Google is making it easier to turn ideas on paper into working prototypes. Its Gemini app can now interpret photos of hand-drawn sketches, generate the code needed to build an interactive app, and let users test and refine the result directly within Gemini.

Announced on November 5, 2026, the capability uses Gemini’s Canvas tool to transform visual ideas into interactive experiences. Rather than starting with a detailed written description or manually coding an interface, users can simply upload a photograph of their drawing and ask Gemini to build it.

Google says the feature can work with different types of sketches, including app mock-ups, habit trackers, and game boards.

How Gemini turns drawings into apps

The process starts with a photograph of a hand-drawn design. Users upload the image to Gemini and select the Canvas tool, which provides the environment for generating and working with the application.

From there, they can enter a straightforward instruction: “Can you code this app?”

Gemini then interprets the uploaded visual and writes the code to recreate the idea as an interactive tool. The generated application appears in a preview window, where users can test how it works rather than simply viewing a static representation of their original drawing.

The experience also supports further adjustments. Users can ask Gemini to change colors, fonts, or features, with modifications appearing in the preview in real time. This makes it possible to refine an initial concept through conversation without having to redraw the design or start the development process from scratch.

From paper sketches to AI-assisted prototyping

The new capability gives Gemini another way to help users move from an initial idea to something they can actually interact with.

For example, someone planning a personal habit tracker could sketch its interface on paper, photograph the drawing, and ask Gemini to turn it into an application. Similarly, a game concept drawn on a notebook or an early app interface could become an interactive prototype that users can experiment with.

It could also be useful for designers and developers who want to explore interface ideas before committing time to writing code manually. A rough visual concept can serve as the starting point, while natural-language instructions help refine the resulting experience.

However, Google’s announcement does not provide technical details about the underlying code generation, the complexity of applications Gemini can handle, or how accurately it reproduces more complicated drawings. It also does not establish whether the generated applications can be exported and deployed as standalone products outside Gemini.

For now, Google is highlighting the ability to generate, preview, test, and modify interactive tools within the Gemini experience.

The announcement adds another dimension to Canvas, which has been evolving beyond a space for working with text and code. With visual sketches now serving as potential starting points for application development, Gemini is bringing more of the early prototyping process into a conversational AI interface.

Google has not announced separate pricing or specific subscription requirements for this particular capability.