A $1.15 Billion Startup Sells AI to Your Accountant. We'd Rather Replace the Accountant.

$263 million was raised in one month to sell AI to accounting firms. None of it helps you, the person paying the accountant. That gap is the opportunity.

$263M raised to sell AI to accountants. $0 raised to replace them.

In February 2026, four companies raised a combined $263 million to bring AI to accounting.

Basis raised $100 million at a $1.15 billion valuation. Accrual launched out of stealth with $75 million from General Catalyst. Fieldguide closed a $75 million Series C led by Goldman Sachs at a $700 million valuation. Accordance pulled in $13 million from Khosla Ventures. And Pilot announced what it called the world's first fully autonomous AI bookkeeper.

That's a quarter of a billion dollars in a single month. For AI that does accounting work.

But here's what every press release has in common: the customer is the accounting firm.

Basis works with 30% of the top 25 US accounting firms. Accrual partners with H&R Block and Armanino. Fieldguide is used by half the top 100, including KPMG. Accordance targets CPA teams. Even Pilot — the closest thing to a consumer product — runs as a managed SaaS service where the AI works for Pilot, not for you.

Not one of these companies sells directly to the self-employed person who actually needs the bookkeeping done.

The Accountant's Margin Is the Automation Opportunity

This isn't about whether accountants are competent. Most of them are. The problem is the business model.

Here's how it works for most self-employed people: you collect your receipts, sort your expenses, organize your documents, track your own deadlines, answer your accountant's questions, then pay the accountant to file what you already organized. The accountant's main value is knowing where to put the numbers and when to submit them. That's not relationship work. That's clerical work with regulatory knowledge.

Basis just proved that AI can complete an autonomous 1065 tax return end-to-end. Their own press release says it. So the technology clearly works. The question is: who benefits?

When Basis sells that capability to a CPA firm, the firm gets faster. Maybe they handle 30% more clients. Maybe they fire a few junior associates. But you — the freelancer, the consultant, the solo operator — you still get the same invoice. Maybe the firm passes along 10% savings. More likely they don't.

The AI made the accountant's business more profitable. It didn't make your accounting cheaper.

$263 Million to Protect a Middleman

A quarter of a billion dollars, in one month, to make accounting firms more efficient. Goldman Sachs, General Catalyst, Accel, Google Ventures, Khosla — serious investors making serious bets.

And every single bet is on the same thesis: AI should help accountants do their jobs better.

Nobody funded the alternative thesis: AI should help you skip the accountant entirely.

There's a reason for that. Enterprise sales are predictable. KPMG signs a seven-figure contract. A top-25 firm deploys across 500 accountants. The ARR math is beautiful. VCs love it.

Selling a $79 product to a freelancer in Berlin who doesn't want to deal with Finanzamt? That doesn't show up in a Series B deck.

But the freelancer in Berlin and the consultant in Zurich and the designer in Melbourne and the photographer in Brooklyn — they all have the same problem. They're paying someone thousands a year to process paperwork they already did most of the work on. And the technology to replace that process exists. It's just pointed at the wrong customer.

The Talent Crisis Is a Feature, Not a Bug

Here's the statistic everyone in accounting media is talking about: 300,000 accountants left the US profession in the past two years. CPA exam candidates are at a 17-year low. 75% of current CPAs are approaching retirement. Fieldguide estimates a 125 million hour capacity shortfall that'll grow to 600 million hours by 2030.

The industry consensus is clear: this is a crisis, and AI will save the profession.

We think they have it backwards.

If you're a self-employed person, the accountant shortage isn't your crisis. You're not the one struggling to hire CPAs. You're the one paying them. And if there are fewer accountants and the same number of clients, what happens to prices? They go up. What happens to responsiveness? It goes down. What happens to that anxious feeling that your accountant might miss something because they're stretched across too many clients? It gets worse.

The shortage means the service is getting worse and more expensive at the exact moment the technology exists to do it yourself.

300,000 accountants quit. The profession calls it a talent crisis. We call it a market signal.

What "Replacing the Accountant" Actually Means

We should be specific here, because "replace the accountant" sounds dramatic and people hear "do your own taxes with ChatGPT." That's not it.

What we actually mean: an AI agent that runs on your own machine, talks to you over chat, processes every document the day it arrives, maintains your ledger, tracks your deadlines, knows your contacts and correspondence history, builds a knowledge base of your jurisdiction's tax rules, and at year-end hands you a clean, organized, categorized filing that's either ready to submit or ready for a tax advisor to review in 30 minutes instead of 30 hours.

That's not hypothetical. We built exactly this and use it for real bookkeeping. The agent processes phone bills, electricity invoices, insurance statements, and tax office letters. It files them to cloud storage, categorizes expenses, applies business-use percentages, updates contact records, and monitors deadlines. It costs about $15 a month in API fees to run.

The hard part wasn't the AI — it was getting the AI to be reliable. To remember that the SVA already sent a letter about this. To know that quarterly payments are due even when you forget. To run startup checks every time it boots and flag anything overdue before anyone asks.

Boring, predictable, reliable. Not impressive. Not AGI. Just an agent that does the same clerical work every session and doesn't forget things between sessions.

Why Self-Hosted Matters Here

Every AI accounting product launched in February runs in someone else's cloud. Your bank statements, your tax returns, your salary slips, your client contracts — they all pass through a third-party API at minimum, and in most cases live on infrastructure controlled by a company whose business model depends on keeping you subscribed.

Think about the business model. Pilot is a managed service. Basis licenses to firms. Accrual charges per engagement. These are all subscription or per-seat models. You pay as long as you use them. The AI gets cheaper to run every month — inference costs dropped 10x in the last year alone — but your bill stays the same or goes up. Sound familiar? It's the same deal as your accountant, just with better branding.

A one-time purchase, self-hosted agent doesn't have this problem. You buy it once. You own the code. When API costs drop — and they will, they always do — you capture the savings directly. Your financial data stays on your server, in your storage bucket, under your control.

The Numbers Keep Getting More Absurd

The AI infrastructure to run a personal bookkeeper: maybe $5 for a VPS, $5-15 in LLM API calls depending on document volume, free storage on Cloudflare R2. Call it $15/month. That's $180 a year.

What Basis sells to a CPA firm to do the same work for each of their clients: enterprise software pricing. The firms then charge their clients the same rates — $150-450/hour, depending on jurisdiction and complexity.

What most self-employed people pay their accountant: $1,500-4,000/year for a straightforward freelance business. More if you're in an expensive jurisdiction.

The technology that Basis proved works — autonomous tax return completion — costs pennies per document to run. But by the time it passes through the accounting firm and reaches you as a service, those pennies become thousands.

That's not a market inefficiency. That's a middleman tax.

Who This Is Actually For

Not everyone should run their own AI bookkeeper. If you have a complex corporate structure, dozens of employees, international transfer pricing — you probably need a human specialist. For now.

But if you're a freelancer. A sole proprietor. A consultant. A contractor. Someone with straightforward income, normal expenses, and a filing obligation that stresses you out more than it should — the $263 million that just got raised isn't for you. Those products will make your accountant faster, and your accountant will keep your invoice the same.

The alternative exists. It runs on your hardware, speaks your language, learns your jurisdiction's rules, and costs less in a year than your accountant charges for a single session.

The market funded the wrong side. We're building for the other one.

Taxclaw

A self-hosted AI bookkeeper. Any jurisdiction. Any LLM. Your server, your data. $79, once.

See what it does →

$263 million to upgrade accountants' tools. $79 to skip the accountant. The math isn't complicated.