Design question

[giteeisdumb]

Okay so I have one of very early versions of Auto-gpt and honestly I liked the simplicity and like a single level of nesting in terms of having to run a barebones minimal example. Auto-gpt was a cool showcase for interfacing a web fast and also doing some local operations with miminal setup using gpt. Why I say minimal is now I look at the repo why the bloat.

I understand the urge to like 'generalize' and 'expand' everything but I do have a feeling that past the first couple of months of development the practicality and simplicity of use of this project went down hill so bad where now I have to sit and dissect entire repo just to get it running. Like what's the point of that?

Plus this repo claims to be operating or existing in the service of openness and accessibility however since a massive improvement of running local llm models both in terms of performance and ease of use no progress of making any such integration showcase was made. Stuff like lm studio especially runs incredibly well on nvidia cards and without sacrificing any api access. To retrieve such info, what is it, one api and one json parse call? Why not focus on that for example? Enable people more openness not less or just abandon such a great start completely...

Is the publicity and eventual grift hole got better of maintainers? Why? Like that basically forces me to eventually substitute and lose hope in the repo long-term.

[ntindle]

We're still here! We had to get our feet under us. Sorry about that taking a bit and us struggling on answering questions like these but we've got a plan to specifically address the issues you've brought up here in regards to vendor lock in

#6969

We also have a general roadmap now! Take a look.

#6971