For accounting practices
The AI-native practice.
Your capacity stops scaling with headcount.
Every practice has two layers: the production work and the judgment. Bookkeeping, reconciliation, chasing paperwork, drafting returns on one side. Reviewing, advising and signing off on the other. AI-native means the first layer runs itself, so your team's time goes where the fees are.
Built for the whole client book. Start at the pace that suits your practice.
Recruit Finn
One accounting agent. Three ways into your practice.
Use the complete AccountsOS practice platform, bring Finn into the AI tools your team already uses, or put the accounting engine behind your own systems. The interface can change. The governed accounting work underneath stays the same.
In short
An AI-native practice is one where the production layer of the work, the bookkeeping, reconciliation and first drafts, is done by AI in the background, and qualified people spend their time on review, advice and sign-off. On AccountsOS that layer is Finn: it works every client's books continuously, prepares VAT returns and annual accounts across the whole book, and surfaces only the exceptions that need a human. The accountant stays the accountant. The difference is what the practice's biggest cost line is spent on, and what happens to margin when taking on ten more clients no longer means another hire.
How it runs
The production layer, running itself
The day-to-day of an AI-native practice on AccountsOS.
Everything feeds in
Bank feeds, receipts by WhatsApp or email, invoices, statements, Xero imports. Clients throw paperwork at Finn; nobody keys it in.
Finn does the work
Categorises, reconciles, chases what is missing, closes month end and drafts VAT returns and annual accounts, across every client at once.
Exceptions surface
You do not review everything. Clean books come through drafted and ready; blocked ones arrive with the blocker named.
Your team signs off
Review, adjust, approve, file. Finn never files anything. Judgment, advice and the client relationship stay yours.
What actually changes
The same practice, without the production ceiling
| Where | Today | AI-native |
|---|---|---|
| Bookkeeping capacity | Growing the client book means hiring another bookkeeper. | The production layer is software. Capacity stops scaling with headcount. |
| Month end | Chase statements, key entries, tick and bash, client by client. | Prepared in the background across every client. You clear the exceptions and lock. |
| VAT quarter | A per-client grind against the deadline. | Batch-prepared across the whole book. Clean books get a draft; blocked ones get a named blocker. |
| Taking on a client | A setup project: software, logins, chart of accounts, opening balances. | Connect Xero or drop the export. Review the import plan. Live in days. |
| The partner's day | Managing the production line. | Reviewing, advising and signing off. The work clients actually pay for. |
The economics
Margin is a capacity question
The biggest line on most practice P&Ls is production labour. Which means the growth question has always been the same one: can we take on more clients without another hire? When the production layer is AI-native, that trade breaks. Take on more clients with the team you have, or run the same book leaner. Either way, the difference lands on your margin, not on a job ad.
The software side is deliberately simple: 12 pounds a month plus VAT per live client, banded down as the book grows, 1.50 pounds plus VAT for dormant ledger files, and no practice minimum. A pilot costs pounds. The decision is about your workflow, not a platform contract.
Run your book through the pricing calculatorFair questions
What practices ask us
Is this about replacing my team?
Our client data is all in Xero. How real is the move?
Do our clients need accounts or logins?
What does Finn say to our clients?
What does it cost?
Is AccountsOS recognised for MTD?
Build the practice around the book you want to run.
AccountsOS is built for the whole client book. Bring the practice across in stages, connect Finn to the AI tools you already use, or integrate the accounting layer into your own systems. You can prove the workflow on one client without making one client the ceiling.
Moving from Xero? The practice migration path