May 22, 2026
Audit me, maybe?
Show HN: My dad is a forensic accountant. I automated ~62% of his job
Dad’s spreadsheet grind gets an AI sidekick — and the comments instantly turned feral
TLDR: A founder says he built software that can handle roughly 62% of his forensic-accountant dad’s workload, especially the painful task of typing up bank records. Commenters split fast: some called it the future, others joked about nightmare exploits and warned that using private financial data with AI could trigger serious legal trouble.
A developer hopped onto Hacker News with a very online flex: he studied his forensic-accountant dad’s work and built software that could automate about 62% of it. In plain English, that means helping sort mountains of bank records, spotting missing documents faster, and even flagging suspicious money moves in messy divorce and fraud cases. It’s a classic "boring job meets futuristic shortcut" story — except the real fireworks were in the replies.
One camp was basically cheering from the bleachers. A commenter said accountants who gave up on artificial intelligence too early are already behind, arguing the tools have improved wildly in just months. That gave the whole thread a "grandpa, the robots are here" energy. But the applause got drowned out by instant paranoia, jokes, and legal panic. One person dropped the delightfully cursed line, "Next week we're going to have prompt injections via ledger," turning bookkeeping into a sci-fi horror movie. Another asked the practical question everyone else was thinking: is this just for forensic accountants, or are regular number-crunchers next on the chopping block?
Then came the biggest drama bomb: privacy. One commenter bluntly warned that feeding highly sensitive financial records into large language models — the kind of chatty artificial intelligence behind tools like ChatGPT — without airtight privacy guarantees could be illegal. Suddenly the thread wasn’t just about saving time; it was about whether this shiny shortcut is genius, reckless, or both. In other words: the software found hidden assets, but the comments found chaos.
Key Points
- •The author analyzed 15 forensic accounting cases comprising 650 billing entries, 1,084 hours, and $345,000 billed to estimate automation potential.
- •The article claims 62% of the sampled work is automatable, with document intake representing 26% of hours and being 90% automatable.
- •A major workflow problem described is delayed detection of missing financial documents after paralegals collect discovery materials.
- •The document parsing system had to handle mixed PDFs, page-level classification, document boundary detection, varied bank layouts, and scanned or angled files.
- •The author says an AI investigation agent can search transactions, run gap analysis, identify entities, trace findings to source pages, and draft court-ready reports, while the human accountant retains final responsibility.