>Then I created HugstonOne Enterprise Edition 1.0.7 and it fundamentally changed how I think about AI.
Who uses versioning numbers like that?
>It’s a paradigm shift.
But we have local LLMs already
>You’re locked into a platform (with no easy way to switch)
I've never think about LLMs providers like a cloud provider. I can jump providers whenever I want. I jumped from OpenAI to Anthropic to Open router. Given the parity of tooling and quality of SOTA models I fail to see where the vendor lock is.
semantic versioning (Major.Minor.Patch) is widely adopted and was standard practice in the 90s.
Yes I know, but when I create and/or launch a software project I don't start with "1.0.7" if I'm being _semantic_. It's like a major version AND a patch already? It doesn't track.
TFA reads like AI slop honestly.
"Who uses versioning numbers like that?" Some coders obsessed with math, among others: Linux kernel: 2.4.0 (1999), Apache HTTP Server: 1.3.34 (2002) etc.
"I've never think about LLMs providers like a cloud provider. I can jump providers whenever I want. I jumped from OpenAI to Anthropic to Open router. Given the parity of tooling and quality of SOTA models I fail to see where the vendor lock is."
You didn't really read the article. You may jump but not your data, not the price and not the dependency. So yes they are a provider and a cloud, while with HugstonOne, the user becomes the provider.
And for the rest, I see you are very curious and have many questions. Maybe you can try the app, I am sure it will satisfy all your questions in an excellent manner.