Your bank isn't in the list — now what?
FinMan imports statements using self-learning templates: it fingerprints a file's layout, matches it to a known template, and parses your transactions. But not every bank is covered yet. When yours isn't, you don't have to type anything by hand.
The AI fallback handles unknown banks
If no template matches, FinMan's AI fallback reads the file directly and builds a template from it — figuring out which columns are dates, amounts, descriptions and so on. This works for CSV, PDF and XLS exports, so it doesn't matter which format your bank offers.
Just send the file to the Telegram or WhatsApp bot, or upload it in the web app. The AI does the parsing and your transactions land in your history.
It only happens once per bank
The template the AI generates gets saved. The next statement from the same bank matches that fingerprint and imports instantly — no AI needed, no quota used. So the slow, AI-assisted import is a one-time setup cost per bank.
Tip: AI parsing of unknown banks is a Premium feature and counts toward your AI quota — check usage anytime with /quota. Once the template exists, future imports are free.
Get a clean import the first time
- Export the full statement, not a screenshot — the AI needs the raw file structure to map columns reliably.
- Pick the longest date range you can; one big file beats many tiny ones for building a solid template.
- Prefer CSV or XLS when your bank offers it — tabular data is the cleanest to parse, though PDF works too.
- Review the first import to confirm amounts, dates and signs look right before you trust the template going forward.
After import: let AI sort it
Once transactions are in, AI auto-categorization assigns categories and subcategories so you're not tagging hundreds of rows manually. You can adjust anything that lands in the wrong bucket.
If your bank happens to be Monobank, skip statements entirely — connect the integration and webhooks capture each card transaction (amount and merchant) automatically as it happens.