Grok Bots become shared AI coworkers with Team Bots
Team Bots turns Grok Bots into shared agents that can work with a team's files, applications, skills, and memories. Available in public beta on Teams and Enterprise plans, they can operate inside Slack, coordinate coding agents, and query internal company data.
One Bot for an entire team
An agent dedicated to a customer account can accumulate months of information without starting over when someone rotates off the team. That is one of the workflows behind the new Team Bots, a collaborative version of Grok Bots.
Each Team Bot is built around a shared role or workflow. It receives files, instructions, and skills, then connects to applications such as Salesforce, Notion, or GitHub through plugins. Shared credentials can also provide access to third-party APIs that do not have a plugin.
Memory completes the setup. The Bot's skills are shared across the team, while each person's conversations, individual context, and memories remain separate. An AI coworker inside Slack
Team Bots have their own Slack handles and can be invited into channels. Multiple people can ask questions, contribute context, and see the Bot's responses in the same place.
SpaceXAI, for example, uses a separate Bot for each major sales account. Overnight, it reviews customer news, Gong calls, Notion documents, and relevant Slack threads. It then posts a morning briefing covering what changed and drafts potential next steps for different members of the account team.
As teams make decisions, the Bot retains that context. When someone new joins an account, they can query that accumulated history instead of manually reconstructing information scattered across multiple tools. From bug reports to coding agents
The engineering workflow described by SpaceXAI goes beyond retrieving information. Its internal Team Bot connects to Notion, Linear, Hex, Datadog, and Cursor. It follows product decisions, triages bug reports, creates tickets, and can hand well-defined fixes to Cloud Agents.
Skills also teach the Bot the team's shipping process, including how PR reviews should be handled, when human input is required, and what evidence a change needs before it is considered complete.
SpaceXAI says it used this setup while building Team Bots itself. A Bot reportedly steered a Cursor Project orchestrating hundreds of Cloud Agents, helping a five-person team ship more than 100 pull requests per day. Those figures come directly from the company and are not an independent measure of engineering productivity. Marketing and data follow the same model
A Marketing Bot can receive a company's brand guidelines, blog posts, and social copy. When a draft appears in Slack, it can review the work against those references. For some website and SEO changes, it can then make the update and post a preview in Slack for final approval.
The internal Data Bot works differently. Shared read-only credentials provide access to approved Databricks tables, alongside connections to Datadog, Hex, and Statsig. Employees can ask one-off data questions without waiting directly for the analytics team.
SpaceXAI says it transferred a two-year-old skills library covering more than 45,000 tables into the Bot. Corrections to its queries can be retained and reused in future answers. Pre-built Bots to get started
Insurance company Harper provides one of the first external examples included in the Team Bots announcement. The company says it built a Bot in 24 hours to identify customers with lapsed policies, retrieve the required information across three platforms, and prepare personalized outreach.
Harper's CEO attributes more than $120,000 in customer savings across hundreds of policies to the system. The figure comes from Harper's testimonial published by SpaceXAI and has not been independently audited.
Team Bots is available in public beta on Teams and Enterprise plans. Pre-built configurations cover sales, product management, marketing, and data analytics. The announcement does not provide separate Team Bots pricing or a timeline for moving beyond beta.