December 16, 2025
When bots meet the tax man
Tell HN: AI coding is sexy, but accounting is the real low-hanging target
Commenters split: Already automated vs “bookkeeping isn’t accounting”
TLDR: A poster says AI should automate small-business bookkeeping because it’s repetitive and rules-based, but commenters clap back that real accounting is judgment and tax strategy. The thread splits between “already in QuickBooks/SAP” and “please automate the crunch,” with LLMs mocked for being too error-prone for money work.
A bold poster claimed the real AI gold rush isn’t writing code—it’s automating small-business finances. Their pitch: bookkeeping is rule-based, verifiable, and boring—aka machine catnip. Think: normalize bank data, apply rules, flag exceptions. The crowd? Instantly divided and very loud.
Skeptics piled in first. One veteran quipped that computers have been checking balanced books “for decades,” and that accountants get paid for the messy stuff—tax strategy, audits, and talking to the government when things go sideways. Another waved the manual, pointing out that most of this already lives in QuickBooks, and if you want big-boy rules and integrations, welcome to SAP.
Then an actual accountant showed up with the hammer: “You’re describing bookkeeping, not accounting.” Applying GAAP (Generally Accepted Accounting Principles) involves judgment—when to record revenue, what to accrue, how to classify gray-area expenses. Translation: spreadsheets don’t argue with auditors, people do. A startup founder countered with real-world pain: finance teams endure brutal crunch cycles, so there’s a huge incentive to automate the grind.
The spiciest twist? LLMs (chatbots like ChatGPT) were roasted as a terrible fit for ledgers—“no hallucinations allowed.” Memes flew: “Wake me when AI survives an audit,” and “Press X to expense tacos twice.” Verdict: automating the grind? Sure. Replacing the CPA stamp? Not today.
Key Points
- •The article focuses on automating small-business bookkeeping, reconciliation, and basic reporting.
- •It argues accounting is often more automatable than programming because it relies on established rules.
- •Verifiability via bank feeds, statements, and prior periods provides ground truth for checks.
- •Routine, repetitive transaction patterns make deterministic, rule-based automation effective.
- •Complex tasks like tax strategy, edge cases, and regulatory interactions still require humans.