In my master thesis, I explored how large language models could function as a core layer of an operating system, shifting interaction from apps toward intent. I frame this as an evolution of operating systems: the first generation organised around single applications, the second around ecosystems of connected services, and a potential third generation built around language , where interaction happens through conversation and systems interpret context, goals, and behaviour. This project reflects my belief that LLMs enable the next step in human–machine interaction, moving beyond the digitisation of tasks toward more personal, adaptive computing experiences.