Why (Not) Build Yet Another Finance App — Part 0

Every developer, at some point, opens their expense spreadsheet, sighs, and thinks: "I could build a better app than this in a weekend."

We thought that too. And we built it.

And the part we expected to be hard, showing where your money goes, turned out to be the easiest of all.

What we thought we were building

What we actually built

• Part 1 A system that notices when the bank sent the same purchase twice, with slightly different descriptions.
• Part 2 Logic to tell whether someone is still a subscriber when Apple doesn't tell you they've cancelled.
• Part 3 A way to understand that a "PIX ENVIADO" (an instant transfer, Brazil's everyday way to pay people) to the same person every 5th of the month is the rent, not "Transfers".
• Part 4 A credit card bill that closes on one day, is due on another, carries an installment from a purchase made eight months ago, and has to match what the bank shows.
• Part 5 A couple who want to share the household account, but not each other's credit cards.
• Part 6 Protection for your bank statement, because it says more about someone's life than any diary.

None of this shows up in an App Store screenshot. All of it has broken in production at least once.

$ git log --oneline | grep -i dangerous
a91d4bbb DANGEROUS change in pluggy for PATCH
4d9cbbe8 dangerous cards upload
6530cc53 dangerous commit in prod - ADD PLUGGY
d8b6ebca workspace e initial sync dangerous changes
2a699cef dangerous sync transaction performance fix

Five commits with "dangerous" in the message. We remember the night of the third one well.

Why write this

It's not to talk anyone out of it. Competition is good, and personal finance in Brazil still has plenty of room for good products.

It's a diary. We want to write down what went wrong while we still remember the details, before it turns into "oh, it's always worked like that". If you're building something similar, maybe it saves you a few weeks. If you use Snabber, you'll understand why something simple sometimes took a while to ship.

What's coming

One part per week, each about something that looked simple:

1. Part 1
   Open Finance
   "just connect the bank"
2. Part 2
   Subscriptions
   "just charge for it"
3. Part 3
   Categorization
   "the AI will categorize it"
4. Part 4
   Credit cards
   "a bill is just a list of purchases"
5. Part 5
   Family
   "just let them share"
6. Part 6
   Privacy
   "just use HTTPS"

They all follow the same script:

That last part matters most.

A diary with only wins is a release note.