Before any app gets AI access to your transactions, ask what data is sent to the AI, whether it is stored, whether it is used to train models, what the AI is allowed to do with your records, and how you switch it off. If an app cannot answer those five plainly, treat that as your answer.
What data is actually sent to the AI?
This is the most important question. Some apps send only a cleaned merchant name and an amount range so that an AI can suggest a category. Others send full transaction text, your balances or the whole uploaded document. Ask the app to say which. A good answer is specific, such as listing the fields, rather than a general promise about privacy.
Where does the data go, and how long is it kept?
- Which company processes the AI request, and is that company named?
- Is the request stored, and for how long?
- Can you delete your data, and does that include what the AI provider saw?
- Can you choose how long documents are kept?
- Does the answer change between the free and paid versions?
Is my data used to train models?
Ask directly, and ask about both the app and its AI provider. The answer should be written in the app's privacy policy or help pages, not given only in a chat with a support person. Provider settings can change, so check the current wording, and note the date you checked.
What is the AI allowed to do?
There is a world of difference between an assistant that can read totals and one that can add, edit or delete entries. Ask what actions are possible, whether you can limit them and whether there is a record of changes. A sensible app lets you start with read-only access and widen only if you want to.
How do I turn it off, and what happens then?
Look for a clear way to switch AI off, ideally for each feature separately, and ask what the app does without it. A good fallback is that items the app cannot categorise are held for your review rather than guessed. Also ask whether you can set a spending limit on any paid AI use.
| Question | A reassuring answer | A warning sign |
|---|---|---|
| What is sent? | A named list of fields, such as merchant name and amount band | Everything, or no clear answer |
| Is it stored? | Stated retention period | No retention information |
| Is it used for training? | Clear yes or no in the policy | Vague wording |
| What can it do? | Scoped access, read-only option | All or nothing |
| How do I stop it? | Switch off per feature, disconnect any time | Must contact support |
How does SpendixAI answer these?
According to its published facts, SpendixAI categorises in stages: your own rules, then your history and similar merchants, then an AI model only if AI is on, for what is left. That model sees the normalised merchant name and an amount band, not your name, account numbers, full statement text or balances. AI can be switched off per feature, given a monthly budget or turned off entirely, and optional features such as reading an unfamiliar statement layout do process the uploaded document. It says it does not sell personal information, has no bank connection and does not ask for net-banking logins. Claude or ChatGPT can be connected via MCP with a scope you choose, starting read-only, and disconnected at any time. Compare these answers with any other app on the same questions.
A list to keep before you agree
- I know what data leaves the app.
- I know whether it is stored and for how long.
- I know whether it trains models.
- I know what actions the AI may take.
- I know how to switch it off and disconnect it.
- I have read the current privacy page, not just the sign-up screen.
Useful ideas start with a clearer picture of your own spending.
Explore your reports

