Kicked it around for years, but CAD mold-making, especially multi-part mold-making, is reslly hard. Geometry, physics, understanding the process of using it… it was way more than I could find the time to develop expertise in.

Claude code, an MCP server to controll grasshopper (MCP server entirely written by claude of course), and maybe 8 hours total and I’ve got a fully algorithic multi-part mold-making program that designs the plastic tools I need to print, so I can pour silicone into plastic, so I can pour plaster into silicone, and the plaster parts fit goether perfectly for the final mold.

And it was fun. AI has brought joy back into coding projects I used to put off because they felt like a grind.

I still think that non-programmers are going to have a tough time with vibe coding. Nuances and nomenclature in the language you are targeting and programming design principles in general help in actually getting AI to build something useful.

A simple example is knowing to tell AI that a window should be 'modal' or that null values should default to xyz.

It's a massive accelerator for my dumb and small hobby projects. If they take to long I tend to give up and do something else.

Recently I designed and 3d printed a case for a raspberry pi, some encoders and buttons, a touchscreen, just to contol a 500 EUR audio effects paddel (eventide H9)

They official android app was over if the worst apps I ever used. They even blocked paste in the login screen...

Few people have this fx box, and even fewer would need my custom controller for it., build it for an audience of one. But thanks to llms it was not that big of a deal. It allowed me to concentrate on what was fun.

https://cdn.discordapp.com/attachments/1461079634354639132/1...

https://cdn.discordapp.com/attachments/1461079634354639132/1...

but there's a weird thing that happens when you build for yourself and don't read the code carefully. I caught one of my little internal tools storing API keys in localStorage because that's what the LLM decided to do and I never questioned it. for a personal tool running locally it probably didn't matter much but it made me think about what else I'm not catching. the whole "audience of one" framing kind of implies you can skip the boring review parts but idk, some of those boring parts exist for reasons even when nobody else will use it. I still vibe code everything now but I've gotten way more paranoid about actually reading what gets generated, at least the parts that touch anything sensitive

i feel like the expectations for a "Show HN" project are too high for a passing around a silly little toy that I had the robot throw together. product hunt is for things that are actual products/businesses. so maybe you throw it in a targetted subreddit for a niche interest group?

seems like there should be a marketplace for silly little side-projects, but i'm not sure how you keep it from getting overrun

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.

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. The extra perceived speed and power of multiple agents and/or Claude Code really didn’t match the output.

With a single Gemini (or sometimes switching to Claude Opus which inexplicably Google provides a generous amount of for free via AG) gives me incremental results so fast that I spend most of my time thinking about what I want (answering unplanned product questions or deciding how to handle edge cases).

I’m fact, sometimes I just get exhausted with so much decision making. However, that’s what it takes to build something useful; we just aren’t accustomed to iterating so fast!

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.

How brutal will the enshittification phase of these products be?

Will the 10x cost or whatever be something that future employers will have to pay, or will it be a more visible impact for all of us? Assuming no AGI scenario here and the investments will have to be paid back with further subscription services like today.

I really hope Open Source (Open Weights) keep up with the development, and that a continuation of Moore's Law (the bastardized performance per € version) makes local models increasingly accessible.

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.

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.