xAI launched Grok Bot on Aug 12, 2026, an always-on agent service that treats Grok as a set of AI teammates users can assign work to, rather than a chatbot they prompt one turn at a time. The bots run in their own cloud-based computer environment, sign into apps and websites the user already uses, and only come back when a task is finished or an approval is needed. It is xAI's clearest move yet into the enterprise agent market that OpenAI, Anthropic, and Microsoft have been staking out for the past year.
The service is available in beta starting today on desktop and iOS for three subscription tiers: SuperGrok Heavy, Cursor Ultra, and Cursor Teams Premium. Teams and enterprise customers can join a waitlist. xAI has not disclosed a standalone price for Grok Bot or given a general-availability date, and the beta requires users to hand the agents credentials to their online accounts.
The pitch is that no workflow authoring is required. Users message the bots "like a colleague" from a phone or desktop to hand off work, and the agents figure out the steps. Multiple bots can run in parallel, with one bot designated as a manager coordinating the others. The bots can "independently message each other" to share context when projects overlap, and can be dropped into a group chat where they assign ownership among themselves.
“Grok Bot (or Bots, as SpaceXAI inconsistently pluralizes it) is the latest push from Elon Musk's AI company to keep up with business services launched by rival AI providers”— Jess Weatherbed, News Reporter, The Verge
Key facts
- 01xAI launched Grok Bot in beta on Aug 12, 2026, an always-on agent service that runs in its own cloud environment.
- 02The agents sign into a user's existing apps, tools, and websites to complete multi-step work, returning only for approvals or when tasks finish.
- 03Multiple bots can run in parallel with one bot managing the others, and can message each other or coordinate in group chats.
- 04Grok Bot is available on desktop and iOS for SuperGrok Heavy, Cursor Ultra, and Cursor Teams Premium subscribers, with a waitlist for enterprise.
- 05The launch follows OpenAI's ChatGPT Work, Anthropic's Claude Cowork, and Microsoft's Copilot Tasks into the agent-for-work category.
xAI says Grok Bot is built on an internal prototype the company was already using for sales outreach, marketing, office operations, and bug fixes. Over time, the agents learn and save existing workflows, retain context about how a given user performs tasks to preserve voice, and follow up on dropped threads. The company frames the endpoint as agents "picking up work before you need to ask and knowing when something needs your attention."
The competitive frame is the more interesting part. OpenAI shipped ChatGPT Work, Anthropic shipped Claude Cowork, and Microsoft shipped Copilot Tasks — all variations on the same concept of an agent that lives inside a company's tools and executes multi-step work. Grok Bot is xAI's answer, and it arrives noticeably later than the incumbents. Elon Musk's company is not defining the category; it is trying to remain credible inside it.
The distribution choice tells the story. Bundling Grok Bot into SuperGrok Heavy and into two Cursor tiers means xAI is leaning on its consumer power-user base and on Cursor's developer subscribers to seed adoption, rather than leading with a direct enterprise sales motion. That is the opposite of how Microsoft rolled out Copilot Tasks, which went through existing Microsoft 365 contracts, and it reflects xAI's weaker enterprise footprint.
The functional bar xAI is setting — bots that share a cloud computer, sign into third-party accounts, and coordinate among themselves — is high, and the beta will be the test. Agent products across the industry have struggled with reliability on long-horizon tasks, with credential handling, and with the tendency to derail when a website's UI changes. xAI has not published benchmark numbers, success rates on standardized agent evaluations, or details on how the bots' cloud environment is sandboxed.
There are also unresolved questions about the security model. Handing an autonomous agent persistent access to work accounts is the exact attack surface researchers have flagged as the weak point of the current agent generation, and xAI has not detailed audit logging, permission scoping, or how a manager bot's instructions to a worker bot are constrained. Enterprise buyers will want answers before the waitlist converts. The company also has not said whether Grok Bot draws on Grok's most recent frontier model or a distilled variant tuned for tool use.
For xAI, Grok Bot is the clearest signal yet that the company sees the near-term business of large models as agentic work rather than chat. The consumer Grok product has traded on personality and access to X data; Grok Bot has to trade on getting tasks done in someone else's software stack, which is a different discipline. If the beta holds up, xAI has a credible seat at the enterprise-agent table alongside OpenAI, Anthropic, and Microsoft. If it doesn't, the gap between announcement and shippable product will widen at exactly the moment buyers are picking their standard.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
