Data handling

What happens to a file after you upload it.

Because this platform handles sensitive business information, here is the full path a file takes.

1. Upload and validation

The file is read and validated: dates, amounts, row counts, duplicate rows, empty files, malformed structure and inconsistent currencies where detectable. Errors are shown to you rather than silently ignored.

2. Normalisation

Rows are converted into a standard internal format with a cleaned description and a detected merchant. The original description is always preserved and never overwritten.

3. Analysis

Analysis runs server-side. Monetary calculations, recurrence detection, annualisation, percentage changes and duplicate detection are performed in code, not generated by a language model.

Where language models are used — merchant interpretation, ambiguous categorisation, written explanations and recommendation wording — they receive structured summaries rather than unnecessary repeated copies of your raw dataset, and they are not permitted to invent monetary values.

4. Storage and access

Transactions and results are stored against your account, with access rules enforced at the database level. No other customer can read them.

Platform administrators do not get casual access to raw transaction rows.

5. Retention and deletion

Data is retained while your account is open so period-on-period comparisons remain possible.

You can delete a single upload, delete an analysis run, or delete your whole account and its data at any time from Account settings.

6. What we do not do

We do not train models on your financial data. We do not send transaction descriptions to analytics. We do not move money. We do not cancel subscriptions or contact your suppliers on your behalf.