Manus 2.0 Brings Remote Agents and Video Editing
Manus 2.0 expands its agent platform from task execution into mobile-controlled video editing, multiplayer game hosting, and remote workflows.
From browser tasks to persistent workspaces
Manus has introduced version 2.0 with a broader set of capabilities that includes video editing, multiplayer game hosting, and remote agent execution from a phone. The release pushes the product beyond the familiar model of asking an agent to complete a short browser task and waiting for a result.
The key change is continuity. A user can start or supervise work remotely, while the agent operates in an online environment that remains available beyond a single chat turn. Video editing gives the system a creator-facing workload; multiplayer hosting tests whether an agent can configure and maintain a live service rather than merely generate text or code.
That product direction reflects a wider shift in agent design. The market is moving toward systems that combine models with tools, hosted computers, files, applications, and long-running state. The model is only one component; reliability depends on how the platform handles permissions, interruptions, cost limits, and recovery when an action fails.
Manus 2.0 also makes the competitive field harder to define. It is no longer competing only with chat assistants or coding agents. It is approaching the territory of cloud workspaces, creative software, and lightweight developer infrastructure. That may broaden the addressable market, but it also multiplies the ways an agent can make an expensive or irreversible mistake.
Why it matters
The release is significant less for any one demo than for the platform thesis: agents are becoming remotely operated services with persistent environments. If the experience is dependable, the phone becomes a control surface for work that previously required a desktop; if not, the breadth of features may expose operational fragility.
Uncle Cat take: Manus’s real bet is not video editing; it is whether users will trust a phone to supervise an agent that keeps running after the conversation ends.