Last week OpenAI launched dots: an always-on AI agent for everyone
They are not alone. In five months Google, Microsoft, xAI, Meta, Apple and Anthropic have all shipped personal AI agents, and they are converging on the same idea: it remembers you, it acts for you, and it keeps working when you walk away.
I have been building my own, LillyRose, for almost a year. This week she learned to take phone calls. If she can't answer straight away, I hang up, she keeps working, and she rings me back when it's done. The ears and voice are OpenAI's GPT-Live; the brain is Claude, with my memory and tools.
It's not perfect. She can take 40 seconds to think, and once she mixed up the names of two sleeper trains. Then again, OpenAI's own live demo stalled on stage too.
I think one day everyone will have their own Doraemon. The question is who holds the pocket, and whether we still make our own decisions.
Would you want one? Tell me what you think in the comments 👇
Want to build your own? My AI Agent course starts from zero. Sign up: anzonwong.com/teaching-booking/
