Guide

What is AI bookkeeping? A plain-English explanation

AI bookkeeping explained: how document AI reads statements, how categorization models assign categories with confidence, what humans still do, and how it is priced.

AI bookkeeping means software reads your financial documents and produces categorized books, with a person confirming only the uncertain parts. Here is how the pieces work.

Reading documents

A document model reads a PDF or photo the way a person would: it finds the account number, the period, the balances and the transaction table, across thousands of bank layouts. Extraction is checked against the balances printed on the statement so errors are caught by arithmetic.

Categorizing

A categorization model looks at each transaction's description, amount and timing, the vendor's history across many businesses, your industry and your own past corrections. It outputs a category and a confidence score. You decide the threshold above which it finalizes without asking.

What humans do

Confirm low-confidence transactions (two clicks each), upload receipts for the items that need documentation, and read the reports. A CPA still files the return and makes the judgment calls.

How it is priced

Old bookkeeping was priced by the hour or the month. AI bookkeeping can be priced per unit of work. ProfitBooks charges $20 a month, billed annually, for the Profit plan ($99 a month, billed annually, for Full Books, which adds a reconciled balance sheet), and one credit per finalized transaction, with 10 free credits every day. If the AI does no work, you pay only the plan.

Your books, done tonight.

Create a free account, drop in a bank statement, and watch every transaction get categorized. 100 free credits to start, 10 more every day. No card required.

Takes about 90 seconds. No sales call, no onboarding fee, no contract.

Start free