It just seems like the applied field of ML/AI - especially with products based on LLMs - the barriers of entry are low to non-existent.

Whipping up some MVP product/functionality and interface where the LLMs do all the heavy lifting is a weekend project.

The only gatekeeper seems to be the cost of eventual pre-training, and the LLM queries / API calls themselves.