I built a receipt organizer and tracker app on iOS in May this year and now it has just past $200 in monthly recurring revenue and >$500 in December. Before this I have been collecting my own receipts and scanning them into a private website I built for myself, then one day I was like, why don't I make it into an app so it is easier to use and I can share it with others?

https://receipt-genie.com

https://apps.apple.com/us/app/receipt-organizer-receiptgenie...

What do you use for OCR, I wonder

if it's LLM isn't that too expensive (or flaky if cheap) to be scalable?

I was using an OCR service but I have migrated to use LLM recently. It is actually cheaper and more scalable because it is accurate enough and I can also get more information (guess currency and timezone, and translate, and even help tag the receipts) with a single prompt. Currently it sits just a tad bit cheaper than my previous OCR service per receipt scanned. Since I am providing this as an app instead of as an automation, there is UI for user to make edits and they seem to be happy enough about the results with very minimal need to edit.

I have also been playing with open source VLMs like Qwen3-VL and it was surprisingly not too far behind. I think the next generation I should be able to switch over and save a bit more on cost.

Thanks for elaborating! Btw, what model is currently cheap yet accurate enough for you?

Nice work and starred! I am curious how you get the embeddings computed, the # of dimensions for the embeddings and if you have run any benchmarks against OpenAI's offering?

Cheers!

The embeddings are computed using llama-cpp, but langchain makes a nice convenience wrapper to directly get them, so I use that. The embeddings are 4096 dimensional vectors.

And no, I haven’t benchmarked them against OpenAI’s embeddings. I should point out that this code will work for any model in GGML format, so if there are fine-tuned Llama2 versions that are optimized for embedding, you could use those instead very easily (or any other model). This project is more about making it easy to go from model to embeddings on demand via an API and then letting you do useful things with those embeddings easily.

Nice review! I am going to try it out now!

I believe they run the dialogs into a program and generate the sounds from it. I play AC in Chinese and the words do sound like Chinese. It is just crazy to wrap my head around the amount of effort they put into this game.

I feel like a big portion of the apps you accomplish similar things. To name a few, Airtable, Notion, and Trello can all do CRM funnel, product roadmap and issue tracking. Even though 2 out of 3 are free, you guys could probably condense them into a centralized spot.

Additionally, one could sacrifice on features and use GitHub projects instead of Trello. That’s what we do at work. Not great, but pretty painless for tracking dev work

They only spend $227/month in total. Why even bother centralizing? That's peanuts. I spend more than that on my personal AWS bill.

Funny and here i am trying to save 5 usd/month because my VPS is overpowered.

It really depends on where you live. 227usd is certainly not peanuts.

They're running a startup. If they can't turn $227 into orders of magnitude more they're not going to succeed.

When I was poor I stuck everything on very cheap machines too so I know what it means to have different constraints. Those constraints rarely mesh well with running a startup.

Is that for a personal or commercial project?

Both actualy i am freelancer and these two are sort of interchangeable.

I think they have a task worker system built back in 2014[1], so they need to have something custom working with it as well. Back then I think they really didn't have many options, but if they were to do it again now, I think either AWS Lambda or AWS Batch will serve this type of scheduled job cases very well.

[1]: https://engblog.nextdoor.com/nextdoor-taskworker-simple-effi...

I think they just didn't publish to other regions. Give it some time and I believe they will cover more than just US App Store ;)

If you think about similar products on the market (fb messenger, WhatsApp, even Line or Kakao Talk in Japan/Korea), this is quite unheard of, because WeChat is taking up everyone's time on the phone. Other apps just simply cannot divide a user's attention out of WeChat the app, on a smartphone.

And just because of that, most of the businesses and companies start to live off of wechat's user base and features. For example, there are subscription accounts where you get subscribed articles regularly so you don't need to build a news app anymore for content, and individual content producers can also publish just as easily, so the spiral gets deeper and deeper into the wechat system. And when payment started to come around on WeChat in 2014 (iirc), even Alipay is taking a hit, because now you literally can just keep WeChat open on your phone.

I think that, if you build an IM product and suddenly become the only app people ever use, this is just amazing.

> because WeChat is taking up everyone's time on the phone

Well, not me. I'm Chinese and I hate hate hate WeChat. It's a cluttered, privacy intrusive app and a very close eco system.

WeChat became this popular is by no means because of the app per se, but because of it is made and heavily promoted by Tencent.

Before WeChat we had much more elegant and simple apps like Talkbox and WhatsApp.

The phone number 114 in China is just like the 411 in the US, it is a directory service, so that's why China uses 114 for the DNS servers.

On that note, the Chinese DNS would not resolve google or facebook properly due to the Great Firewall of China, I wonder if Freenom can resolve those domains within China without any proxy solutions.

For Chinese tech news, you can get on Wechat and subscribe to 36kr 36氪. Or simply visit http://36kr.com/ . It is one of the biggest site for tech news, and they regularly translate popular articles from techcrunch or medium.