Grok Bot and OpenAI Dots look like they are pointing at the same thing.

Always-on agents with their own computers just became a category. Not one vendor in a funny costume. A class that includes Dots, Grok Bot, and peers like Meta Muse.

The stack has been walking this path for a while. Chat. Then chat with tools. Then agents. Now these Tamagotchi-like entities that live on a machine that is theirs, keep working while you are away, and ping you when something needs a human.

Personal experiments with OpenClaw and Hermes already felt like that. The enterprise-shaped versions are the same trajectory with harder lanes: narrow rights, isolated specialists, clean handoffs. An exploded production line where the system itself enforces the lanes.

This is harness and operating system evolution more than model cheerleading. Agent-first headless services are showing up where agents get first-class citizenship. It starts to look like a new App Store for agents: interactive, voice-ready, collaborating, scaling as needed.

The hard part is still ambiguous work. If you cannot capture it in a process or a runbook, the always-on agent stalls in the same place a human would.