also fwiw for small things like this, unless you want to really learn image recognition, just send the image to gemini-flash-3 or something. Sure it's 0.5-1s latency, still faster than entering it manually and it's pretty cheap, I'd reckon it's under the free tier at least for you and your family.
OurCar: What I learned making an app for my family
21–30 of 82 posts
Re: OurCar: What I learned making an app for my family
#22Earlier quoted context omitted.
It has a button, records button presses for the last 7 days, saves them to local storage. Then it presents totals per day and detailed timestamps for today's button presses. This is how I described it to the LLM (which wasn't even one of the coding assistants, just Gemini free). Not the exact prompt, that was more detailed, but that's the idea. I did like 3 iterations just to add features, because everything worked t…
Every time you finish a pack, save it. Seeing the empty packs will be more impactful. And they’ll take up space as an additional consequence of smoking more.
I'll try any advice but that doesn't really work for me because I already save them.
One day I'll run into something that works for me...
Re: OurCar: What I learned making an app for my family
#23Plenty of comments about using an LLM to assist with this, and I was happy to be able to read about a learning experience where the stakes were pretty low and the feedback loop pretty tight. Thanks for writing it up; for me, it reminds me that some of the use cases where an LLM might be an efficient tool are also the places where it can be wise to take the opportunity to learn and sharpen new skills.
Re: OurCar: What I learned making an app for my family
#24Best family app to me is home assistant.
It's so powerful and you can build so many custom UIs on it.
I started it for smart home automations but on daily basis I use it more for managing tasks,scheduling reminders.
And with Claude code remote even my not so technical wife uses it to build her tiny utility apps.
Re: OurCar: What I learned making an app for my family
#25Fun project. The article is a bit light on details. I find it astonishing that a project like this runs into performance issues. I would have liked to learn more about what these resource-constrained widgets looked like, what they did, and what caused the performance issues.
Re: OurCar: What I learned making an app for my family
#26Great writeup and also a great example of where LLMs can step in to help fill the gaps in areas where you don't have as much skill or interest. For instance, your wife used ChatGPT to come up with a name and you used AI to generate the admin flows that you weren't interested in building. Sounds like Flutter was a good technology choice too, given its flexibility across platforms. As a designer, I know how frustrating…
Flutter was good, but now with Liquid Glass™ I find react native (specifically expo) using expo-ui far far better at designing apps that match their native look and feel.
Heh, this is indeed rather unfortunate. Right as I finally got the look and feel to a relatively polished point, Liquid Glass dropped, making it rather obsolete. But to be honest, the app doesn't feel that out of place anyway. I don't know if that says something about me or about Liquid Glass, but it does say something.
Re: OurCar: What I learned making an app for my family
#27also fwiw for small things like this, unless you want to really learn image recognition, just send the image to gemini-flash-3 or something. Sure it's 0.5-1s latency, still faster than entering it manually and it's pretty cheap, I'd reckon it's under the free tier at least for you and your family.
That's actually not a bad idea, if I pick this up again, I will definitely try that. Thanks.
Re: OurCar: What I learned making an app for my family
#28Next: "How I made one whole website to host one HN-featured blog post"
Re: OurCar: What I learned making an app for my family
#29Manual data entry is just too unreliable and time-consuming. I don't see how this could work short of integrating OBD-II fuel consumption data combined with some sort of presence tracking.
I agree with you that at a large scale this is a problem. The one thing I decided very early on (when thinking about if/how I would grow this app beyond my family), is that this is for people who trust each other. If you don't trust the other people in your group to be responsible with the vehicle, then this app isn't gonna help. I'm trying to make the experience more smooth, not necessarily completely painless.
Regarding automatic tracking tho, the easiest option is Bluetooth based, the app knows the identity of the car, when the phone connects, it knows you're probably starting a trip. And the same goes for connected car APIs, those can much easier take the place of an OBD-II reader.
Re: OurCar: What I learned making an app for my family
#30Hmm an app where you can count the users on your fingers, and where it's not a big deal if it's slightly wrong. Safe to LLM generate it, unless you want to learn something in the process, in which case do whatever parts you want to learn about manually. Had an 100% generated app with one user - me - on my phone's home screen since some time last year.
I don't strictly disagree, but as I mentioned in the article, those weren't really an option at the time. Claude code was nascent and not very reliable, and I actually like writing code sometimes, heck, I'm currently rewriting this site by hand.