Sector primer: AI in accounting

TODO: placeholder article — replace with finished editorial copy before launch.

Accounting is one of the sectors where the gap between “AI hype” and “AI in actual workflows” is widest. The hype says everything is being automated; the reality is more interesting.

What’s already changing

• Bookkeeping classification. Most modern bookkeeping software now suggests categories using ML trained on millions of transactions. Accuracy is high enough that the bookkeeper’s job has shifted from data entry to review.
• First-pass document review. Receipts, invoices, statements — extraction is fast and cheap. The skill that’s gone up in value: noticing the things the model missed.
• Drafting client communications. Routine emails, follow-ups, fee-quote summaries. The accountant’s voice still matters; the blank page doesn’t.

What’s not changing

• Judgment calls on positioning. Whether something is a capital expenditure or a current expense, how to structure a business sale to minimize tax — these are still human conversations about specific situations.
• Trust and relationships. Clients hire a person, not a model. The texture of how an accountant explains a tough year hasn’t moved.
• Compliance signing. Someone is on the line if the return is wrong. That signature is a person.

The actual skill that matters now

It isn’t “knows how to prompt.” It’s “knows how to verify quickly.” If you can spot the line item the model classified wrong in 5 seconds instead of reading the whole report, you save hours per week. That’s the literacy.

Where to start if you’re in the field

Pick one workflow. Run it both ways — the way you do it now, and the way you’d do it with AI assistance. Time both. Note where AI helped and where it didn’t. Repeat next week. That’s it. There isn’t a course that beats this.

We’ll have a podcast episode on this with two practicing accountants soon — subscribe to be notified.