Comment by combocosmo
2 days ago
Nice project! I built a CLI budgeting project a long time ago, and what made me stop using my own project was the lack of automated integration with my bank accounts. At that point I had many credit cards, multiple bank accounts, in different currencies, and integrating all expenses was just too much manual work.
I wish financial institutions were better at automated exports of your financial data, given the right permissions of course.
That’s a fair point. Automated bank imports sound essential at first, especially with many accounts and cards.
In practice, though, I found them less useful for budgeting than expected. A bank statement tells you how much was spent and where, but not what the expense actually was. “$100 at a supermarket” could be groceries, pet food, a lawn mower, or business expenses — that context is what makes budgeting meaningful, and it usually has to be added manually anyway.
At that point, entering the expense directly with the right category often turned out to be simpler and more accurate for me. Automated access would still be nice for reconciliation, but it’s not the silver bullet it’s often perceived to be.
This is something I kept bumping into when building my own tracker (Simple Wallet - https://simplewallet.app).
You're right that "$100 at a supermarket" is useless but I found even knowing "I spent $400 on groceries" wasn't that useful either. I kept asking myself "okay, but on what?"
So I leaned hard into making categories the starting point instead of the endpoint. Groceries breaks down into what I'm actually buying. Turns out I was spending way more on coffee than I realized.
Did you ever consider going deeper into categories, or do you find users just want the high level view? I've been torn on how much detail is actually helpful vs. overwhelming.
This was true, but today I would much rather have an llm categorize my expenses. Me doing it manually will never happen.
That’s fair — and I agree if enough context exists.
The key limitation is that a raw bank transaction usually contains very little semantic information: amount, merchant name, date. From that alone, an LLM can only guess based on patterns or prior behavior, not actually know what the expense was for.
“$100 at a supermarket” could be groceries, pet food, a household item, or something work-related. An LLM can infer probabilities once it has enough historical data and feedback, but that still means the user has to correct or confirm things at some point.
So I see LLMs as very helpful for assisting categorization (suggestions, defaults, learning over time), but they can’t fully replace intent unless the underlying data becomes richer than what bank statements provide today.
3 replies →
I have been pondering on this issue as well. Budgeting apps which have integration seem to have access to too much data.
Why do we need such a detailed breakdown for personal finance. I am pondering on this idea of using under or over type prompts to capture daily expense. The app send a notification asking me did you spend $100 or less in the day. Eventual goal of all budgeting apps is to reduce spending. this simple prompt can capture immense information without needing to break it down. breaking it down to categories can be a step if there is a problem detected in the savings pattern. For that banks are already adding the feature
keen to hear thoughts on this
it's very sad that in Europe we have laws to guarantee "open banking" but in practice it's only B2B
one way to go around this is to use apps like Toshl which connect to banks (it is far from perfect but usable) and then if you are unhappy with the app you can use their API to sync with your own system