Manus has introduced a new generation of products with Manus 2.0 and a separate app for personal AI agents called Cue. The interesting step goes beyond better text responses: Agents are supposed to receive their own contact addresses and computers, while creative work becomes editable projects. Announced in late September, these offerings target people who want to delegate tasks and then develop the results themselves.
Key takeaways
- Manus introduced Manus 2.0 and Cue on September 28, 2026. Bloomberg independently reported on the product launch.
- Cue is a separate app for personal agents. The company lists an agent’s own email address, phone number, wallet, and computer as its equipment.
- Manus Studio includes a video editor with an editable timeline and a game development environment.
- Cue launches in a free early access phase requiring an invitation code. The announcement still described the iOS version as pending App Store review.
- These capabilities are company claims. They do not establish independently verified reliability in everyday tasks.
Cue gives agents a way to be reached
Bloomberg describes Cue as a standalone platform built on Manus’s technical infrastructure. Its own communication and payment tools are meant to enable personal tasks on a user’s behalf. That distinguishes the product idea from a chat window that is useful only during a conversation. A dedicated point of contact could keep a task together across multiple responses, rather than making the person act as a constant intermediary.
The underlying Manus documentation already describes an isolated virtual computer environment with internet access and a persistent file system. That explains the computer part of the idea: It is a workspace for software, rather than another physical machine that has to sit on a desk. Such a system can bring files and tools together in one environment. Which capabilities a particular Cue account actually unlocks must be checked in the relevant offering.
An illustrative assignment might be to contact several event venues and compile their responses into a comparable overview. The agent’s own address could serve as the return channel. This is a use case example, not a test we conducted. Its value would lie in keeping questions and results attached to the assignment. Having an email address alone, however, says nothing about whether prices, dates, or availability are understood correctly.
For payments, the company’s wording is narrower than the image of a completely unrestricted digital employee: The agent is meant to pay within a budget set by the user. The announcement therefore establishes an intended spending limit, rather than universal acceptance by merchants. Nor does it establish confirmed availability of every phone feature in Germany. Anyone trying the service should check those specific conditions before assigning a concrete task.
Manus Studio turns the first draft into a project
The creative side of the launch is more immediately tangible for many readers. Manus’s official video editor uses a full timeline: Clips, text, and audio remain editable. The product page lists the desktop app for macOS and Windows. Users can import their own footage, change individual elements, and export the finished video. Further editing is therefore part of the offering.
This could provide a practical starting point for a small community organization or business: Have existing footage assembled into a first clip based on a clear description, then make targeted changes to the sequence, text, and music. This is a suggested application, rather than a claim about measured time savings. The benefit of a timeline is that a useful shot can be preserved while only another component changes.
This complements recent developments in AI voice models at the level of the work tool: A good audio track is one building block; a finished film that remains editable is another. This connection does not establish an integration of Eleven v4 into Manus. What matters here is making generated media accessible within a project so that people can make the final decisions themselves.
According to its product page, Game Dev also takes an iterative approach. Users describe a game idea and style, play the first version, and request changes to rules, controls, or visuals. Published games run in a browser and can be shared through a link. A small game with a clear objective would therefore be a good first attempt. Keeping the scope manageable makes it easier to see whether a change actually improves the game.
Separate access from performance promises
The launch announcement lists Manus 2.0 for the web, desktop, and mobile devices. Cue is described separately as free early access requiring an invitation code; iOS was to follow App Store review. This is the status at announcement, rather than a guarantee that every user can join immediately today. A limited early access phase also says nothing about later pricing. Cue’s free phase must not be generalized to every Manus offering.
For the technical foundation, Manus introduces the name Cascade. It refers to the company’s agent orchestration system, which organizes the agent’s work. What matters more to readers is what their particular assignment yields: a result they can assess, a project they can continue editing, and an effort that is worthwhile. General efficiency promises do not replace that concrete evaluation, even when a company publishes measurements from a single configuration.
Anyone who can try the new product generation should therefore begin with a task whose result is easy to assess. With a video, users can inspect the picture sequence and audio directly; with a game, they can try the controls themselves. Research involving responses from other people is harder to evaluate. These differences determine which starting point makes sense, without dismissing the underlying product idea.
The advance lies in continuing the work
Manus 2.0 brings together two interesting directions: Tasks receive their own digital workspace, and creative results remain changeable. Cue could extend the first direction into everyday communication; Studio makes the second visible as a concrete interface concept. Whether that translates into dependable assistance can only be judged through real assignments and the applicable access conditions.
The most compelling idea is therefore that work does not stop after the first generated result. A useful AI tool should allow changes, incorporate existing material, and build the next attempt on what came before. This provides a better measure of the product launch than the notion of a digital employee that supposedly does everything.