Skip to main content
Back

xAI's Grok Bot 1.0 Joins Video Calls and Remembers Every Team Decision

xAI's Grok Bot 1.0 turns AI assistants into shared teammates that join Slack, Teams, and video calls, taking initiative without being asked.

  • xAI launched Grok Bot 1.0, turning its assistant into a shared AI teammate for workplaces.
  • Bots can now join video calls, participate via chat, and track team decisions across sessions.
  • Team Bots work inside Slack and Microsoft Teams, learning from everyone on the team over time.
  • Proactive mode lets bots spot work, prep materials, and hand off tasks to other bots.
  • Available on Super, Ultra, Team, and Enterprise plans; Meetings require Ultra tier or above.
  • New safeguards enforce per-user permissions and flag sensitive content before sharing publicly.

xAI has released Grok Bot 1.0, the first stable version of its workplace assistant after an extended beta. The release expands the bot across Slack, Microsoft Teams, video calls, and connected apps, where it can follow group conversations, retain team context, and take on work. Meeting participation is the central addition.

The rollout announcement lists availability for Super, Ultra, all Team and Enterprise plans, and equivalent legacy tiers. Feature access varies by plan, while Enterprise administrators can enable or disable meeting participation.

Grok Bot 1.0 access at launch
Plan or role Access described
Super Grok Bot 1.0; the announcement does not detail Team Bot or Meetings access
Ultra Grok Bot 1.0 and Meetings
Team Grok Bot 1.0, Team Bots, and Meetings
Enterprise Grok Bot 1.0, Team Bots, Meetings, and an administrative meeting toggle
Equivalent legacy tiers Corresponding access based on the matching current plan

One memory across the team

Team Bots let several people work with the same AI agent across Slack and Microsoft Teams. According to xAI, each bot builds context from its interactions with the group, including project history, prior decisions, and information about how team members contribute.

Shared memory allows work to move between people and applications without rebuilding the prompt each time. A bot that helped research a launch can use that context when another team member asks it to update the brief or prepare material for a meeting.

Access can be limited to named people or groups, and connected services use the permissions of the person invoking the bot. That design constrains access to calendars, repositories, and inboxes, although teams will still need to test how reliably the bot handles information shared across private and public channels.

Slack thread showing Research Bot and Copy Bot coordinating on a headphone launch
Research and copy bots coordinate inside a Slack thread.

Meetings join the workflow

During a video call, Grok Bot appears as its own participant tile, listens to the discussion, and responds through meeting chat. Team members can ask questions, retrieve information from earlier conversations, track decisions, and assign follow-up work that continues in Slack after the call.

Continuity across meetings and messaging distinguishes xAI’s approach from dedicated meeting recorders. The same bot can attend a stand-up, retain its decisions, and later edit the launch brief, preserving one working context across those tasks.

Video call interface with a Research Bot tile alongside human participants
A Research Bot participates alongside human attendees in a video call.

Proactivity, with permission checks

Grok Bot can monitor ongoing project context, identify work that needs attention, and suggest or prepare next steps without waiting for a direct request. For example, a launch bot could detect an approaching deadline and assemble the required materials.

Bot-to-bot handoffs extend that model into simple multi-agent workflows. A Research Bot can pass findings to a Copy Bot, receive a draft for review, and involve a person when the task requires approval or judgment.

xAI describes several controls intended to limit unwanted actions and information exposure:

  • Permission inheritance: Bots operate with the access rights of the person invoking them.
  • Sensitive-action approval: Bots ask before performing actions that require confirmation.
  • Channel awareness: Bots check whether information is suitable for a public channel before sharing it.
  • Remembered authorization: Approval flows can reuse prior authorization to reduce repeated prompts.
  • Administrative oversight: New settings and analytics show what a Team Bot knows and how people use it.

Browser and connector upgrades

The stable release also includes capability and reliability improvements developed during the beta:

  • Browser use: More reliable navigation, smoother sign-ins, and fewer stalls during multi-step tasks.
  • Documents and presentations: Stronger writing tools and interactive Pages that teams can co-edit or refresh on request.
  • App connectors: Improved integrations with Gmail, Google Calendar, and GitHub.
  • Task efficiency: Faster execution, higher completion reliability, and lower resource use for complex browser, document, and presentation work.
Campaign bot summarizing lessons learned from edits to a project brief
A campaign bot retains lessons from edits to a project brief.

The continuity bet

Grok Bot 1.0 enters a market that includes Anthropic’s Claude for Chrome, OpenAI’s business connectors for ChatGPT, native assistants in Zoom, Google Meet, and Microsoft Teams, and numerous meeting-recording services. These products increasingly connect models to company data and let them act inside workplace software.

xAI’s pitch centers on one persistent agent carrying context and assigned work across chat, meetings, browsers, and connected applications. That continuity could reduce repeated prompting and manual handoffs, while increasing the need for precise memory boundaries, reliable authorization, and complete action logs.

Questions for a pilot

A controlled pilot should establish how Grok Bot behaves with real company data, changing permissions, and overlapping team contexts. Developers and administrators should verify:

  • Identity and access: Whether permissions update promptly when roles, groups, or connected accounts change.
  • Memory boundaries: Who can inspect, correct, export, or delete retained context, and how long that context remains available.
  • Meeting governance: How participants are notified, what the bot stores, and which consent or retention policies apply.
  • Action safety: Which operations require approval and whether prior authorization can be narrowed or revoked.
  • Auditability: Whether logs identify the requesting user, bot, source data, approvals, and resulting external actions.
  • Failure handling: How teams detect partial completion, retry failed tasks, and reverse unwanted changes.
  • Integration options: Whether xAI provides the APIs, SDKs, webhooks, or export tools required for internal systems.

The release defines xAI’s workplace strategy around a persistent agent that follows work across communication and productivity tools. Its performance in production will depend on permission fidelity, memory controls, audit coverage, and predictable behavior under real team workloads.