A tangential note -- these days when I'm casually reading through job postings, a lot of companies mention in their job requirements something along the lines of "Must be familiar/knowledgeable about GenAI technologies (RAG/MCP)" even for roles that are not strictly related to GenAI.
When you look deeper into most of these companies, they seem better off using third-party services (like Claude/ChatGPT) instead of building their own GenAI models, MCP servers and whatnot. This is not considering that the domain they are operating in would have no apparent and obvious gain from utilizing GenAI (sure, they can buy ChatGPT/Claude organization subscription and that's all they need). I don't know if this is the indicator of how big this AI bubble is, but it sure is interesting...
When you look deeper into most of these companies, they seem better off using third-party services (like Claude/ChatGPT) instead of building their own GenAI models, MCP servers and whatnot. This is not considering that the domain they are operating in would have no apparent and obvious gain from utilizing GenAI (sure, they can buy ChatGPT/Claude organization subscription and that's all they need). I don't know if this is the indicator of how big this AI bubble is, but it sure is interesting...