Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

I Tried Building a Real App with AI. It Took a Year

alexhyett.com

Post reply on HN
Print view View original

51–60 of 90 posts

Re: I Tried Building a Real App with AI. It Took a Year

#51

Post by InsideOutSanta »

oblio wrote: post #33
InsideOutSanta wrote: post #7

Earlier quoted context omitted.

If you read the article, you'll see that the author took the same approach one would use when writing an app manually from scratch: > I started with a simple hello world application. Then added in the habit list, the ability to add a habit, edit a habit and so on. This hands-on, incremental approach works, of course. But I don't think it is the most effective way of using LLMs to write apps.

> This hands-on, incremental approach works, of course. But I don't think it is the most effective way of using LLMs to write apps. 1. Which apps should be built like this? 2. Which apps shouldn't be built like this? My instinct is that any app worth building and using for a long time should be built incrementally. Otherwise, slop.

I don't think any app should be built like this. Rather, I think this approach should be used when you already have a large codebase and are making atomic changes to it. In fact, I think taking this approach for greenfield work is more likely to result in slop because early accidental decisions turn into fundamental design choices.

Instead, I've settled on a top-down approach: I write an in-depth spec for the whole app first, then have the LLM implement reviewable sections of it.

Re: I Tried Building a Real App with AI. It Took a Year

#52

Post by SpicyLemonZest »

Chance-Device wrote: post #20
over_bridge wrote: post #14

I've used AI for building a hobbyist note app myself. It's also a mixed bag but a lot of that comes down to how I've changed. Before AI dev was expensive so I'd spend hours planning features. I filled 5 notebooks with UI sketches and data schemas and lists of names of what to call things. Building was more or less final as I only had 5-10 hrs a week max and wanted to move on. Now I type directly into the Claude Code…

> AI has made it very easy to make a lot of bad apps quickly Is it bad from a user or performance perspective, or just untidy and unaesthetic from a developer’s perspective? Because the developer’s perspective is becoming increasingly unimportant as LLMs become the main readers and writers of the code.

Both. I debugged an issue a few days ago where a messy, incomplete migration was causing every agent that drove by to very confidently assert the wrong behavior and try to make breaking changes.

Re: I Tried Building a Real App with AI. It Took a Year

#53

Post by Chance-Device »

Ekaros wrote: post #49
Chance-Device wrote: post #20

Earlier quoted context omitted.

> AI has made it very easy to make a lot of bad apps quickly Is it bad from a user or performance perspective, or just untidy and unaesthetic from a developer’s perspective? Because the developer’s perspective is becoming increasingly unimportant as LLMs become the main readers and writers of the code.

Depends on how it manifests. If things change every app release sometimes for better sometimes for worse and nothing really stays consistent over time it is clearly bad for user. Users really like stability of their experience. They want the things that are working to keep working. If something is bit weird but acceptable they expect it to continue like that. Going back and forth between things might be the worst. Af…

For sure, but I think solid automated testing is what gets you to a good and consistent user experience. Validation of the outcome rather than the code itself.

Re: I Tried Building a Real App with AI. It Took a Year

#55

Post by oblio »

InsideOutSanta wrote: post #51
oblio wrote: post #33

Earlier quoted context omitted.

> This hands-on, incremental approach works, of course. But I don't think it is the most effective way of using LLMs to write apps. 1. Which apps should be built like this? 2. Which apps shouldn't be built like this? My instinct is that any app worth building and using for a long time should be built incrementally. Otherwise, slop.

I don't think any app should be built like this. Rather, I think this approach should be used when you already have a large codebase and are making atomic changes to it. In fact, I think taking this approach for greenfield work is more likely to result in slop because early accidental decisions turn into fundamental design choices. Instead, I've settled on a top-down approach: I write an in-depth spec for the whole a…

Most apps, just like most buildings, aren't greenfield, though :-)

Re: I Tried Building a Real App with AI. It Took a Year

#56

Post by malfist »

diskzero wrote: post #28
malfist wrote: post #21

Earlier quoted context omitted.

My weightlifting app sold me a pro unlock to get weight calcualors years ago. Then they introduced a subscription model to get that plus some features I didn't want, they promised my purchase would be respected for all time. Last September they declared that the app was "too valuable to give away for free" and told me I had to subscribe to use it. Even though it hadn't been free, I'd paid for it, and nothing about it…

That is rough. I assume this is a web application? I went through a previous thought experiment about what would happen when we moved a paid up front feature behind a subscription. This was the outcome I came to. Unless the user bought and installed a physical app, on the desktop, outside of an app store, they were going to get the rug pulled out from under them.

No, its the stronglifts app. But I switched to hevy and I'm happy with it. But I did paper for a while and liked it until I missed mu graphs

Re: I Tried Building a Real App with AI. It Took a Year

#57

Post by jorisw »

spprashant wrote: post #42
dpoloncsak wrote: post #36

Earlier quoted context omitted.

I will say, when working with the recent batch of frontier models, even the last batch, if you ask "Do you think we should xyz" it may sometimes push back for an alternate solution. Now, I can't promise it's advice is worth taking, but they do seem to be taking strides at judging the relevancy of some needless additions.

I agree with your observation. Although the framing of the question signals some of the response "Do you think.. " forces the model to check for the cost-benefits. We can definitely mould AI agents to think more critically about these things, I dont know how effective it will be in the long term honestly.

Recommend checking out Ponytail — a pi extension that behaves like a senior engineer who keeps the LLM code generation in check, and the processing of your prompts just the same

https://ponytail.dev

Re: I Tried Building a Real App with AI. It Took a Year

#58

Post by jorisw »

smalltorch wrote: post #15
spprashant wrote: post #11

I think we are moving from "can you build this product?" to "do i trust you to support this product?". Software is cheap, trust is a premium. Developers and companies who put real care into their product will come out on top. A large part of that is going to be, what not to build - something AI agents will always say yes to. To be fair, was always the case but its only going to be made more apparent as AI scales out.

What are the actual maintenance items you will run into for offline utilities if the initial version has been throughly tested and works exactly how you want it? Is it typically bug hunting on edge cases? Feature requests? If the initial product is polished enough, what can I realistically expect to run into down the line?

Software will never be truly stable as long as their environment (the world, ultimately) is not stable

Re: I Tried Building a Real App with AI. It Took a Year

#59

Post by fragmede »

customguy wrote: post #50
torginus wrote: post #32

Earlier quoted context omitted.

I have noticed the exact behavior change on myself that you describe. But, thanks to the magic of keeping track of time and tasks, I have discovered: - I am not all that faster when project timelines are concerned - I've often built things with AI with the latest and greatest technologies that I've always wanted to learn but never did, like write a microservice backend with the trendiest frameworks with infrastructur…

> I carefully considered execution flow and made sure the simple looking code actually did everything correctly That is what I love about coding. Not just insofar as it's necessary and useful (it is both), but just as a puzzle in its own right, to a fault. It's just a type of "imagination" that I don't really apply often otherwise, since e.g. I'm not into math, not an organizer or manager etc. Though I did have jobs…

omfg all those days and money lost to a single character.

Re: I Tried Building a Real App with AI. It Took a Year

#60

Post by Grombobulous »

I think the author arrived at the sort of bad middle trench of AI: where you use it knowing how the code works and trying to fit it into a paradigm (Xcode) that it’s not well-suited for.

The 6 months of cleanup is something a true no-code vibe coder would have never done. I also think that kind of person might be inclined to keep AI going down the simplest development lane possible, which probably means using a more web-native technology rather than a native app that relies on an IDE the AI can’t work well with.

Something like Expo, for example.

I found that the most time consuming part of iOS app development in a vibe coder scenario was the various “red tape” of the process for the App Store including submission.

Post reply on HN