue and immediately running into the staffing problem. Any one of them would rather buy a working firm with no staff than try to build one.
The zero-people firms that come next
If this works in accounting, the same shape fits anywhere the business is documents, deadlines and correspondence rather than physical work. Some obvious candidates:
- The zero-people ad agency. Briefs, concepts, copy, media buying and reporting, with no creative department.
- The zero-people recruitment agency. Finding candidates, screening them, booking interviews and keeping in touch, which is nearly all correspondence already.
- The zero-people translation firm. The closest to solved of all of them, and still sold today at agency prices.
- The zero-people property management company. Tenant requests, sending out repair people, chasing rent and issuing statements, with contractors doing the physical work.
- The zero-people medical billing firm. Coding, submitting, appealing rejections and invoicing patients. Pure paperwork.
- The zero-people freight broker. Quoting, booking, tracking and settling up between shippers and hauliers.
- The zero-people market research firm. Survey design, panel management, analysis and reporting.
- The zero-people PR firm. Media lists, pitching, monitoring and coverage reports. It looks like a relationship business and is mostly admin.
Some of these need a licence and some do not, and that difference decides how hard each one is. The ones that need no licence could be started tomorrow by anybody.
Which is why this is open to people who are not engineers
Here is the part that should interest anyone reading this who does not write code.
Building one of these does not require inventing anything. You buy the models. The tools are off the shelf. What it does require is knowing, in detail, how a particular kind of firm actually works: what clients ask for, what the deadlines are, where the work goes wrong, what a good result looks like. That comes from having worked in the industry, not from having studied machine learning.
If your field is something other than computer science, that is now the rare half of the pair. The engineering is getting cheaper and more available every month. Knowing an industry inside out is not, and no clever person picks it up in a weekend.
This is about as close to easy money as the moment offers: take the trade you already understand, rebuild it with nobody in it, and sell the result to the people who already have the clients.
The catch, and it is a real one
None of this gets rid of the question that has always made professional services valuable.
What a client buys from an accountant is not just the sums. It is a name on a signature line, attached to a licence that can be taken away, a person who can be sued, and an insurer standing behind them. The rules are built around exactly that, by holding individually licensed preparers responsible. A firm with no humans has to say who carries that, and a buyer has to answer it too, because they inherit it.
That is the unresolved thing at the centre of this company, and it is why the answer matters beyond accounting. If the guarantee can be put up as money rather than as somebody's career, then every licensed profession whose fees are explained by expertise but really bought as insurance becomes reachable. If it cannot, the zero-people firm stops at the edge of the regulated trades and does very well everywhere else.
Why it is bigger than it sounds
The short description is an AI accounting firm that is ninety per cent cheaper. The structure underneath it is the more valuable claim.
- A firm with no payroll and nobody who can resign is a structural asset. That is a different kind of object from a cheaper bookkeeper, and it is what makes the company buyable rather than merely useful.
- Parity plus zero headcount is the clever target. Matching a good accountant and removing the staffing is an easier goal than beating one, and a far stranger asset once achieved.
- Whoever answers the signature question first defines the category. Who signs off is the live frontier here, and it is a frontier with very few people standing on it.
What to watch
The number that settles it is not customers or cost savings. It is whether Rational can file at volume and stand behind what it files: how many returns go out, how many get questioned, and what happens when one is wrong.
The second thing to watch is the sale. If a zero-people firm gets bought by an established player inside the next two years, that deal will be studied closely, and a great many people will set out to build the next one.