As of 13 August 2026, AI can categorise your bank transactions.
This still needs a person who signs their name to it.
Can you do it?
5 minutesto a draft.
30 minutesto something you’d act on.
Cost, all in£0
Skill neededchat-fluent
Who has to check ityou
What the alternative costsThe supplied data gives no price for a human alternative to transaction categorisation.
If this goes wrong: some payments land in the wrong category and distort your budget until you correct the rows.
What to actually do
Hand it to a person
The route this page recommends
A person who owns the outcome does this end to end, worth it when the failure is dear.
Do it yourself
Second choiceA chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.
How to actually do it
- Open your banking app or online banking and export the relevant transactions as a CSV or spreadsheet, including the date, description and amount.
- Remove full account numbers, card numbers, addresses and any transaction details that are not needed for categorisation, while keeping enough description to identify each payment.
- Create a category list that matches how you want to track spending, then upload the cleaned file and paste the prompt above into the AI tool.
- Check the returned row count, dates, descriptions and amounts against the original bank export so that no transaction has been omitted, duplicated or changed.
- Open the exceptions list and decide each uncertain transaction from your own records, such as a receipt, invoice or merchant account, then change the category in the spreadsheet.
- Add the corrected categories to your budgeting spreadsheet or banking tool and keep the original export alongside it so you can trace every decision.
Prompt
Categorise the transactions in the attached CSV or spreadsheet for a personal UK spending review. Use these categories unless a transaction clearly belongs elsewhere: housing, household bills, groceries, eating out, transport, shopping, health, entertainment, subscriptions, debt payments, savings and investments, income, transfers, cash withdrawals, fees, refunds, and unknown. Preserve the date, description, amount and account reference exactly as supplied, but do not reproduce full account numbers or other unnecessary private data. Add a category, a short reason, and a confidence level to each row. Do not guess what an ambiguous merchant sells. Put uncertain transactions in an exceptions list and explain what information would resolve each one. Do not give tax advice, decide whether a payment is tax-deductible, or treat transfers between my own accounts as spending. State any assumptions before the table. Return the categorised table in CSV-friendly format, followed by the exceptions list and a summary of totals by category. This is not professional advice.
Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.
Use a tool built for this
The distant third
What it gets wrong
- AI cannot know what an unfamiliar merchant bought unless the transaction description or your records make it clear.
- AI cannot decide your personal definition of categories such as household, shopping or eating out without rules from you.
- AI cannot reliably determine whether a payment is tax-deductible or how HMRC would treat it.
- AI cannot access missing bank transactions or detect a payment that your bank failed to include in the file.
- Sensitive transaction data may be exposed to the service you upload it to, so you have to decide whether the convenience is worth that privacy risk.
Even on a YES, the friction has a name: private data access, judgement under ambiguity and verification cost.
How we scored this
Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.
| Axis | Score (0–2) |
|---|---|
| Output | 2 |
| Inputs | 2 |
| Verification | 1 |
| Liability | 2 |
| Effort delta | 2 |
| Total | 9 / 10 |
The methodology and its thresholds are published in full.
FAQ
- Can AI automatically categorise my bank transactions?
- Yes. Give it a cleaned CSV or spreadsheet and clear category rules, and it can classify the rows and flag uncertain payments. Check the returned amounts, descriptions and exceptions against your bank export before using the result.
- Is it safe to upload my bank statements to AI?
- It can expose sensitive financial information to the AI service, so remove account numbers, card numbers and unnecessary personal details first. This is not professional advice; ask your bank about security concerns or an accountant about financial records that matter for tax.
- Can AI categorise transactions for my tax return?
- It can sort transactions into a review list, but it cannot reliably decide which costs HMRC will accept or which evidence you need. This is not professional advice; ask an accountant before using categorised transactions in a tax return.
- Can AI find spending mistakes in my bank transactions?
- It can flag duplicates, unusual descriptions, refunds and transactions that do not fit your category rules. You still need to compare those flags with your bank records and contact your bank about a payment you do not recognise.
Nearby answers
Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.
The newsletter
AI news, new answers and product picks, straight to your inbox.