Google is preparing a Canvas mode for NotebookLM that would let users pin generated outputs, drag them across a workspace, and synthesize multiple Gemini chats into a single assembled document. The feature has not been confirmed by Google; TestingCatalog reported it on May 29 based on leaked screenshots and internal build flag inspections.
Canvas is the strategically significant item in the batch. NotebookLM already has a loyal base in research-heavy academic and professional-services workflows, but until now it has been a summarizer and a source organizer, not a place where the final document lives. A Canvas workspace changes that positioning. It makes NotebookLM a direct competitor to ChatGPT Canvas, which lets users edit and iterate on generated documents inside the ChatGPT interface, and to Claude Artifacts, which renders and refines structured outputs in a persistent side panel.
Two other features surfaced in the same TestingCatalog report. Personal Preferences would allow NotebookLM to retain user-level settings such as tone, citation style, and default sources across sessions; currently those preferences reset each time. Connectors would enable live-sync ingestion from Google Drive, Gmail, and reportedly third-party tools including Notion and Confluence, replacing the current model of documents frozen at the moment of upload. Together, the three features sketch a product that is trying to move from “research session” to “persistent research surface.”
The skepticism is structural. TestingCatalog’s evidence is leaked screenshots and internal flag inspections, not a Google product announcement. Features visible in internal builds frequently slip, ship with reduced scope, or get folded into other products. Google has given no confirmed launch date for any of the three. The gap between internal testing and general availability at Google has historically ranged from weeks to over a year; Google Assistant’s preview features in 2023 and 2024 provide recent precedent for features that cycled through internal builds without a clean public release.
The Connectors roadmap suggests Google wants NotebookLM to hold the live data connection, not just the summarized output. That is a meaningful architectural bet: a notebook that stays synced to a Drive folder or a Confluence space becomes harder to migrate away from than one that accepts uploaded PDFs.
The Canvas direction also re-anchors NotebookLM against the broader workspace race. ChatGPT introduced Canvas to make ChatGPT the place where the artifact actually gets written. Anthropic shipped Artifacts to make Claude the place where the artifact gets refined. Google has the structural advantage of already owning Docs, Drive, and Workspace, but it has not until now tried to make NotebookLM the assembly surface. If the Canvas mode ships at the quality the screenshots suggest, NotebookLM stops being a research utility and starts being a Workspace-adjacent product that competes for the same workflow.
For knowledge-work teams currently standardizing on ChatGPT or Claude as their primary document surface, NotebookLM’s roadmap is closing the feature-parity gap. Which surface owns the document is the lock-in question, not which model answers the query.
TestingCatalog reported the three unconfirmed NotebookLM features on May 29, 2026, based on leaked screenshots and internal build inspections.
