A common expectation among people new to AI expense categorisation is that the tool simply connects to a bank account and starts working. Some setup does happen automatically, but the default configuration often does not reflect how a specific business actually tracks its spending.
What defaults look like in practice
Most tools arrive with a standard chart of accounts - broad categories like Travel, Meals, Office Supplies, and Utilities. For a business that needs to track expenses by project, client, or department, these defaults are too broad to be useful. A marketing agency billing expenses back to individual clients cannot do that with a generic Advertising category.
The configuration that actually matters
Useful setup involves mapping custom categories, setting rules for recurring merchants, and connecting the tool to the accounting software already in use. In Dext, for example, a user can create a rule that automatically assigns any transaction from a specific supplier to a named project code. That rule does not exist until someone creates it.
This is not a criticism of the tools - it is a description of how they work. The AI learns from patterns, but the initial pattern framework comes from the user.
A realistic starting point
Spending one to two hours on initial configuration - defining categories, setting merchant rules, reviewing defaults - produces noticeably better results than simply connecting a bank feed and walking away. The AI improves faster when it starts with a coherent structure to work within.