Google adds coding and creative tools to Canvas in Search

Google opened Canvas in Search’s AI Mode to every user in the United States searching in English, removing the Search Labs opt-in and adding creative-writing and coding tools that can produce executable prototypes inside the browser. The company announced the change on its The Keyword blog, positioning Canvas as a persistent workspace where people can organize projects, draft documents and build simple apps without leaving Search.

Users enter AI Mode, select the Canvas option from the tool menu and describe what they want to create; a Canvas side panel produces an initial draft or prototype that can be refined through conversational follow-ups with Gemini. Google says Canvas draws on the real-time web and the Knowledge Graph to ground results, and that users can toggle to view underlying code, test functionality and iterate until the prototype behaves as desired. The company also makes Canvas available inside Gemini, and its paid tiers offer access to larger models and expanded context windows for more complex work.

The expansion follows a year of broader AI Mode rollouts. Alphabet disclosed in its Q3 2025 earnings call that AI Mode reached about 75 million daily active users after successive moves that opened the feature to all U.S. users, extended it to Workspace accounts and then to more than 40 countries in late 2025. Canvas had been limited to Google Labs experiments and early testers until this week, when the company removed the experimental barrier and folded the workspace into mainstream Search.

Google suggested a range of practical uses: students could build study guides by uploading notes, researchers might convert reports into web pages, quizzes or audio summaries, and everyday users could create dashboards to track information such as scholarship deadlines and requirements. For developers and hobbyists the headline capability is code generation: users can describe a game, app or interactive tool and watch Canvas generate runnable code that they can test and refine by chatting with Gemini.

“With Canvas in AI Mode, we’re making Search not just a place to find information, but a space where people can create and experiment with ideas in real time,” said Maria Chen, Google’s vice president of AI products.

The move formalizes Google’s strategy to turn Search into a productivity hub that brings generative-AI capabilities directly to billions of users. Embedding drafting and prototyping tools into Search lowers the friction for people who do not use standalone AI chat services and lets Google surface creation features during ordinary queries.

But the rollout raises immediate questions that Google did not resolve in the announcement. The company did not publish details on how user content in Canvas is stored, whether prompts and generated code are retained for model training, or how it will prevent dangerous or insecure code from running in users’ browsers. There was no timeline for expanding Canvas beyond U.S. English searches, and Google did not provide adoption figures for Canvas itself.

The capability to generate executable code inside a widely used search interface brings clear benefits for productivity and rapid prototyping, but also heightens risks around intellectual property, verification of source material pulled from the live web, and security of shared prototypes. Independent scrutiny of Canvas’ data-handling and code-safety controls will be important as more people use Search as a creation platform.