Why we built Cassis

For about four years I ran my money out of a spreadsheet. Not a good one. A tab per account, a manual paste of each statement at the end of the month, and a summary row at the bottom that was accurate for roughly two days before it drifted again.

It worked, in the sense that it existed. It didn't work in the sense that I only ever looked backwards. By the time the spreadsheet told me something, the money was already gone.

The thing that finally broke it was a question I couldn't answer. I was deciding whether to book a flight in March, and I wanted to know one thing: will I have enough on the 15th? I had four accounts. Each of them could tell me what I had today. None of them could tell me what I'd have in six weeks, because none of them knew about the other three, and none of them knew about the annual insurance premium that was about to land.

That's the whole product, really. That question.

What existing tools got wrong

We spent a couple of months using everything we could find. There were three recurring failures.

The first was that most tools are backward-looking by design. They categorize what already happened, produce a pie chart, and stop. That's an accounting view of your money, and accounting is a record, not a decision tool. Nobody has ever avoided an overdraft by looking at last month's pie chart.

The second was effort. Almost every budgeting app assumes you'll show up. You'll categorize the uncategorized. You'll set the envelope amounts. You'll review the weekly digest. The apps that require the most discipline are also the ones marketed to people who feel out of control — which is a strange thing to ask of someone.

The third was trust. Half the tools we tried wanted write access to move money on our behalf. The other half were free, which meant we were the product, and their privacy policies did nothing to argue otherwise.

What we decided

Three rules, written down in the first week, and we've kept all three.

Cassis reads, it never moves. Every connection we make is read-only. Savings rules tell you what to transfer and when; you confirm the transfer in your own bank. This closes off a whole category of features, and we're fine with that.

Cassis looks forward. Categorization is table stakes. The point of knowing where your money went is being able to say where it's going. Every feature we've built since has had to answer to that.

Cassis is paid. We charge money for the product, so we don't have to make money from the data. It's a less exciting business model than the alternative and it's the only one we could explain to a friend without wincing.

Where we are

We're a team of six. We've been live for fourteen months. People connect an average of 3.2 accounts, and the number that surprises us most is that the median user opens Cassis four times a month — not four times a day.

That was always the goal. This isn't a product you're supposed to live inside. It's a product that should tell you the one thing you needed to know, and then let you get on with your life.

The flight, by the way, was fine. I'd have known that six weeks earlier if the thing I've just described had existed.