You got to the second screen of an expense app, it asked for your online banking login, and you closed the tab. That reaction is reasonable. It is also, in most apps, the end of the road \u2014 no login, no product.<\/p>\n
PennySlice has two doors, and they both lead to the same place. You can hand it files you exported yourself, or you can authorise a read-only connection through your bank. Neither one is the “real” way. They differ in effort, in coverage, and in how much standing access you hand over.<\/p>\n
Every bank with an online portal lets you download your own transactions. Usually as CSV, sometimes Excel, often OFX, and almost always PDF statements. That file is the entire input PennySlice needs.<\/p>\n
Drop it in and the AI reads the structure itself. There is no template to choose and no column-mapping screen where you tell it which field is the date and which is the amount. Banks format exports inconsistently \u2014 different column orders, debits as negatives or in a separate column, dates in three regional formats \u2014 and the detection step is there so you never have to think about any of that.<\/p>\n
The accepted formats:<\/p>\n
PDF deserves a note. If your bank only offers statements as PDF, that works: PennySlice reads the transaction table out of it. But a CSV or OFX from the same bank will parse more cleanly, so if the export menu offers both, take the structured one.<\/p>\n
What import costs you is a recurring errand. Once a month you log into your bank, download a file, and drop it in. Two minutes, but two minutes you have to remember. What it gives you is that PennySlice holds no standing access to anything. There is no token, no connection, nothing to revoke \u2014 just a file you chose to hand over.<\/p>\n
It also works everywhere. A bank in Berlin, Dublin, Madrid or Melbourne exports CSV the same way a bank in Ohio does. There are 104 export guides in the product covering banks across 14 countries \u2014 heaviest in the US and Canada, then the UK, then eleven more European markets \u2014 because “where is the export button on my bank’s site” is the only genuinely hard part of this path. The Canadian bank export guide<\/a> is a good example of what those look like: click paths for RBC, TD, Scotiabank, BMO, CIBC and National Bank, no interpretation required.<\/p>\n Linking runs through Plaid. The part that matters: you enter your banking login at your bank’s own site, not at PennySlice. What comes back to us is a read-only token for transactions and balances. We never see or store your bank password, and you can revoke the token at any time \u2014 from inside PennySlice, or from your bank’s own connected-apps settings.<\/p>\n Read-only is literal. The token permits reading transactions and balances. It cannot move money.<\/p>\n What you get in exchange for that access is the disappearance of the monthly errand. Transactions arrive on their own, which means the background detectors have something current to work with. Penny Spotter runs seven of them without being asked \u2014 budget pace, category anomalies, recurring cost creep, predicted expenses, savings opportunities, income changes and merchant concentration. A subscription that quietly went up in March is more useful to know about in March than in a batch import in June.<\/p>\n The constraint is geographic. Bank linking<\/strong> is available in the US and Canada only at launch. This is worth separating from the export guides above: guides exist for 14 countries, but linking does not. If you bank in the UK or the EU, import is the path \u2014 and it is the whole product for you, not a lesser version of it.<\/p>\n Linking is also on paid tiers only, with a cap on connected accounts: three on Insight, ten on Predict. Paid plans come with a 30-day trial and no card required, so you can see whether an automatic feed changes how much you actually use the thing before you decide.<\/p>\n Most people with messy finances end up here. Link the two accounts that generate the most volume \u2014 the daily-spend checking account, the main credit card \u2014 and import the rest when you need them. A once-a-year business account, a joint account at a bank Plaid does not reach, a foreign card from a trip.<\/p>\n The distinction stops mattering after ingestion. Both paths land in the same data, and the conversational interface does not care where a transaction came from. You ask in plain English across transactions, accounts, categories, budgets, imports, receipts, predictions and what-if scenarios, and you get figures out of your own history rather than general guidance.<\/p>\nWhat a linked bank actually gives you<\/h2>\n
Mixing both<\/h2>\n