Dots vs Grok Bot vs Meta Muse: three always-on agents, one disclosure

OpenAI Dots, SpaceXAI's Grok Bot and Meta Muse all promise an agent with its own computer. Here is how they compare, and why I disclose that I use Grok Bot.

Disclosure: I use Grok Bot. No sponsorship, no affiliate links.

Three companies now sell the same idea: an agent with its own computer that keeps working when you are away. Here is how they differ, based on the companies' own pages.

Meta Muse

Meta introduced Muse on September 8 as a personal agent. It runs on Muse Secure VM, a dedicated cloud computer with its own browser. A separate Sentinel agent on the same machine approves anything that reaches the internet. Muse checks with you before sensitive actions such as sending an email or buying something, shows an audit trail, and can check out with Link by Stripe. It is rolling out in the US on iOS, Android and muse.ai and is free for most uses, with subscription plans. Muse for Small Business, announced on September 29, adds connectors such as Shopify, Slack and QuickBooks.

OpenAI Dots

Dots launched at DevDay on September 29, powered by GPT-6 Astra. Each dot has its own cloud computer and browser, and you can reach it in ChatGPT, Slack, Teams or by call. It is available on Pro plans outside the EEA, UK and Switzerland, on Business Premium, and on Enterprise when an admin turns it on. Before an action affects your accounts, a review step decides whether it can proceed or needs your approval.

Grok Bot

SpaceXAI describes Grok Bot as a team of always-on agents with their own computer that sign into your tools and come back only when something needs approval. It is in beta for SuperGrok, SuperGrok Plus and SuperGrok Heavy, and for Cursor Pro, Pro+, Ultra and Teams subscribers, on desktop and iOS. It has its own usage allowance, separate from Grok and Cursor plans.

The fine print

All three lean on approvals, and none of them removes the need to check the work. The New Stack reports that OpenAI's own Astra system card shows more boundary problems in longer chains: when tasks in a sequence went from five to ten, flagged samples rose from 8.6% to 19.7%, with no high-severity breaches found. That is a useful reminder that permissions drift as work piles up.

Dany's take

Same promise, different fine print. Pick based on where you already work: Slack and ChatGPT for Dots, Meta's apps for Muse, Cursor and Grok for Grok Bot. Whatever you choose, keep sending, buying and deleting behind approval, and re-state the limits when a task chain gets long. Which one would you trust with your inbox?

Sources: Grok Bot, Meta Muse, Muse for Small Business, OpenAI Dots, The New Stack.