OpenAI shipped dot this week, making "AI that works across your computers on your behalf" a primary product form — a clear upgrade from "chatting with you" to "executing tasks for you." GrokBot and a product called Muse are doing nearly the same thing in parallel. Three players are competing for the same seat: your personal AI employee.

What this is

Early users showed dot being used to: fix a project entry hidden by a new UI (edit in the cloud, test locally); review student video submissions — install Whisper to convert speech to text and read the video, switch to GitHub to inspect the code, and deliver feedback; even produce a 30-second product demo video in Blender running in the cloud. GrokBot leans on multi-bot group collaboration. Muse sells itself on "managing what your AI remembers." The common thread: let AI get the work done itself — you don't need to sit at your computer.

Industry view

What we find worth acknowledging: this wave of "personal assistants" is finally getting close to actually doing work. The tool-entry fix in the source — code upload, GitHub pull, UI check, local test, merge back to main — ran end to end on its own. The user only confirmed the result on their phone. That's the kind of execution past ChatGPT-style conversational AI couldn't deliver.

But the risks are equally clear. First, dot is still early-stage; the author themselves ran into issues that needed manual code syncing and entry fixes — "doing it itself" doesn't mean "doing it right." Second, dot can hop across computers, read GitHub, connect to Feishu, view local files — and enterprise IT will face a new problem: when employees let personal AI assistants connect directly to company machines, how does data compliance work? The source has zero discussion of permission boundaries or data flows — precisely the biggest blocker to enterprise rollout. Third, all three players are moving in nearly the same direction, all betting that "personal assistant" is the next entry point, but who ultimately wins remains unanswered.

Impact on regular people

  • Enterprise IT: Employees letting personal AI assistants connect directly to company computers and cloud tools means data leakage and permission holes will escalate from "asking ChatGPT a question" to "AI directly reading and writing company files." Compliance teams need to engage early.
  • Individual careers: Roles centered on looking up information, reviewing materials, and scoring code or video (analysts, operations, teachers, content moderators) will see value shift from "producing yourself" to "reviewing AI output" — not disappear.
  • Consumer market: We expect AI products priced by "task completed" rather than "conversation count" to appear within 12–18 months, redefining how personal AI assistants are priced.