You can automate most of expense tracking without handing over any net-banking login: download your statement yourself, upload the file, review the rows, and let rules and history categorise the repeat merchants. The trade-off is that this is periodic rather than live, because the tracker sees your spending when you upload a statement, not the moment you pay.
How does file-based automation work?
The idea is that you stay in charge of the one step that needs your bank credentials, which is downloading a statement from your own bank app or site. After that, only a file travels.
- You download a statement for a date range from your bank or card issuer.
- You upload that file (PDF, CSV or Excel) to the tracker and choose which account it belongs to.
- You review the rows, then save them.
- Categories are applied from your rules, then your history, then an AI model only if you have it on.
- Reports update from the saved rows.
SpendixAI works this way. It does not connect to a bank and never asks for net-banking logins, and it does not read your SMS. Sign-in is through Google only, so there is no separate password for the tracker either.
What is different about connecting to a bank instead?
A bank-linking tracker pulls transactions in on its own. That can feel effortless, and it is more current. But it asks you to authorise access to your account, and you are relying on that link staying healthy. A file-based approach asks for a few minutes of your time each month and little trust, because the only thing handed over is a document you chose to send.
| Question | File upload | Live bank link |
|---|---|---|
| What do you hand over? | A statement you downloaded | Permission to read the account |
| How fresh is the data? | As of the statement date | Close to real time |
| What do you do each month? | Download, upload, review | Mostly review |
| What if the link breaks? | Nothing to break | Reconnect, or wait |
How automatic can the categorising be without a login?
More than people expect. Rules turn a merchant you have named into a category every time it appears. History handles similar merchants and your earlier corrections. If you switch AI categorisation on, a model looks at what is left, and it is shown the normalised merchant name and an amount band, not your name, account numbers, the full statement text or balances. You can switch AI off per feature, set a monthly AI budget, or turn it off entirely. If AI is off, unmatched items are held for review so nothing is guessed.
What about the AI features that read a document?
Be clear about this one. Some optional features, such as reading an unfamiliar statement layout, do process the uploaded document itself. That is separate from categorisation, where only the cleaned merchant name and an amount band are used. If you prefer not to send documents to an AI feature, leave those options off and use a CSV or Excel file, which is usually easier to read anyway.
Is monthly good enough?
Ask yourself what you do with the information. If you are looking for where the month's money went, to set a budget, or to hand something to an accountant, a weekly or monthly upload is plenty. If you want a nudge the second you overspend, a file-based tracker will feel slow, and cash you add by hand will be fresher than the card rows.
A privacy-minded routine
- Download statements only from your bank's official app or site.
- Delete the downloaded file from your phone or laptop once the import is done, if you no longer need it.
- Keep AI features off until you have read what they use.
- Use your own workspace's export and deletion controls if you ever want your data out.
- Never type a net-banking login into anything that is not your bank.
Useful ideas start with a clearer picture of your own spending.
Explore your reports


