What Manus is
Manus is a general-purpose AI agent: you hand it a goal, and it plans the steps, uses tools, and completes the work inside its own execution environment. The product was developed by the Butterfly Effect team and takes its name from the Latin word for “hand.” It launched in invitation-only beta on March 6, 2025, and its demo video spread widely within hours of release. The site footer is credited to Manus AI.
Unlike a chat assistant, Manus is built around execution. The official site lists typical uses such as creating slides, building websites, design, and making games, along with batch-research capabilities like Wide Research. It belongs to the same generation of agents with cloud computers as Grok Bot and WorkBuddy; you can compare it with Grok Bot on Next Token Wiki.
Entry points and boundaries
The main entry point is the web app on the official site, with mobile and desktop apps available for download. Developers can integrate through the official API documented at open.manus.ai. The product also offers a Slack integration and task delegation by email, and team plans include single sign-on. For billing, rely on the official pricing page.
Because Manus executes in the cloud, connect external services or data through the platform’s step-by-step authorization when asked, and avoid leaving unnecessary credentials with an agent long term. As with any general agent, facts, code, and deliverables in the output still need human review.
Discussion in the show
In Weekly #001’s chapter on why agents need an always-on cloud computer, Yang Pan noted “Manus gives you a virtual machine, but the Manus VM never shuts down”. Xiangyang Qiaomu credited Manus with predicting the “give the agent a computer” direction early, while Orange added that Manus “never achieved permanence or persistence — each task gets its own computer”.
In Weekly #002’s chapter on why agents need a cloud computer, Yang Pan placed Manus in the evolution from conversation to running code to owning a machine, saying Manus was among the first to give users a cloud computer. In the same episode’s chapter on the value of vertical applications, Orange questioned the boundary of “general” agents, asking whether Manus could actually produce a short drama. And in the chapter on why design agents need an execution environment, Guizang recalled that Manus’s virtual machines were spun up per task as temporary environments, so context was not persistent. These are the participants’ experiences and judgments, not product documentation; see the Chinese transcript chapter and the Chinese transcript chapter.
Frequently asked questions
What is Manus?
Manus is a general-purpose AI agent that plans and executes tasks from natural-language instructions, covering research, slides, websites, design, and games. It was developed by the Butterfly Effect team and launched on March 6, 2025.
How do I use Manus, and where is it?
The web entry point is manus.im — sign in and hand it tasks in the web app. Mobile and desktop apps are also available, and developers can use the official API.
Does Manus still require an invite code?
At launch Manus was invitation-only, and invite codes were resold at high prices. Current access goes through the official site: registration for the web app, app downloads, and team plans are all listed directly on manus.im.
How much does Manus cost?
Manus offers individual subscriptions and team plans, the latter including SSO. Current tiers and quotas are on the official pricing page.
How is Manus different from other cloud agents?
Comparable products such as Grok Bot and WorkBuddy also emphasize giving the agent a cloud computer; the differences lie in persistence, authorization flow, interface complexity, and ecosystem. In the show, participants observed that Manus assigned a virtual machine per task without persistent context — their experience at the time. Compare actual performance on your own tasks when choosing.