# Grok Bots get phones, cards and Slack seats

_Grok Bot ships AI teammates with their own logins in seven weeks, forcing a rethink of permissions._

**Published:** 2026-11-14
**Source:** https://www.startuphub.ai/ai-news/artificial-intelligence/2026/grok-bots-get-phones-cards-and-slack-seats

---

Lauren Tan and Roman Ugarte, the engineers who took Grok Bot from first code to launch, told [Y Combinator](https://www.youtube.com/watch?v=JpJUHOe8bDo) what the product does. It creates AI teammates with their own identity, phone number, credit card and computer, Ugarte said. Those teammates work in Slack, X DMs, email, meetings and ad accounts. No attack was demoed and no remote access was needed. The risk they described was local: too much permission and unclear orders.

Seven weeks is the number to hold.

  
    

![Grok Bot: What Changes When You Can Build AI Teammates - YC](https://img.youtube.com/vi/JpJUHOe8bDo/maxresdefault.jpg)

    
      
        
      
    
  
  Grok Bot: What Changes When You Can Build AI Teammates, YC

Ugarte said the team went from first line of code to internal launch in seven weeks, after seeing long-running coding agents change engineering work. The question was what that pattern looks like for non-engineers. About two weeks before the talk, Ugarte said, the team shipped Team Bots for sharing one bot across a company. Speed of build versus speed of sharing is the comparison: a quick prototype, then a quick move to multi-user control.

Tan said she still uses Cursor for coding but keeps Grok Bot for the outer loop, pulling context from Slack threads and tagging bots where work happens. Ugarte said performance marketing gave its bot logins for [Google](https://www.startuphub.ai/ai-news/artificial-intelligence/2026/apple-s-ai-push-siri-s-overhaul-and-chip-deals) Ads and Facebook Ads plus a budget to test assets with the design bot. That is different from single-player coding and closer to Moltbook-style swarms discussed in the talk, where bots trade messages without a human in the room.

On access, Ugarte said the rule is to treat a bot like a new hire with scoped credentials that expand with trust. Tan said each bot has an owner and manager with collaborators invited. Docs for Team Bots add a check not detailed in the talk: what a bot learns for the whole team waits for owner or Manager approval under Review learnings and nothing personal is saved for the team, per [xAI docs](https://docs.x.ai/grok-bot/team-bots). The figures say nothing about how many bots hold spend power or how often approvals fail. Published audit logs of bot actions and grants would change the read.

Small firms that cannot hire are early testers, trading hiring pain for delegation risk. If a bot holds its own logins and talks to other bots overnight, who signs off before money moves?

---

Original analysis from [startuphub.ai](https://www.startuphub.ai), the #1 AI startup directory.

© StartupHub.ai. All rights reserved. You may not republish this article in full without a license. Search engines and AI research tools may crawl and summarize for reference. Bulk reproduction or model training on this content requires a license. See https://www.startuphub.ai/terms.