grin on Warpcast

grin pfp

just spent an hour helping a friend debug some lovable-generated code. he's on hour 6 with this bug super bearish on vibecoding with llms. they're great for prototyping but cannot be trusted we need a different model architecture that can actually understand and reason about the code

13 replies

3 recasts

37 reactions

phil pfp

give it six months

4 replies

0 recast

17 reactions

Drew Volpe pfp

Part of me agrees but also so much software in the world is little, in-house apps that are used by 3 people and are very janky. If people can write these themselves without needing a programmer, there’s huge value in that.

1 reply

0 recast

2 reactions

Mark Fishman pfp

agreed, I think the path forward isn't "non-coder uses LLMs to code their product it's "non-coder uses LLMs to prototype their product, gets enough traction and clarity to hire engineers to build the product"

1 reply

0 recast

2 reactions

Shira Stember pfp

I do not have a technical background, but feel more tech saavy than most. I was PUMPED to experiment with these new tools that claim to "turn your ideas into apps with AI" so I could to start building things on my own. But after spending time learning how tools like replit & cursor work and trying to turn big and small ideas into something I am convinced these tools are not yet built for me. They still require a foundational level of technical understanding in order for the code to be clean, functional, secure and not overly complex. I'm optimistic that they are moving in the right direction and things will only improve in parallel to me learning more, but perhaps there needs to be another layer or application developed that would provide a trust or security score to give vibe coders more confidence in their applications?

1 reply

0 recast

1 reaction

Garrett pfp

we're like 2-3 months in feels like the models will get much better at debugging and helping us mere mortals understand what's actually happening in the code

1 reply

0 recast

1 reaction

Peter Kim pfp

@ace was able to whip out a frontend prototype with lovable in a few days, but as soon as I wanted to make it production ready for a farcaster frame I had to make a new copy of the repo because of the limitations still not fully there yet

0 reply

0 recast

2 reactions

Ben pfp

Story of my life every time I try to push to production!

0 reply

0 recast

2 reactions

scottrepreneur pfp

@scottrepreneur.eth

great tools for shared understanding of a workable mvp. use demo data, anticipate the code is shit and hopefully you can use 30%. you still need to code after prototype. wireframe -> high fidelity design vibe proto -> real implementation The former are great for reducing rework in the latter.

0 reply

0 recast

1 reaction

Tudor 🟣🟡 pfp

that fits my experience of the past year and a bit. Having experience across the entire stack helps though, to ask the right prompts and guide it.

0 reply

0 recast

1 reaction

Sam Griffin pfp

I think if you're an excellent developer and can write excellent documentation, from what I hear from devs better than me, you can do a sort of documentation-driven-development approach But you have to have a fine intuition for the LLM limitations and be prepared for careful code review

0 reply

0 recast

0 reaction

Ivy pfp

0 reply

0 recast

0 reaction

L3MBDA pfp

debugging LLM-generated code can be a nightmare

0 reply

0 recast

0 reaction

Michael Zogot pfp

@michaelzogot.eth

Yeh, sometimes it gets thing done in a wrong way.

0 reply

0 recast

0 reaction