GnuCash 5.9

I think the only thing that will ever really make sense is flat horizontal property lists.

A given item and a given transaction could have any number of different properties, and no single heirachical category can express the reality.

Is it a work expense or is paid through paypal or is it a subscription or is it from amazon or is it food? A single item could be all of those at once, and sometimes when you want to know "all food" you want to include that regardless what it's other properties are.

IE, pork doesn't really have to be under some heirarchy like meat. It is both pork and meat, and it also may be art supplies or photographic subject etc.

I’d expect something like claude 3.5 or chatgpt 4 to do OK at this. (Maybe ocr the receipt, or just send a copy to a multimodal model).

You might be able to use one of the open weight models instead. (Maybe one of the apache 2.0 qwen’s?) Scan a batch, then hand check the results in fifo order. That way you can probably get away with a local gpu.

The hard part will be getting structured output that gnucash likes. I’d try a simple json schema (stick the schema and some example input/output in the context), and then write some code to convert it to a format gnu cash can import.

It's great for personal or very small businesses but god help you if you are building a real startup with GNUCash. I am speaking from experience here, GNUCash zealotry is a plague. I honestly wish this project basically did not exist, because the business world despises GNUCash, ONLY cares about QuickBooks, and using anything else is a giant waste of everyone's time. Believe me, I have been fighting this fight in non-profits and startups since the early 2000's. It's not a fight ANYONE should ever have. I USED TO BE THE "WE MUST USE GNUCASH" GUY!

Now, in a perfect world, yes, GNUCash and literally ANYTHING other than Quickbooks would be an option for small business accounting. But we do not live in a perfect world. We live in a world where Intuit has fought VERY hard to make damn sure no one can use anything but Quickbooks, and their iron fist is clad in very specific APIs and file formats.

If you do not use Quickbooks:

* Your bank will hate you.

* Your investors will hate you.

* Your payroll system won't work.

* Your tax systems won't work.

* Your accountant won't work.

* Grants are even off the table in some cases.

* Some places won't audit without Quickbooks.

I constantly run into well intentioned open source zealots who demand the use of GNUCash. This is terrible. Don't be that person.

The world has chosen QuickBooks. This choice was made under duress and with corrupt power brokering. But the decision has been made. Maybe there are some OK SaaS options, but they only exist as long as Intuit allows them to exist. Anything competing with Quickbooks is going to be bought and killed by Intuit, so you're going into a dead-end alley. I know that leaves GNUCash on the table, but...

Please, do not make the mistake my many non-profits and businesses have repeatedly made when I was not paying enough attention to scream bloody murder about it. I turn my back for one minute, and engineers are installing GNUCash on the accountant's Mac Laptop.

It's 100% always come back to bite us in the ass, as we've had to change platforms on-demand in order to meet a funding deadline, a bank requirement, a loan ask, or a grant application. And it's ALWAYS the accountant in the org that gets saddled with the 60+ hour work weeks it takes to redo everything. If I was an acocuntant, I'd quit if told to use GNUCash, but most of them kinda don't know any better because, hey, why not try some thing the techies are all excited about!

GNUCash is a wonderful project. I wish we could all use it. But we cannot. Not for real business. And honestly, this is only for contrived, arbitrary reasons. But these reasons exist. The world is 100% built to prevent people from not using Quickbooks at every turn, and you only harm yourself by demanding open source software for accounting. As I said to the director of my most recent non-profit: "You would not tolerate the accountant coming in here and demanding you use NetBeans. Please, give them the same courtesy you'd expect them to give you on tool choice."

I was going to say what GP said, but, yeah, XSLT is just no fun to write by comparison to SQL.

You don’t need that level of accuracy, you just need “close enough”. Split things out that are important, but if you’re trying to get the sales tax calculated to the cent you’re going way past a level of utility that matters. Eyeball it and move on.