Small publishers lost 60% of search referrals. Google beat a publishers antitrust suit. For brands, the game is becoming a source inside the answer layer.
Gemini Canvas Expands to U.S. Search — Google Turns Answers Into Workspaces
Google made Canvas in AI Mode available to all U.S. users — no Labs opt-in required. Search is no longer just about answers. It is becoming a workspace for creating documents, tools, and prototypes.
Google just turned Search into a workspace. Canvas in AI Mode is now available to every English-speaking user in the United States — no Labs opt-in required. Instead of just getting an answer, you can open a persistent side panel directly inside Search and start building things.
This is not a minor feature addition. It is a structural change to what Google Search does.
What Canvas Actually Is
Canvas is an interactive workspace that lives inside Google's AI Mode. When you enter AI Mode and tap the "+" icon, you can select Canvas to open a side panel that stays active as you continue your search session. According to Google's support documentation, Canvas is "an AI-powered collaboration tool in AI Mode that helps you work through plans, projects, and things you want to get done."
The workspace supports two primary modes. The first is document drafting — you can write plans, reports, business proposals, creative writing, and study materials directly in the panel, refining through conversational follow-ups. The second is interactive tool creation — Canvas generates functional HTML, CSS, and JavaScript code from plain-language prompts, producing working prototypes that run directly in the browser.
That second capability is where it gets interesting. You can describe a budget calculator, a meal planner, or a quiz, and Canvas will generate a working version in the side panel. You can toggle between the live preview and the underlying code, test functionality, and iterate with natural language instructions.
How It Works Under the Hood
Canvas is powered by Google's Gemini model architecture. According to technical guides, users can choose between Gemini 2.5 Flash (optimized for speed and quick iteration) and Gemini 2.5 Pro (which produces "much higher-quality, stronger code" for serious projects). The recommendation from practitioners is to start with Pro for anything beyond throwaway prototypes.
What differentiates Canvas from standalone AI coding tools is its integration with Search infrastructure. Canvas taps into Google's Knowledge Graph and live web results to ground its outputs. When you ask it to build something, it pulls the freshest information from the web and structured data relationships to contextualize the output. AI Mode's query fan-out architecture — running multiple related searches concurrently across subtopics — feeds into Canvas, giving it a richer information base than a standalone chatbot would have.
The feature supports multi-modal input: text, images, documents, and video screen recordings. Outputs are shareable via links, and code can be exported for use in professional environments.
From Answers to Starting Points
Canvas's arrival in mainstream Search represents a fundamental shift in what a search engine is supposed to do. Google Search has spent 25 years optimizing for one interaction pattern: query in, links out. AI Overviews already disrupted this by synthesizing answers directly. Canvas goes further — it makes Search a place where you create, not just consume.
Consider what this means in practice. A user searching for "budget tracker spreadsheet" no longer needs to click through to a template site, download a file, and customize it. They can describe what they want in AI Mode, open Canvas, and have a working interactive tool in seconds. The search result is the product.
This has direct implications for content publishers, template marketplaces, and any business whose model depends on search-driven traffic to functional resources. When Google embeds the creation step inside Search itself, the incentive to click through to external sites drops further.
Competitive Context
Canvas is not the first AI workspace to ship. OpenAI's ChatGPT Canvas and Anthropic's Claude Artifacts both offer similar functionality — side panels where users can generate, edit, and interact with code and documents. But Google's version has a structural advantage: it lives inside the world's dominant search engine.
That distribution matters. ChatGPT and Claude require users to navigate to a separate product. Google Canvas meets users where they already are — in a search session — and extends that session into a creative workflow. The integration with Knowledge Graph data and live web results gives Canvas a contextual grounding that standalone tools have to work harder to achieve.
There is also a cost advantage. Canvas is free, with no subscription required. Competitors like Lovable, Bolt, and Base44 charge subscription fees for similar code-generation capabilities.
The Limitations
Canvas is not a production development environment. It generates browser-based prototypes — functional for demos, personal tools, and quick utilities, but not architected for enterprise-scale applications. It lacks user authentication, database integration, and advanced security features that production apps require. Google is positioning Canvas as a rapid prototyping and planning tool, not a replacement for professional development workflows.
Currently, the feature is limited to desktop and available only in the United States in English. The U.S.-only deployment likely reflects Google's strategy of monitoring adoption patterns before expanding internationally.
What to Watch
The signal here is directional. Google is moving Search from an information retrieval tool to an action platform — a place where the output of a search is not just an answer or a link, but a functional artifact you can use immediately.
For marketers and publishers, this accelerates the need to rethink what "search visibility" means. If users can build a budget tracker, draft a business plan, or generate a quiz directly inside Search, the traditional funnel — where search drives traffic to your site — loses another step.
The brands and publishers that will maintain relevance in this environment are the ones whose expertise gets embedded in AI-generated outputs, not just linked to from search results. Structured data, authoritative content, and Knowledge Graph presence become the mechanisms through which your information surfaces — even when no one clicks through to your site.
Google is not just changing how people find things. It is changing what people do after they find them.
James Calder is the editor of The Search Signal, covering AI-powered search, generative engine optimization, and the future of brand discovery.
Small publishers lost 60% of search referrals. Google beat a publishers antitrust suit. For brands, the game is becoming a source inside the answer layer.
Perplexity's Personal Computer turns a Mac mini into a persistent, always-on AI agent that works across your local files and apps around the clock — a new category of ambient computing.
