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

Building for an audience of one: starting and finishing side projects with AI

codemade.net

Post reply on HN
Print view View original

21–30 of 70 posts

Re: Building for an audience of one: starting and finishing side projects with AI

#21

Post by ath3nd »

echelon wrote: post #20
michaelteter wrote: post #16

I have more than a few side projects that began as late night discussions with an llm. A couple of those projects reached a level of completion where I use the products daily, and one project reached production (a game you can find referenced on my profile). I have had similar experiences to the author, and I’ve found that just working with a single agent in Antigravity (on the Gemini Pro subscription) is adequate. T…

Anyone who hasn't tried AI for coding absolutely should. This is the future. I don't think we'll ever manually write code again. It's just so much faster.

> This is the future.

NFTs and crypto were also the future.

> I don't think we'll ever manually write code again. It's just so much faster.

More work for the people who like to fix tech debt.

Re: Building for an audience of one: starting and finishing side projects with AI

#22

Post by xyzsparetimexyz »

> Start with a conversation, and explore the problem space with the LLM. The idea here is to gather options and ideas. Once you have a clear vision of what you want to build, ask for a detailed specification. Iterate on the spec until you understand it fully and are happy with it.

Maybe its just the specific language being used here but I really hate talking to these things. They inject way too much personality into things, especially Claude and are still too sychophantic and could lead you down a wrong path. I'd much rather just give them instructions.

Re: Building for an audience of one: starting and finishing side projects with AI

#23

Post by sicher »

I'm in the same camp. The last few months I've been building a couple of applications (editors) for my own work - and since it's so fast I've had Claude spin off to build Zig tools and libraries for markdown parsing, PDF generation, a Scheme implementation for embedding and more. (If anyone's interested they are at my Codeberg: https://codeberg.org/sicher)

Re: Building for an audience of one: starting and finishing side projects with AI

#25

Post by echelon »

ath3nd wrote: post #21
echelon wrote: post #20

Earlier quoted context omitted.

Anyone who hasn't tried AI for coding absolutely should. This is the future. I don't think we'll ever manually write code again. It's just so much faster.

> This is the future. NFTs and crypto were also the future. > I don't think we'll ever manually write code again. It's just so much faster. More work for the people who like to fix tech debt.

You need to try it before you knock it.

I was a doubter. This will literally work 100x faster than you. It can one-shot 1kLOC across dozens of files in mere minutes and understand the context.

You'll need to pay back a lot of those performance gains in reviewing the code, but the overall delta is a 2x speedup at minimum. I'd say it's closer to 4x. You can get a week's worth of work done in a day.

A human context switches too much and cannot physically keep up with these models. We're at the chess take off moment. We're still good at reviewing and steering.

Re: Building for an audience of one: starting and finishing side projects with AI

#26

Post by arjunbajaj »

This resonates with me a lot, and well-timed too!

I've always been unhappy with the way tasking/todo app (don't) work for me. I just started building a TUI in Zig (with the help of Codex) to manage my daily tasks. And since I'm building it just for me, the scope is mine to determine too.

Re: Building for an audience of one: starting and finishing side projects with AI

#27

Post by embedding-shape »

hmcamp wrote: post #24

I’ve had similar results with multi-agent approach. It consumes lots of tokens, required more setup and at the end of the day had pretty much the same output as if I had used a single agent.

You have any concrete examples of source code that your multi-agent approach has produced?

Re: Building for an audience of one: starting and finishing side projects with AI

#28

Post by embedding-shape »

realharo wrote: post #17
parpfish wrote: post #6

Earlier quoted context omitted.

i really want to see the silly little side projects that everybody is making! not because i think i'll actually use any of them, but because they could inspire me to do something different in my silly little side projects the goal isn't "product release", it's elementary school "show and tell"

Usually online communities have dedicated days for such things. Like a "side project Sunday" or such, with one large thread.

HN has that every day, and a dedicated section for it: Show HN. Link in the top bar :)

Re: Building for an audience of one: starting and finishing side projects with AI

#29

Post by citizenpaul »

>The age of actually finishing side projects is here

This is a really good summary of how I've experienced AI put into words. I'm not really sure how this can be monetized though.

I'm not going to burn $200-1k per day on agents to do some side projects that have been on the back burner. The only reason I'm doing it now is the heavily subsidized or free available models all over the place.

Post reply on HN