zoozla

This is the result of about 10 months of sporadic work and two restarts from scratch. I never thought it would be worth sharing, but here we are.

Among other things, I needed to manage accounts in 4 currencies (USD, CAD, ILS and EUR), track transactions from ILS to USD to USDC (!) to CAD to EUR (don't ask), figure out all the fees I paid for this madness (please do ask), and deal with a dip so low that I only had 2 weeks of runway left, Canadian banks that randomly break Plaid imports, and Israeli banks that export XLS (no even XLSX!) as the only export format.

There is no way in hell an off the shelf budgeting app could handle any of that, and a bunch of vibe coded shell scripts that I started with can't tell me at a grocery store if I can afford beef or just chicken today.

A lot has gone into my financial recovery, but this app was one my most important tools. The other one is an emotional processing and stress relief tool that helped me deal with the series of shit storms, but that's for another day.

Oh that's odd. We use it both on Android and an iPad. Will look into it, thanks!

Thanks for pointing these things out, although dangerous is probably overstating it a bit. The points you mentioned at are easily fixable - there are few missing phonemes that I haven't recorded, the voice can be changed from British (it's actually Australian) to US in the settings, and the word no is often taught as a sight word because it comes much earlier than the full "o" sound, but you can tweak which words are treated as "tricky" in the settings as well.

That's one of the reasons I made it open source - you can point your agent to it and adapt it to any other language. You'll need to record the phonemes yourself through, haven't found any AI model that can do it well.

Made it encrypt the data at rest and support multiple users and published it on https://whatsapp.finereli.com if you don't want to install your own and just want access to your WhatsApp messages (groups included!) with your agents.

Limited to the first 100 logins for now so I can figure out how much this costs and if I can support this at scale.

Enjoy!

I used some skills I picked up 20 years ago refactoring a huge and very messy codebase I knew nothing about and built a skill that allows Claude to do the same.

Seems to work really well by executing very small and very safe refactoring steps that don't break functionality.

Very curious to hear how this works on large codebases and in different circumstances and how else we can improve it so that refactoring becomes a solved problem.

This is excellent, just what I needed from a barebones SEO tool as a total newb.

Simple, clear, gets the data right from my Google Console setup and shows a simple graph and a list of keywords.

Would love to connect on Twitter and offer more feedback/support if you're there.

Sounds like bad codebase compounded by a toxic team. A lethal combo.

That sounds aweful. I didn't think there were still shops that don't even have source control and just edit code in prod.

Mine was huge Java monolith that ran (and maybe still runs) part of the foreign exchange system that banks use to trade currency.

There were 8 layers of abstract classes with only one concrete class at the bottom.

There was a whole half-assed quasi-functional almost-but-not-quite-turing-complete language in XML for parsing legacy binary files.

There were unit tests, but they modified the dev DB so you needed to manually reseed it after every run (defintely no CI).

I spent months trying to refactor some of the more afwul bits. Not sure I actually made things any better.

I still shudder.

I can imagine a bot that posts a "Show HN" for each of those ideas that points to a landing page generated with GPT3 and DALL-E and tracks traffic to figure out the best one to build.

Might need a little tweaking. GPT3 skews to 5 y/o humour without deeper guidance.

...

Generate a list of 5 cringy app names and descriptions.

1. "PooPee!" - A game where you must help a little brown turd make its way through a maze to the toilet.

2. "Bathroom Break!" - A game where you must help a person take a dump in as little time as possible.

3. "Pimple Popper!" - A game where you must pop as many pimples as possible in a set amount of time.

4. "Hangman With a Twist!" - A game where you must guess a word or phrase, but with a hangman twist!

5. "Who Can Pee the furthest?" - A game where you must see how far you can make your pee go.

I removed the worst of the cringy ones (e.g. AnalConnect for business analysts) and the "Inspiration" bit in each app can sometimes ring a little off, but I'm pretty happy with the result overall.

Heard it a million times before but it's a nice reminder.

You're right, that's what the article is about, because that's easier. Finding paying customers and product market fit for an existing product is harder, and that's what I teach at growthlab.so.

People seem to be enjoying this over on IndieHackers, thought you guys might be interested in this too.

I do and it is!

Ping me on Twitter @finereli, let's connect!