I integrated an LLM into the bookkeeping app I built.

I recently asked some friends to try out the bookkeeping software I developed, and the feedback was blunt: for the average person—or even a lazy person like me—the learning curve for this system is just too steep.

It’s true that for bookkeeping data to be meaningful, precise "tagging and categorization" is an essential step, but it’s also often the most tedious part. To simplify this process, I decided to integrate an LLM (Large Language Model) into the system.

Now, users only need to type a single sentence, and the AI automatically handles the categorization. All you have to do is confirm that the tags and payment methods identified by the system are correct, hit record, and the entire process is complete.

In addition, I’ve added an end-of-month analysis feature. The system automatically determines whether your spending is reasonable based on local price trends and generates two detailed financial reports.

Interestingly, after analyzing my spending habits, my AI actually suggested that I "quit smoking." Looking at the report, I didn't know whether to laugh or cry; it seems this system doesn't just manage my money—it’s started managing my health, too.