Will personal and work AI split apart in the agent era?
The author argues chatbots could combine personal and work use in one product, but agents involve identity, permissions, data ownership and security boundaries, so the two will split into separate stacks, citing people who use Grok Bot for work and Muse or Instinct for personal tasks.
In the chatbot era, @ChatGPT / @claudeai could comfortably combine personal + work in one product. The more agentic AI becomes, the harder it is to put these two worlds into one “super app.” Chat is fundamentally passive. For chatbots, the difference between personal and work is mostly just context. In the agentic era, the split is much deeper: identity, permissions, data ownership, security boundaries, and risk models are all different. I’ve been hearing people say “I use @bot Grok Bot for work, and @Muse / Instinct for personal.” ChatGPT / Claude today still bundle both use cases. @ChatGPT consumer usage is 70% non-work; http://Claude.ai is roughly 42% personal. If agentic computing really splits by principal, that bundle may start to unwind. Personal AI could increasingly be owned by $Meta / $Google / $Apple / dedicated personal-agent startups? Work AI could increasingly be owned by solana:7gnye7MymAa2iN8iZsZvH6ybjxeUiaxSoavxgeqG8kkh / $Google / enterprise software / vertical agents?