I'm also developing solutions with this, and everything will depend on how you contextualize the agents so they understand how your product should work within your market vision. Generating consistent and decent code works, but you need to manually persist (for now) the operational principles and inviolable fundamentals of your program. Bugs will always appear, but the speed of implementation is impressive and, in practice, becomes cheaper than hiring an entire team of developers.
The idea is a children to install it on their parents devices and do nothing more. No smart watch, no need to the parents to frequently click anywhere. Just keep their device with them. I did it that way on purpose. To be as less intrusive as possible. The app stays there and run 24/7 and detect locations, moment of activeness and rest, sleep, and creates behavior patterns. If any of the pattern gets violated, it send a notification. I give an example: my father do a daily walk between 8AM and 10AM. If he misses that one I will receive a notification. If my mother fails on the street and she does not move for some time, I will receive a notification with the exact coordinates.
:) It will be really stupid from my side to just let it go, without being sure that it will work. I am a professional after all, with almost 20 years of experience. There are 2000+ tests, 50K+ lines of testing code. I tested myself, and a lot of other people tested, with different devices. Covering all cases that we can think of. This is serious project, that I really hope that it will help a lot of people, and I really invested a lot of time and effort to make it stable.
I intentionally delayed the push to production with a few weeks, so that I am 99.9% sure that I covered everything.
Of course, we are human after all, and I will highly appreciate every effort and support from others, so that we can make the product do better. :)
Wow, you spent a crazy amount of time on something an experienced dev could hack together in a weekend without AI, and probably as little as a day with Claude Code.
I doubt. Experience dev can't make 400+ files and write 2000 tests in a weekend. At least, I was never that fast and productive ;)
With my experience, developing android apps in Java, 10-15 years ago, such project would need at least 2 developers, for 5-6 months of development and testing.
I think that you are underestimating the complexity. You need to deal with the specifics of different OEMs. You need to be absolutely sure that you are covering all cases, where the OEMs can stop the app. Testing and fine-tuning took more time, than I really expected
Then monitoring itself is not that easy as well. You need to distinguish between a planned vacation trip, and visit to a friend. And a lot, a lot more things.
To be honest, yes, it took me more time than expected. And now, with my current knowledge, next time, I will do it much, much faster, with less token. But this is how it is with every development. A junior developer will waste more time and money that an experienced one. With AI, most of us are like junior developers.
Most of the time, I waste, not with the app itself. But to keep it running 24/7, without in any way asking the parents to do any manual work. And this is because I did not know that Android based OSes are so aggressively killing long-running applications. I took my note :)
But to write it for a weekend, I just cannot agree. For a weekend, you can't make real and deep testing. A simple example: you need to run it for a few weeks, without opening it to see if Android will kill it, because it treats it as 'never used' app.
I'm also developing solutions with this, and everything will depend on how you contextualize the agents so they understand how your product should work within your market vision. Generating consistent and decent code works, but you need to manually persist (for now) the operational principles and inviolable fundamentals of your program. Bugs will always appear, but the speed of implementation is impressive and, in practice, becomes cheaper than hiring an entire team of developers.
Yes. you can't just leave it and go somewhere. Constant monitoring is required
does this go on your phone, or do you need a smart watch, or what? did you actually do any physical testing?
The idea is a children to install it on their parents devices and do nothing more. No smart watch, no need to the parents to frequently click anywhere. Just keep their device with them. I did it that way on purpose. To be as less intrusive as possible. The app stays there and run 24/7 and detect locations, moment of activeness and rest, sleep, and creates behavior patterns. If any of the pattern gets violated, it send a notification. I give an example: my father do a daily walk between 8AM and 10AM. If he misses that one I will receive a notification. If my mother fails on the street and she does not move for some time, I will receive a notification with the exact coordinates.
That simple ;)
> Is the code quality good? Honestly, I don't care. > The app shipped and looks stable — so the code is in decent shape
not generally things you want to fuck with when making claims about health or safety.
:) It will be really stupid from my side to just let it go, without being sure that it will work. I am a professional after all, with almost 20 years of experience. There are 2000+ tests, 50K+ lines of testing code. I tested myself, and a lot of other people tested, with different devices. Covering all cases that we can think of. This is serious project, that I really hope that it will help a lot of people, and I really invested a lot of time and effort to make it stable. I intentionally delayed the push to production with a few weeks, so that I am 99.9% sure that I covered everything.
Of course, we are human after all, and I will highly appreciate every effort and support from others, so that we can make the product do better. :)
Wow, you spent a crazy amount of time on something an experienced dev could hack together in a weekend without AI, and probably as little as a day with Claude Code.
This is like an anti-ad for AI and vibe-coding.
I doubt. Experience dev can't make 400+ files and write 2000 tests in a weekend. At least, I was never that fast and productive ;)
With my experience, developing android apps in Java, 10-15 years ago, such project would need at least 2 developers, for 5-6 months of development and testing.
An experienced dev wouldn't need 400 files and 2,000 tests just for this.
I think that you are underestimating the complexity. You need to deal with the specifics of different OEMs. You need to be absolutely sure that you are covering all cases, where the OEMs can stop the app. Testing and fine-tuning took more time, than I really expected
Then monitoring itself is not that easy as well. You need to distinguish between a planned vacation trip, and visit to a friend. And a lot, a lot more things.
To be honest, yes, it took me more time than expected. And now, with my current knowledge, next time, I will do it much, much faster, with less token. But this is how it is with every development. A junior developer will waste more time and money that an experienced one. With AI, most of us are like junior developers.
Most of the time, I waste, not with the app itself. But to keep it running 24/7, without in any way asking the parents to do any manual work. And this is because I did not know that Android based OSes are so aggressively killing long-running applications. I took my note :)
But to write it for a weekend, I just cannot agree. For a weekend, you can't make real and deep testing. A simple example: you need to run it for a few weeks, without opening it to see if Android will kill it, because it treats it as 'never used' app.