Manus 2.0 Adds Editable Games, Message-Triggered Tasks and Remote Computer Use

Manus 2.0 is available now, adding three ways to put its AI agent to work: users can edit generated games and videos, start tasks when messages arrive, and direct work on their computer from a phone. The release joins hands-on creation with tasks that can begin without a fresh prompt or happen away from the desk.

A game you can change after generation

The desktop app is now called Manus Studio. Its Game Dev and Video Editor environments let people manually adjust what the agent generates. A creator can revise the output rather than treat the first version as finished. That makes the editing tools part of the workflow, not just a way to view what Manus made.

For multiplayer projects, Manus says dedicated cloud computers can keep game servers running. That addresses a separate step after creation: keeping a game available for people to play together. Hosting does not, by itself, show how well a game plays or how much revision its maker will need to do. The editing environment is where that work would fall.

Tasks that start when a message arrives

An incoming email or Slack message can now trigger a Manus automation. Someone need not open the app to start each task when a selected event can set it off. That changes the decision from when to issue an instruction to which incoming events warrant work. The trigger tells the agent when to begin; users still have to judge whether the work it produces is right.

Computer access is a different decision

Computer Use lets Manus work with approved files and apps on a user’s own machine. The user can also direct it remotely from a phone. This is not the cloud computer Manus offers to host a multiplayer game: one keeps a game server running, while the other involves software and files already on the person’s computer.

Phone control separates directing a task from sitting at the computer where it happens. Approval still determines which local files and apps Manus can work with. For someone considering remote use, that choice comes before the convenience of sending an instruction from elsewhere: the useful setup is one that gives the agent access to the work intended, without approving more than the user wants it to reach.