Manus, the AI agent company, has added an image design feature that combines generation and targeted editing in a single interface. Design View, announced in late December, is available to all Manus users on web and mobile and is aimed at one of the persistent annoyances of AI image tools: having to re-prompt or switch applications to change a single detail.
The centerpiece is what Manus calls the Mark Tool. After generating an image, a user can select a specific region, such as a piece of furniture, a color or a decorative element, and give an instruction that applies only to that area. The rest of the image is left untouched, which preserves parts that were already correct and avoids the drift that occurs when an entire picture is regenerated.
How the workflow works
On desktop, the user clicks on the part of the image to change and types the revision. On mobile, a long press selects the region, after which the instruction can be given by voice or text. Manus says projects carry across devices, so work started on a phone can be continued on a desktop.
According to the company’s documentation, Design View is intended for tasks such as posters, mood boards and product shots, and it accepts multiple reference images so that a series of edits keeps a consistent look. Manus says the system preserves the original image’s characteristics during edits, a claim based on its own testing.
Built on Google’s image model
Design View runs on Nano Banana Pro, Google’s latest image generation model, which is known for accurate text rendering, use of world knowledge and visual consistency across edits. Manus is one of several third-party products to adopt the model since its release, rather than training its own image system.
That choice places Manus in a crowded field. Adobe Firefly, Canva and a range of specialist tools already offer AI-assisted image editing, including region-based changes. Manus’s pitch is less about the image model itself than about integration: it positions Design View as part of an end-to-end agent workflow, where producing a visual is one step in a larger task the assistant is carrying out, rather than a standalone activity.
Early reception
Coverage from product trackers such as TestingCatalog has focused on the speed of the Mark Tool and the cross-device flow. Whether the feature changes how design professionals work will depend on edit quality in practice, which has so far been demonstrated mainly through the company’s own examples. For existing Manus users, the addition is included in the product rather than sold as a separate tier.
