Login
You're viewing the functional.cafe public feed.

Replies

  • 💬 1🔄 0⭐ 0
  • Nov 9, 2026, 10:46 PM

    @civodul I don't think it is relevant what Anthropic/OpenAI are doing here? Existing open weight models work for this use case. I have used them. You give them a 150MB APK, tell it a few things about the app and you get back the backend api endpoints you can build code against. (This isn't making anything possible that wasn't before. But it makes things feasible that I just had no spoons to spend on previously)

    Takes like 30 minutes and costs 20ct or so in inference cost.

    💬 1🔄 0⭐ 0
  • Nov 9, 2026, 10:50 PM

    @civodul (And this is a tradeoff between hands on time spend, gaining a better understanding of this and just doing it autonomously. The autonomous thing works and is impressive but I'd much rather spend a few hours/days on this and understand it better than the half hour. But the point I was trying to make still stands.)

    💬 0🔄 0⭐ 0
  • Nov 9, 2026, 11:20 PM

    @civodul I have one more point I wanted to make here (but was searching for the reference), then I'll shut up.

    LLMs are actually (which surprised me a lot) really good for learning. Learning to navigate/understand a new codebase. Learning a new framework/language/etc.

    Nextcloud's Developer Relations person made a similar point here: youtu.be/K0s1XWklI7M?t=1850

    1/2

    💬 1🔄 1⭐ 0
  • Nov 9, 2026, 11:21 PM

    @civodul I don't agree with everything that's said there, but they also mention that they got an increase in quality and quantity of contributions for their desktop client (which is a complex, legacy codebase) due to LLMs, from people who use it well.

    2/2

    💬 1🔄 0⭐ 0
  • Nov 10, 2026, 6:32 AM

    @Bubu I am not a programmer. My professional work is in health care. But I am a techie.
    I got an EV and a tuya charger. There's a project for a local Home Assistant. It didn't support my device. Quickly the code was patched by ChatGPT, this charger is blocked off the tuya cloud and usable with local control and energy usage.
    LLMs helps people with less education with expressing their interests in a written form. Immigrants understanding and replying to written communication. The list goes on.

    💬 0🔄 0⭐ 0