xAI launches Team Bots, AI teammates for teams

xAI has launched Team Bots, Grok bots that work with a team, use its tools, and build shared context so everyone can collaborate from the same information base.

The idea is simple: you create a bot for a shared role or workflow, give it access to the necessary files and applications, and share it with the people involved in that task. Each member can chat with it separately, but the bot retains the team’s shared skills.

Personal context remains private. Each user’s conversations and memories stay separate, while the bot can use the instructions, knowledge, and connections the team has configured. It can also participate in Slack under its own user account, answer questions in channels, and receive information from multiple people.

What a Team Bot can do

xAI presents four internal use cases to explain how the tool works.

Sales and customer support

At SpaceXAI, each major commercial account has its own bot. The sales agent, customer success manager, solutions architect, and sales lead all work with the same context.

Overnight, the bot reviews company news, recent Gong calls, Notion documents, and relevant Slack conversations. Every morning, it posts a summary in the account channel with important changes, next steps, and drafts tailored to each role.

It also records the decisions the team makes. When someone joins or leaves the account, the bot helps bring the new person up to speed without rebuilding the entire history from scratch.

Harper, an insurer specializing in small businesses, created a bot to identify customers with expired policies and help restore their coverage. According to xAI, the system reduced manual work and helped recover significant savings for its customers.

Product and engineering

The engineering bot connects to Slack, Notion, Linear, Hex, Datadog, and Cursor. It can track product decisions, classify bugs, create tickets, and launch Cloud Agents for well-defined fixes.

The team taught it its release process through skills, meaning reusable instructions about how to review changes, when to ask for help, and which tests are required before closing a task. It then reports progress and blockers in Slack.

xAI says it used this system while building Team Bots. One bot coordinated a Cursor project with hundreds of agents and helped a team of five people publish more than 100 change requests per day.

Marketing

Marketing Bot reviews copy and materials against brand guidelines, blog posts, and the company’s approved messaging. This allows regional teams to get feedback without waiting for headquarters.

It can also track website or SEO changes from review through preview. When the content meets the criteria, it publishes a link for final approval. The team can move forward without opening an engineering ticket, although every change still goes through the same checks.

Data and analytics

Data Bot answers targeted questions about data using read-only queries against approved Databricks tables. It also connects to Datadog, Hex, and Statsig to investigate results and create charts.

At SpaceXAI, this bot uses a skills library built over two years to work with more than 45,000 tables. It includes instructions for finding the right data, defining metrics, analyzing feature usage, and safely updating dashboards.

When someone corrects a query, the bot can retain that correction to improve the team’s later responses. The promise is not that it will always be right, but that its answers will become more consistent within the shared environment.

What changes for you

Team Bots is available in public beta for Teams and Enterprise plans. You can share a Grok Bot you have already created with your team or start with bots prepared for sales, product management, marketing, and data analytics.

The important change is that AI is no longer just an individual conversation. It can become a shared hub for tracking decisions, consulting information, and carrying out repetitive tasks. Even so, the result will depend on the permissions, connections, and instructions the team configures. The more sensitive the information, the more important it is to control what the bot can read and which actions it can perform.