Balancer

Balancer is a personal finance tracker I built for my household. It presents our bank and credit card activity in a categorized, easy-to-digest way, makes it clearer where money is coming from and where it's going each month, and suggests ways to improve our finances.

I used and before building Balancer. Both had useful parts, but neither solved the problems that mattered most to me and my partner.

Most importantly, neither app gave us the broader financial picture we wanted. One place that could tell us how we were doing financially, what was helping us, what was hurting us, what was changing over time, and what we could do to improve.

Second, their categorization systems made it hard to match the way we actually spend, which created an inaccurate picture of our monthly cash flow. Cheques were one example: we pay both rent and after-school care by cheque, but those transactions were lumped into one category even though they should have been split. I had to keep correcting them every month.

Finally, they mixed savings contributions into expenses, which caused two problems. In one app, savings contributions inflated the expense total, which meant the monthly spending total was unreliable. In the other, savings were buried among regular spending categories, making them harder to track.

I'm also building Balancer to practice designing with AI agents. And while it started as a tool for my household, I'd like other households to benefit from it too, so I keep expanding it to make it easier to use and flexible enough to handle different needs (e.g., Hebrew interface, custom categories, etc.).

To track finances and get recommendations in Balancer, users upload bank and credit card statements (to be replaced by automatic account sync soon). Balancer then auto-categorizes the transactions. If a transaction is miscategorized, users can apply a one-time fix so similar transactions are categorized correctly.

Once the transactions are in good shape, users move into review and decision-making. They can understand the current monthly cash flow (income, expenses, and savings), whether the month is likely to end positive or negative based on current pace and past activity, how much they can likely still spend on unplanned expenses, and which actions would help most. Balancer provides up to three recommendations for how to continue the month, such as spending less in specific categories or saving more to stay on track with goals, including building an emergency cash fund.

From there, users can explore trends across past months to see how their income, spending, and savings are changing over time, how much they usually have left after expenses and savings, how much they usually save, how much they usually spend in a given category, and whether there is any suspicious spending. If they need more details, they can drill down to the specific transactions associated with the period they are exploring.

Balancer is fully responsive, so users can do all of the above from any device.

Guidance and certainty rather than just "what happened"

Rather than only reporting what has happened so far in the month, Balancer tells users what is likely to happen and how they should respond. That reduces uncertainty and helps them make decisions that improve their odds of finishing the month in a healthier financial position, or at least moving in that direction over the coming months. Historical analysis still matters, but it lives in separate views so the main experience stays focused and easier to act on.

Flexible rules instead of repetitive cleanups

If Balancer miscategorizes a transaction, users can correct it once and turn that into a reusable rule for similar transactions. Those rules do not rely only on the transaction description. They can also use the amount and the account the transaction came from. That matters because the same transaction description can mean different things in different contexts. A transaction with the same description might belong in one category on a bank account and another on a credit card, or it might need to be split based on the amount. Matching on more than one signal makes categorization more accurate and cuts down on manual cleanup each month.

Savings separated from expenses

If savings contributions are counted as expenses, the monthly expense number becomes misleading. Balancer separates savings from expenses so users can clearly see both how much they spent and how much they intentionally set aside.

As I was building Balancer, I tested and refined it by asking my wife to use it for real tasks, such as uploading statements, reviewing categorization, and deciding what to do with the month's numbers. For example, at one point the current-month view and historical trends lived on the same page. At first, I thought keeping them together would make comparison easier, but in practice it created data fatigue. The more information I added, the harder it became to see what actually mattered. Splitting them let the main view stay focused on monthly cash flow, recommended actions, and savings guidance, while Trends became the place for historical exploration.

Balancer has already helped us save money. Part of that came from seeing how much we spend on specific categories (e.g., eating out) more clearly over time and cutting back where we were spending more than we thought was reasonable. Part of it came from the emergency cash guidance, which showed that too much money was sitting in the bank instead of being moved to places where it could grow.

It also improved my personal finance tracking experience in other ways. Manual categorization dropped because the rule system became increasingly tailored to my household. My monthly spending number became more trustworthy because savings were no longer mixed into it. And it became faster to find the things that mattered instead of hunting for them across multiple screens.