Grok Bot Agentic AI
xAI quietly shipped the clearest signal yet that AI labs have stopped competing on chatbots.
Grok Bot went into public beta on August 11. Not a chat window — a system of persistent AI agents running on dedicated cloud computers that can log into websites, operate applications, work through your inbox, and execute multi-step tasks end to end, only pinging you when a real decision is needed.
It's available now to SuperGrok Heavy and Cursor Ultra and Teams Premium subscribers. And as of a week into the beta, xAI still hasn't published a standalone price or a flat per-agent fee.
That last detail is the real story. Every major lab — OpenAI, Anthropic, Google, now xAI — is racing to ship autonomous agents before anyone has settled how to price autonomy itself. Usage-based, outcome-based, seat-based: nobody's converged, and that ambiguity is quietly becoming a budgeting problem for the enterprises adopting these tools first.
The product capability is arriving faster than the commercial model built to support it.
If you're evaluating agentic AI tools right now, are you more worried about what they can do, or about not knowing what they'll actually cost you at scale?