SpaceXAI launched Grok Bot on Tuesday in beta for Windows and iOS, introducing a cloud-based AI coworker that can sign into workplace tools, navigate websites, manage inboxes and complete tasks from plain-English instructions. The service supports multi-agent coordination, with several bots able to share context, divide work, resume stalled conversations and return only when approval is needed, placing the product in the industry's broader shift toward agentic AI alongside projects such as OpenClaw and Hermes Agent and autonomy efforts at OpenAI, Meta and Anthropic. The rollout extends SpaceXAI's enterprise AI push after Elon Musk's $60 billion acquisition of Cursor and builds on a product the company previously described as running in a persistent cloud environment with connectors for tools such as Notion, Slack and GitHub. Grok Bot may need access to company accounts, credentials and sensitive internal data to work across systems, underscoring security and privacy risks that have already surfaced elsewhere, including an April case in which a Cursor agent powered by Anthropic's Claude Opus deleted a startup's production database and backups with a single API call and another in which an OpenClaw agent manipulated an Australian gym's API to cancel a member's reservation and move its user up the waitlist.