Three years of history from the tool you're leaving. A backlog you let pile up. A business that runs enough transactions a day that sorting them one at a time was never realistic. Fin is built to take on volume, not just tidy it up a row at a time.
Nobody sets out to have eight thousand uncategorised transactions. It happens because you migrated off another tool with years attached, because the books got away from you for a quarter, or because the business genuinely does that many transactions and manual sorting was never on the table. None of this is size-gated — a sole trader with one account and a business running several are handled the same way, at whatever scale each actually is.
Bring your history across from wherever it lives now — CSV or PDF statements from virtually any bank worldwide, mapped automatically, not against a template you have to match.
A year, ignored, is not a special case. It is imported and categorised the same way one week is — you correct the exceptions, not the whole thing.
Retail, hospitality, anything with a high transaction count: rules and bulk actions do the sorting a spreadsheet never could keep up with.
One more thing volume brings with it: the same import is rarely all one thing. A single bulk drop can mix business and personal, span several accounts, or cut across half a dozen projects or clients at once. Categorising it is only step one — Views (also called Lenses) are the second, independent way to slice that same set of transactions, by project, client, department or anything you define, without touching the tax categories the bulk import already assigned.
Not a promise — how it actually behaves when the volume is big.
Select a large batch of transactions and re-categorise every one of them in a single action. Built for clearing out real volume in one go, not for nudging a handful of rows at a time — and it runs on the categories and logic you actually set up, not a generic default, because every pass is driven by your own custom rules.
See how rules shape it to your business (opens in a new tab)Past a size threshold, an import is broken into manageable batches behind the scenes instead of processed all at once — so a multi-year history does not time out, and does not slow down anything else while it works through.
Every transaction — bulk-imported or arriving live — is categorised in under 200 milliseconds, starting at 70-80% accuracy and improving toward 95%+ as Fin learns your patterns from what you correct.
CSV or PDF statements from virtually any bank worldwide — the columns are mapped for you. If your bank connects directly, this step disappears entirely and transactions arrive on a live, read-only feed.
A year of transactions you haven't looked at is not just an import job — it is a year of deductions nobody has found yet, and a year of receipts scattered across emails, folders and a shoebox. Bulk processing only solves half of that problem if it just sorts the rows. The other half is what happens to them next.
Deduction surfacing runs on every transaction that lands, imported or live — a purchase from fourteen months ago is scanned exactly the same way as one from this morning. Catching up does not mean catching up late; it means finding what was missed the whole time. See what Fin looks for, by occupation.
Forward a year of receipt emails to your own AI2Fin address, or snap the ones sitting in a folder, and each one files itself against the transaction it matches — extracted, categorised, encrypted at rest. What you had was a pile of documents; what you get is a line-by-line record.
Once the backlog is sorted, hand it over in the shape it needs to travel: the Accountant Ledger for accounting software, a Profit and Loss Summary for the overview, or your tax-agency's own deduction format — for whatever date range the catch-up actually covers. See the three formats side by side.
Three things, per transaction — not per year. Fin builds toward all three as the backlog is imported and sorted; this is the checklist behind that.
None of this is advice on what you can claim — Fin surfaces likely deductions from your actual spending; what applies to you is a question for your accountant or tax authority.
Volume you handle once is a chore. Volume you handle with a rule is solved. The two work together: bulk-assign what has already landed, then write the rule that catches the next one automatically. There are two reasonable orders to do that in.
Drop it all in
Bring the whole backlog in at once and let automatic categorisation take the first pass on everything. Then use Smart Filters and bulk-assign to clear up whatever it got wrong, and write rules for anything that keeps recurring. Fastest to get started; best when the spending is mostly straightforward.
Calibrate first
Import a month or two first, write the rules that cover your recurring merchants and edge cases against that smaller set, then bring in the rest of the backlog once those rules exist. More setup up front; best when the business has a lot of repeat spending worth getting exactly right before the volume hits.
Step one
Connect your bank for a live feed, or import CSV/PDF statements — one file or several, covering whatever date range you have. Large files process in the background; you are not stuck watching a progress bar.
Step two
Every transaction is categorised on arrival. Use Smart Filters to pull up anything that needs a second look — by date, amount, category, merchant or tag — and correct those instead of reviewing the whole set.
Step three
Select the transactions that share a category, however many there are, and assign it in one action — the fastest way to clear a backlog that a rule would not have caught cleanly.
Step four
If a whole category kept needing manual correction, that is a rule waiting to be written — a condition and an action, matched against description, merchant or amount. The next time that volume shows up, it sorts itself.
CAFE and the amount is greater than 50 then set the category to Meals & Entertainment and set the note to "client meeting — confirm receipt".That note is what turns a bulk pass into an audit trail — it stays attached to the transaction inside Fin, so the reason a borderline expense counts is not something you have to reconstruct from memory eight months later.
Bulk processing, answered
Connect your bank or import your history, and let volume stop being the reason the books are behind.