[wrs]

This and other data analysis front ends could be a fantastic application for LLMs + tool use.

It’s also a market where getting the wrong answer could result in huge liability, so at this point you’re really rolling the dice that you’re a great LLM whisperer. (There’s no such thing as an LLM engineer, at least not yet.)

[presentation]

Yeah, I’m biased since my startup is a very non-AI payroll app, but trusting my finances to an LLM sounds frightening and the money saved is not much since just hiring an accountant whose neck is on the line to get it right just isn’t that expensive.

[JohnnyRebel]

Fair point—though to be clear, the LLM isn’t doing the math, just the interface. The numbers come from structured data, so accuracy isn’t left to chance. Where this really helps is for small business owners who are overwhelmed by QuickBooks data entry and classification. Our goal is to continually improve the experience, making bookkeeping as simple as possible.

[JohnnyRebel]

I totally agree; the liability is real, which is why we don’t let the LLM “invent” numbers. We use the model as the interface, but all financial data comes from a structured database. In practice, it works like RAG: the LLM interprets the user’s question, retrieves the right data, and explains the result in plain English. That way the math is deterministic, the answers are grounded, and the AI layer just makes it accessible.

[wrs]

I can see that this is potentially a good sweet spot for the current state of AI. More complex and custom enterprise BI queries can get totally bollixed up in interpretation — even humans can’t agree on definitions so there’s no way to know if the query is “correct”. Perhaps in small business accounting SaaS you have the luxury of saying “this is the model, no substitutions please” and produce clearly interpretable answers.

[FredPret]

LLM engineer -> silicon psychologist who can sometimes sell the beast into making the year-end postings pass all tests?

[JohnnyRebel]

We sidestep the “silicon psychologist” issue: the LLM simply interprets questions, while all numbers come from structured data. AI explains results, but it can’t rewrite the books.

[FredPret]

Sounds like a powerful model if you can get it right