xAI opened an early beta of Grok Bot on August 11, 2026, a system of “always-on” AI agents built to work inside a user’s existing apps and finish multi-step tasks without constant supervision, the company said in a post announcing the product.
Unlike a chatbot that answers one prompt at a time, each Grok Bot runs on its own cloud-based computer and keeps working after a person logs off, according to xAI. “Grok Bot is your team of always-on agents,” the company wrote. “They have their own computer, work inside tools and apps like you do, and keep working 24/7.” The bots can sign into apps that lack a clean API, coordinate with each other by sharing context in group chats, and split a task among several bots at once.
Learns by watching
xAI said Grok Bot builds its own routines by observing how a user completes a task, then saves the sequence to repeat later without re-explaining it. The system is meant to run autonomously but still hands control back to the user when a step “needs your approval,” according to the announcement — a check meant to keep sensitive actions from completing unsupervised. xAI pitched the release around finishing work fully rather than partially: “There is a huge difference between 90% done and 100% done,” the company said. “Most AI gets you almost there. Grok Bot can finish the swing.”
Bundled with Cursor
The early beta runs on macOS and iOS, with xAI listing Android as a download option but no fixed timeline for a wider rollout. Access requires a SuperGrok Heavy, Cursor Ultra, or Cursor Teams Premium subscription, plus a separate waitlist for enterprise customers — tying the launch to Anysphere’s Cursor coding tool and extending Grok Bot into developer workflows alongside sample use cases xAI lists, including sales outreach, CRM upkeep, recruiting, and expense processing.
xAI describes the release as unfinished and says it is still working through known issues before widening access. Elon Musk said a broader rollout is planned alongside Grok 4.6, a successor to Grok 4.5, later in the week.
The launch adds xAI to a widening field of labs shipping autonomous AI agents meant to operate with less step-by-step human input — the same category where a recent industry index found most deployed agents carry no published safety evaluation.