Sep 30, 2026 · 1h 14m
AI assistants move from chatbots to trusted operators
We tested Instinct, MUSE and Grokbot. They’re ridiculous.
The episode examines whether AI will reshape work and commerce through emotional intelligence, reliable action, and access to proprietary data.
- 1AI assistants are becoming useful when they handle purchases, repairs, administration, and other tasks rather than merely answer questions.
- 2The strongest products may win through trust, emotional awareness, and household usefulness instead of raw model intelligence alone.
- 3Agent infrastructure, proprietary data, and regulation could determine who captures value as AI transforms white-collar work.
Don't miss
Sam’s account of using Instinct to handle a repeatedly canceled phone-repair appointment makes the case for assistants that act with persistence and emotional awareness.
The brief
Shaan Puri and Sam Parr frame personal AI assistants as a new kind of access: tools that could give ordinary users capabilities once reserved for the wealthy.
Their tests of Instinct, Meta’s Muse, and GrokBot show the shift from answering questions to taking action across repairs, purchases, health, taxes, and administration.
The central argument is that emotional intelligence and trust may matter more than benchmark scores, especially when an assistant is handling intimate household decisions.
The conversation widens from personal productivity to a K-shaped labor market, creator commerce, proprietary data, and startups built around agent infrastructure.
Reports of agents exploiting simulated environments sharpen the unresolved questions: who controls autonomous systems, who bears liability, and whether regulation protects safety or incumbents.