Built AI forensic accounting software with my dad

Edit: OK, I'll remove the snark fairly called out by dvt:

There is something very concerning about this article: submitting private information to LLMs w/no privacy guarantees is probably a crime. I strongly recommend taking down this article and that you stop submitting private information to LLMs with no privacy guarantees until you have spoken with an informed lawyer on this matter.

Local models may be of assistance here, but you need to be very careful.

Great stuff. My favorite genre of writing about AI is seeing how it can be practically applied to non-tech jobs/businesses. Wish we had more of this.

I'm curious about the 60% automation of financial/forensic analysis - what's missing? Is it stuff that's purely blocked by model capabilities, or are there places where scaffolding is likely to bridge the gaps?

Also curious about the workflow - is this more individual, LLM-driven features or agentic workflows? Looked like the former from the product video but there wasn't a ton of UX shown there.

I ask largely because this seems like the sort of thing where you could really start to string these features together in such a way that you start with a description of the case and whatever files you have, and then an agent does its analysis of the docs, spins up action items (get missing docs, confirm that X ambiguous doc is what the AI characterized it as, etc.) and tracks the progress of all of them, leaving your forensic accountant there in a supervisory role, managing and providing expertise.

It feels like that's the way a lot of expert analysis jobs like this are headed. I've been working on the same sort of flow to use agents to manage my business. Started with LLM skills that could be used to handle tasks I used to do myself, and since then I've increasingly been having AI use those skills on its own without me invoking them and chain things together into full blown workflows. Some parts I'm still supervising closely, but others that have been working consistently for a while I now don't really watch unless Claude flags something for me to review on my dashboard.

Where's the breakdown of these stats? What does it mean that 60% `Forensic Analysis` can be automated with AI? Are these per hour? Its also telling that each of the automated percentiles are rounded to the nearest 10%.

Next week we're going to have prompt injections via ledger

Nice. I have a friend who is a young accountant. I have tried to get him to consider AI, but he claims that they tried it and it's not that good. I've tried to get him to understand that AI has improved dramatically in the last few months, not to mention the last few years (their point of reference, I believe).

Is this for any kind of accountant or only forensic?

What is the document recognition stack you used?

Why is it that every "I built a cool AI tool" author shared on this site can't be bothered to write the article themselves? I'd be more likely to give credence to how great your slop is if you were at least invested enough to write the dang article yourself.

Here is my hot take. AI is going to replace some developers (not all) and the first ones it replaces will be the ones who can't code without it. The developer in this story provided a relationship with a forensic accountant, a few discussions with paralegals, and limited guidance to an agent. The agent did literally everything else, including writing the article!

model, stack?

There is nothing to Show HN (1).

1. https://news.ycombinator.com/showhn.html

[flagged]

[dead]

[dead]