Comment on Claude Code is suddenly everywhere inside Microsoft

Nioxic@lemmy.dbzer0.com ⁨1⁩ ⁨month⁩ ago

• 1984@lemmy.today ⁨1⁩ ⁨month⁩ ago

  My girlfriend is forced to use only Copilot at her work, and its just completely useless. She frequently posts me failed tasks it couldnt do. :).

  source

  • gravitas_deficiency@sh.itjust.works ⁨1⁩ ⁨month⁩ ago

    It’s only core competency is convincing idiot c-suites that they should force their org to use copilot

    source
  • 3abas@lemmy.world ⁨1⁩ ⁨month⁩ ago

    GitHub copilot within visual studio is not bad actually, if it doesn’t find anything to but typos then congrats your team is all competent senior developers…

    It does cut down review time for juniors’ work, and it’s capable of implementing full features given the correct documentation and instructions. It’s a useful tool that assists and doesn’t replace competent developers.

    source
  • Kissaki@feddit.org ⁨1⁩ ⁨month⁩ ago

    For copilot in visual studio at least you can select between models it will use, depending on the org setup showing them.

    source

    • lepinkainen@lemmy.world ⁨1⁩ ⁨month⁩ ago

      They’re all objectively shit.

      Even with Opus in the backend Copilot is so bad compared to Claude code

      source
  • JustEnoughDucks@feddit.nl ⁨1⁩ ⁨month⁩ ago

    I don’t know, that is about all Mistral can do too.

    That and python scipy is like a 50/50 for relevant code snippits.

    It had never once output compilable embedded code, even when I have tried to directly lead it there.

    source
  • k0e3@lemmy.ca ⁨1⁩ ⁨month⁩ ago

    Seems to be the worst Ai by far.

    Which is weird because isn’t ChatGPT supposed to be one of the best (least shittiest)? Isn’t CoPilot rebranded ChatGPT?

    source

    • lepinkainen@lemmy.world ⁨1⁩ ⁨month⁩ ago

      The model (sonnet, opus, gpt) is the engine, the framework (copilot, Claude, codex) is the rest of the car

      If you put a V8 Hemi in a shitbox, it’s still a shitbox.

      source
    • 1984@lemmy.today ⁨1⁩ ⁨month⁩ ago

      No, gpt is much better. I dont know how it works behind the scenes but I think Copilot is what Microsoft built and gpt is what open ai built.

      source
• MadPsyentist@lemmy.nz ⁨1⁩ ⁨month⁩ ago

  Like intelisence which has been doing it faster, more consistantly and offline since the mid 2000’s?

  source

  • lepinkainen@lemmy.world ⁨1⁩ ⁨month⁩ ago

    Intellisense can review a 1000 line PR?

    source

    • MadPsyentist@lemmy.nz ⁨1⁩ ⁨month⁩ ago

      Of course? Just as quick as it can candy colour the braces and highlight your references of your 1000 line pr

      source
• douglasg14b@lemmy.world ⁨1⁩ ⁨month⁩ ago

  We have ours configured pretty well at this point and it does an absolute bang-up job at code review. It will point developers towards tools and utilities that our monorepo has If they appear to be reinventing the wheel. It will catch common problems and issues that we have in our instruction file, and does a pretty good job of catching bugs and small issues like typos and what not.

  Perhaps you need to tighten up your instruction file?

  source

  • mech@feddit.org ⁨1⁩ ⁨month⁩ ago

    Yeah, LLMs are pretty great contextual search engines if you curate a good dataset for them to draw from.
    They aren’t a magic bullet that’ll find what’s relevant to you from the vast ocean of shit that is the internet.

    source
  • MonkderVierte@lemmy.zip ⁨1⁩ ⁨month⁩ ago

    Can you share yours?

    source