SpaceXAI Launches AI Agent 'Grok Bot' in Beta

SpaceXAI, the AI development subsidiary of SpaceX, unveiled the beta version of AI agent 'Grok Bot' on August 11, 2025. With clear pricing starting at $120 per month, it marks the official market entry of an enterprise-focused AI work automation tool.

The fundamental concept of Grok Bot is that 'users can do other things while entrusting AI with their work.' Traditional AI assistants activate each time a user asks a question and provide an answer. In contrast, Grok Bot functions as a 'worker' assigned a specific role and operates continuously. It continues processing even when the user closes their computer, sending notifications only when approval is needed or when tasks are completed. Each Bot has its own independent computing environment, and multiple Bots can hand off work to one another.

According to SpaceXAI, Grok Bot was originally developed as an internal prototype and has been used by internal teams for purposes such as sales outreach, marketing campaigns, office tasks, and bug fixes. Based on these results, the company decided to roll it out as an external product. Supported platforms include macOS, Windows, Linux, and iOS, with Android support planned for the near future.

The pricing structure comes in two types. For individuals, it is included in the Cursor Ultra plan at $200 per month. For enterprises, it is offered as Cursor Premium Teams at $120 per seat per month. Both include a computing environment for Bot operation, access to various tools, scheduled execution, desktop and mobile operation, and expanded AI token limits. SpaceXAI acquired Cursor, known for its code editor, for $60 billion in June of this year, and Grok Bot is provided as a service integrated with Cursor's existing plans.

Grok Bot is entering an already intensifying AI agent market. Anthropic introduced computer operation capabilities for Claude in 2024 and released Claude Code for developers and Claude Cowork targeting white-collar desk work in 2025. OpenAI added app operation functionality to Codex in April of the same year and subsequently launched Workspace Agents, which can integrate with third-party apps, and ChatGPT Work, which handles long-term, multi-step tasks. This demonstrates the major companies' focus on this field.

Grok Bot distinguishes itself from existing AI agents through its design philosophy of a 'persistent digital worker' with four key elements: role, memory, habit, and handoff. Rather than simply executing instructions, it learns accumulated work patterns and can pass tasks to other Bots. However, SpaceXAI has not published benchmark metrics regarding Grok Bot's task execution performance as an agent. It remains to be seen how trustworthy it will be in real-world settings, which will become clearer through ongoing beta operations.

The AI agent market is currently at a stage where companies compete to prove 'whether it works.' For enterprises to integrate these tools into daily operations, processing accuracy and risk management for operational errors are essential. While Grok Bot's track record of actual internal use provides some credibility, how reliably it performs in diverse external environments requires future verification. Competition in this market is not just about product maturity but also about trust—the extent to which enterprises can delegate operational control to AI. All eyes are on the next moves of each company.