AI giants trying to understand the output of LLM+"Agentic tools"=idea

I don't think the definitions of agents from OpenAI, Claude, and so on are being understood in the same spectrum. Microsoft understood it at some point when Satya Nadella took over, but then abandoned it, probably the reason they bought OpenAI. Google and Apple fumbled with their AI assistants that only ran on phones to schedule appointments. Meta is probably a bit more alive, as shown by its open-source projects. But none of the big players are building their foundations on the AI problem.

AI, once treated as candy or a toy by big tech companies and used internally or occasionally released to the public, is now something else. Just like Apple and Microsoft once fought over processor power, pricing, and design in consumer-level computers, the market had defined categories. Now, those categories shift with AI models with non-deterministic answers or application. There are real clients with real needs, companies with no guaranteed profits from their products, and the most flexible technology of all time.

In physics, the capabilities of an element being both flexible and resistant make it applicable to almost any mechanism or structure. AI is that element, and the scientific industries that discovered it can’t really figure out what structures or mechanisms to adapt it. The point is, big tech companies convinces itself of its project of creating AI, but they are not AI companies. They are the most hyped consumers of all, and that’s exactly why this is the right time to sell to the rich.