Comment by scarface_74
1 month ago
How is being a wrapper for LLMs you don’t own any different from being a company based on cloud infrastructure you don’t own?
LLMs are a platform.
Bill Gates definition of a platform was “A platform is when the economic value of everybody that uses it exceeds the value of the company that creates it.”
It's relatively easy to move to different cloud infrastructure (or host your own) later on down the line.
If you rely on an OpenAI LLM for your business, they can basically do whatever they want to you. Oh, prices went up 10x? What are you gonna do, train your own AI?
Anyone who says it’s relatively easy to go to a different cloud has never led a major migration (I have). That’s kind of part of my day job - cloud consulting.
And if you think it’s hard to move to another LLM you haven’t done a major implementation using an LLM and used LangChain (I have). It abstracts a lot of the work and people can choose which LLM they want to use.
You don’t train your LLM. You use your LLM along with RAG.
I have no direct experience with this, but I’ve read that prices went down by 10x or so in 2024, and it seems that OpenAI has plenty of competition?
https://simonwillison.net/2024/Dec/31/llms-in-2024/#llm-pric...
A LLM wrapper adds near-zero value. If I type some text into a "convert to Donald Trump style" tool, it produces the exact same output as typing it into ChatGPT following "Convert this text to Donald Trump style:" because that's what the tool actually does. Implementing ChatGPT is 99.999% of the value creation. Prepending the prompt is 0.001%. The surprising fact is that the market assigns a non-zero value to the tool anyway.
Startups that use cloud servers still write the software that goes on those servers, which is 90% of the value creation.
I also think the proliferation of "GPTs" really took the wind out of thousands of these "NextJS Frontend + Custom System Context + LLM" wrapper apps as well.
https://chatgpt.com/gpts
That’s not what I see from the companies I work with (cloud consulting).
Almost all of them are using LLMs along with “tools” and RAG.