Agents
Which AI agents work on their own?
In August and September 2026, three companies launched AI agents that keep working after you close the app: xAI’s Grok Bot, Meta’s Muse and OpenAI’s dots.
Each one gets a computer of its own in the cloud, signs into the apps you allow, and comes back to you for approval on the steps that matter. A chatbot answers when you ask. These agents keep going between conversations, and some of them talk to each other.
- Agents tracked
- 3
01
The agents, oldest first
- 11 Aug 2026
Grok Bot xAI
A team of bots that share a cloud computer of their own, sign into your tools and keep working around the clock. xAI says “Bots can independently message each other and share context in threads.” Its launch post doesn’t say which model they run on.
- 8 Sep 2026
Muse Meta
A personal agent on Meta’s Muse Spark model, in a virtual machine of its own, which “keeps working after people close the app.” A second agent on the same machine, called Sentinel, has to approve anything Muse sends to the internet.
- 29 Sep 2026
Dots OpenAI
Always-on agents on GPT-6 Astra, each with its own cloud computer, reachable in ChatGPT, Slack and Teams. OpenAI published safety tests for them on launch day. What are OpenAI’s dots? covers them in full.
When one of these agents does something it wasn’t asked to, and the record shows it meant to, the case goes on AI gone rogue.
02
Sources
These are the sources behind this page. We checked them against the original text on 1 October 2026.
- Lab publication
xAI. Introducing Grok Bot. 11 Aug 2026.
What Grok Bot is, and its bots messaging each other.
- Lab publication
Meta. Introducing Muse: The World’s First Personal AI Agent Built for Everyone. 8 Sep 2026.
What Muse is, and Sentinel.
- Lab publication
OpenAI. Introducing dots. 29 Sep 2026.
What dots are.