Gemini gets Notebooks; Colab learns
Google added persistent project workspaces called Notebooks to Gemini that sync with NotebookLM, and it also launched a Learn Mode in Colab that behaves like a personal coding tutor inside existing developer workflows. These moves shift product design away from one-off prompting toward stateful, task-scoped workspaces and embedded assistance inside tools engineers already use. The practical takeaway is that product ML is leaning into continuity, context management, and tighter integration with developer tooling rather than isolated chat experiences. (blog.google)(blog.google)
Most chatbots forget the project the moment the chat ends. Google’s April 8 update tries to fix that by giving Gemini a saved workspace called Notebooks and giving Google Colab a tutor mode that stays inside the coding notebook people already use. (blog.google 1) (blog.google 2) A workspace is just a place where the files, instructions, and past conversations for one job live together instead of being scattered across tabs. Google says a Gemini notebook can hold chats, documents, portable document format files, and custom instructions so the model has the same context each time you come back. (blog.google) Google already had NotebookLM, which is its research tool for asking questions over your own sources with citations. In January 2026, Google let people pull a NotebookLM notebook into Gemini as a source, and this week it turned that link into a two-way notebook that appears in both apps. (workspaceupdates.googleblog.com) (support.google.com) That two-way link is the part to watch. Google’s help page says that if you rename a notebook, add sources, or change custom instructions in Gemini or NotebookLM, the same notebook updates in the other app too. (support.google.com) The two apps still do different jobs. NotebookLM answers from the notebook’s sources only, while Gemini can answer from those sources plus web search and other Gemini tools, so the same notebook can act like a strict study binder in one place and a broader assistant in another. (support.google.com) Google also kept some NotebookLM features exclusive to NotebookLM. Audio Overviews, Video Overviews, infographics, and slide decks stay in NotebookLM even when the notebook started in Gemini. (support.google.com) (blog.google) Access is not universal yet. Google says Notebooks in Gemini are rolling out on the web first to Google Artificial Intelligence Ultra, Pro, and Plus subscribers, with broader access coming later, while the NotebookLM help page says the integration is available only in supported regions. (blog.google) (support.google.com) The Google Colab side is aimed at a different problem: code assistants often dump finished code into a notebook, which helps you ship faster but can leave you no better at Python the next day. Google’s new Learn Mode tells Gemini in Colab to teach step by step instead of just writing the answer for you. (blog.google) Google paired Learn Mode with Custom Instructions, which are saved at the notebook level inside Colab. That means a teacher can set rules like a preferred library or coding style once, and anyone who opens that shared notebook gets the same tailored assistant. (blog.google) Put together, these two launches point in the same direction. Google is moving its artificial intelligence products away from one-off prompts and toward saved environments where the model remembers the project, inherits the right instructions, and shows up inside the tool where the work is already happening. (blog.google 1) (blog.google 2)
