Research idea:
MCP standardizes AI tools, but local models (Ollama, llama.cpp, etc.) still need a runtime to discover, call, and govern those tools.
I'm exploring a lightweight, model-agnostic runtime for MCP, local tools, and APIs. Bad idea? Tell me why.
@theo Things are way too chaotic right now to really know, imo. If the big guys have any say in it, they're going to want us using their datacenters lol. My bet is the future ends up being a hybrid approach.
@IamAroke AI wouldn't be able to build apps without frameworks built by developers. It still makes dumb calls all the time, so humans still need to lead development.
@Glensonis@kaaaash____ A new level of fraud >.< making it harder for us real developers get jobs cuz ya'll all out here faking expertise. I have no prob with AI assisted development for tinkering and prototyping, but pushing real production grade apps users actually use? That still requires expertise
@theo You mean you had AI do it lol. Not hating, just saying. I feel AI is making us devs lazier x.x. It is me at least. Some days I love it, some days I hate it. I don't miss diving into docs for hours upon hours, but the dopamine rush I'd get then vs now just isn't the same.
I think AI is making me lazy x.x . Why dive into the codebase when you can have AI do it? There is a million reasons, but I'm catching myself doing it less and less unless I absolutely have to... I don't think that's good for long term maintainability.... so much tech debt :,(
@thenowhereway cuz everyone one uses next and vercel + next were made for each other? The frontend +backend being in same repo + one click github deploys is pretty magical. makes it easier to work with for noobs and next is tightly integrated with vercel. Vercel abstracts so damn much.
@Im_IrushiK It means the exact same thing? The bar to entry is just lower. It does sicken me a little bit though lol. I'm glad I learned the hard way though ^.^