Gemini Notebook now allows you to frame a presentation or video before its final rendering: the tool offers themes from the sources of the notebook, lets the user confirm the direction, then can display a slide plan or production notes to read before generating. For a trainer, consultant or creator, real progress is not just speed. This is the possibility of correcting the intention, order and level of detail at a time when the changes still cost little.
Google announced this development on 19 October 2026. Deployment is global and progressive over the week. However, there is an important nuance: in 20 October 2026, official help pages indicate that co-creation routes with a detailed review of the plan are reserved for Google AI Ultra subscribers. The actual availability therefore depends on the account, subscription and interface used. This article distinguishes what is announced, what is deployed and what still needs to be checked before integrating the function into a professional process.
What Google has officially announced
So far, the simplest scenario was to load sources, ask for a Slide Deck or Video Overview, and then wait for the result. Gemini Notebook now adds an intermediate phase. In the chat, the user requests a presentation or video. The tool suggests topics based on the selected sources. After confirmation, it becomes possible to orient content, visual style and format. With the edit option enabled, the user can view the slides or production notes for the video, add their returns, and then choose to launch the rendering.
This function was presented ingoogle's official announcement of 19 October 2026. Google also states that it is possible to keep a fast track: from the Studio panel, it is still possible not to activate the rereading and generate directly. In other words, the new stage is not meant to slow down all uses; It is especially useful where the issue justifies prior validation.
The exact status deserves to be read with precision. The announcement talks about a global deployment to Gemini Notebook users during the week. It does not mention a list of excluded countries. France and La Réunion therefore enter the announced global perimeter, without guaranteeing that each account sees the function immediately. In addition, the help pages devoted to Slide Decks and Video Overviews indicate, at the time of our audit, that the co-creation steps with rereading are available to Google AI Ultra subscribers. Geographical deployment should be distinguished from commercial eligibility.
Why validate the plan before rendering actually changes the work
A final rendering impresses because it is visible. However, the most costly mistakes are often upstream: bad audience, too wide an angle, demonstration placed too early, conclusion without next step, example that proves nothing or video that goes over the central subject. Once the layout or video is generated, the user is tempted to correct the form rather than question the structure.
The rereading of the plan reverses this logic. It transforms the generation into an editorial process: sources, intent, architecture, validation, and then rendered. It is close to the work of a editor or director who validates a synopsis before launching the production. IA no longer receives only one order; It proposes an interpretation that the human can accept, reorder or refuse.
This step is particularly useful when the content must support an activity. A training presentation is not only used to summarize documents. It should guide educational progress. A consultant video must not only be correct: it must make an expertise understandable and guide a decision. A commercial medium must show the offer, the proof and the next step. To avoid clean but interchangeable results, the VIFLY guide on use of AI without producing the same content as everyone remains a good complement: the plan must look at you, not just reflect the average of the sources.
Two concrete cases for trainers and consultants
Example 1: a trainer prepares a workshop from its sources
A marketing trainer runs his program, three case studies, product documentation and frequent questions from his learners. It requires a slideshow of 25 minutes for beginners. Gemini Notebook offers a plan that starts with features, continues with market numbers and ends with an exercise.
Before rendering, the trainer finds that the order does not correspond to his pedagogy. It asks for an opening by a concrete problem, adds a short demonstration, moves the figures to the annex and places the exercise before the synthesis. It also requires a slide distinguishing what is demonstrated by the sources from what falls within a practical recommendation. The gain does not come from a prettier automatic design. It comes from the fact that the final presentation finally follows the mental path of the learner.
After the workshop, the trainer may propose additional resources in a LinkHub VIFLY structured: support, exercise, training and contact. The generated support then becomes a step in a path, not an isolated file.
Example 2: a consultant turns an audit into an explanatory video
A strategy consultant is responsible for an audit, data capture and recommendations. She requested a short video to present the three priorities to the leader. The production notes offer a general introduction, a succession of graphs and a vague conclusion.
When rereading, it replaces this structure with three blocks: the observed symptom, the probable cause and the recommended action. She asked that each statement be linked to a source of the notebook, remove an overly certain formula and add a methodological limit. Finally, it concludes with a precise decision to be taken this week. If the customer wishes to deepen, a link to VIFLY Booking can naturally extend the video by a dedicated work session, rather than by a series of private messages.
In both cases, the re-read plan serves as an editorial contract. It defines what the content must accomplish before the appearance takes full attention.
The viewpoint VIFLY : the best IA button is sometimes "not generate yet"
Progress is not that AI produces more. It makes human validation earlier, more visible and more difficult to forget.
Many tools promise to reduce the time between idea and rendering. For experts, this shortening can become a trap. A credible presentation, educational video or sales medium is not worth its volume of slides. It is worth the accuracy of its hierarchy: what the public must understand, believe and do.
The true value of this evolution is therefore a useful friction. Before clicking on "Proceed", the user must look at the plan and ask: Is the angle mine? Do the sources really support the assertions? Is the example suitable for the public? Does the conclusion lead to legitimate action? This control joins a broader rule: too smooth content can weaken confidence. VIFLY also explains in its analysis on too perfect content that no longer convinces.
The role of the creator or expert is not to monitor each pixel made by the tool. It is to protect intent, evidence and responsibility. If these three elements are low, the visual quality will not save the content. If these three elements are solid, the AI can speed up an already well-directed execution.
Nuances, limitations and points to bear in mind
The first limit concerns access. Google's blog announces a global deployment of co-creation capabilities, but the help pages indicate that the rereading of plan and production notes is reserved for Google AI Ultra subscribers at 20 October 2026. A free user or subscriber to another formula can therefore have the generation of slides or videos without seeing exactly the same validation path. Check the option on your account before promising this method to a customer or team.
The second limitation concerns reliability. Google warns that generated slides may contain factual or visual errors, and that videos may have inaccuracies or audio defects. The presence of sources does not automatically transform the result into verified content. Names, numbers, citations, dates, causalities and recommendations must be checked.
The third limit appears after validation. For videos, Google specifies that once the production notes are approved and the generation is launched, they can no longer be refined in this journey. For slides, revisions are still possible, but each generation creates a new version and sources are not currently taken into account during slide modifications. A late correction can therefore move away from the original material.
Finally, the motive may have differences or limitations. Presentations are indicated as being reserved for users over 18 years, and certain video features have their own conditions. These parameters change: they must be rechecked on the official help pages prior to contractual use, paid training or sensitive production.
What to do now? Seven-step protocol
- Choose a result, not just a format. Write what the public needs to understand or decide after support.
- Clean your sources. Remove duplicates, date documents and distinguish evidence from opinions.
- Define audience and duration. A beginner's deck and a video for decision makers do not follow the same progression.
- Ask for a plan first rendered. Even without a dedicated button, require a structure with objective, proof, example, limit and conclusion.
- Reread with five questions. Is the order logical? Is every fact tainted? Does the tone match your voice? Is a shade missing? Is the next step clear?
- Validate and then control the final file. Compare the rendering to the approved plan and check text, visuals, audio and quotes separately.
- Link content to a useful path. Guide, resource, offer, form or reservation: the suite must be relevant, visible and proportionate.
If you do not see the option of rereading yet, do not suspend your method. Ask the tool to produce only a text plan, correct it, and then run the generation in a second step. The editorial principle remains valid regardless of the button: one validates the thought before validating the appearance.
Frequently Asked Questions and Sources
Does Gemini Notebook replace NotebookLM?
Google now uses the Gemini Notebook brand in its ads and help center. Depending on the interfaces and accounts, some people can still recognize the old name NotebookLM. It is the same environment of source-based notebooks, enriched with creative functions.
Is the rereading of the plan available in France and in La Réunion?
The announcement of 19 October 2026 speaks of a global deployment during the week and does not publish an exclusion list. France and La Réunion are therefore within the announced geographical scope. However, real access depends on the progressive deployment and, for the detailed routes described in the help, on a Google AI Ultra subscription.
Can we ignore the validation stage?
Yes. Google indicates that the user can go through the Studio panel, do not enable the reread option and generate directly. For an issue-free draft, this choice can be reasonable. For customer, educational, commercial or public support, a plan validation remains preferable.
Are slides and videos guaranteed to be accurate because they use sources?
No. Help pages explicitly indicate risks of factual, visual or audio inaccuracies. Sources improve scoping and traceability, but responsibility for final verification remains human.
Should we subscribe to Google AI Ultra?
At 20 October 2026, official help pages indicate that co-creation paths with rereading slide plan notes and production notes for videos are reserved for Google AI Ultra subscribers. Basic generation functions may have a different availability.
Sources
- Google, "How to shape your slides and videos with Gemini Notebook", 19 October 2026.
- Gemini Notebook Help Center, generation and co-creation of Slide Decks, consulted on 20 October 2026.
- Gemini Notebook Help Center, Generation and Co-creation of Video Overviews, consulted on 20 October 2026.