End-to-End AI Video Creation: NotebookLM to Gemini Canvas to Google Vids

---

We have spent a lot of time exploring how individual tools can transform the way we approach teaching with AI. From exploring the NotebookLM 2026 update for knowledge databases to learning how Google Vids turns slides into AI teaching videos, the possibilities are massive.

But what happens when you combine them?

Today, we are moving beyond AI prototypes and building a robust, production-ready system. I am going to walk you through an end-to-end AI video-creation workflow that connects NotebookLM, Gemini Canvas, and Google Vids to generate a complete, professional video lecture featuring AI avatars and voiceovers.

(Note: To follow along and create videos like this, you will need a paid Google AI Pro plan, Gemini for Students, or a Google Workspace for Education account.)

https://www.youtube.com/watch?v=RDKVcaE52hg

End-to-End AI Video Creation: NotebookLM to Gemini Canvas to Google Vids

Step 1: Ground Content for AI Video Creation in NotebookLM

The biggest advantage of starting with NotebookLM is that it acts as your dedicated knowledge database. This ensures that all content generated later in the pipeline is consistent, accurate, and properly sourced.

1. Open NotebookLM and create a new notebook.

2. Use the Fast or Deep Research feature to pull in relevant, up-to-date information. In our video example, we searched for recent AI models on MongoDB (such as Voyage 4 and Native Automated Embedding).

3. NotebookLM will scan hundreds of web pages and pull in the best resources, such as reports and YouTube videos. Select the ones you want and import them.

Now, you have a solid foundation of facts.

NotebookLM interface showing the imported MongoDB resources on the left panel and the Fast Search results.

Step 2: Outline Your AI Video Creation in Gemini Canvas

Instead of starting from a blank page, we can seamlessly push our NotebookLM data directly into Gemini to start building our presentation. If you've ever built no-code AI apps using Google Opal, you know how powerful chaining AI agents together can be. Here, we are doing it manually.

In Gemini, import your new NotebookLM session into the chat. Then, open Canvas mode. (Note: If you want a deep dive into building interactive AI dashboards with Gemini Canvas or other data visualizations, you can generate and import those here, too!)

 Gemini Chat interface showing the NotebookLM session successfully attached to the prompt box

Troubleshooting Tip: If you ask Gemini to create a fully formatted Google Slides presentation with text, images, and charts all at once, you might encounter an error. To bypass this, ask Canvas to generate a slide outline first.

Gemini Canvas mode opens side-by-side with the chat, displaying the generated text outline for the slides

Once the outline is generated perfectly, save it to a Google Doc. Then start a new chat, import the Google Doc outline, and ask the Pro model to generate a Google Slides deck based on that structure.

The final, beautifully formatted presentation was exported to Google Slides, complete with text, charts, and images.

Step 3: Automate Speaker Notes in Google Slides

Once you export the generated file to Google Slides, you can refine the design, change the presenter's name, or alter colors.

To save hours of scripting, open the Gemini side panel directly in Google Slides and prompt it to: "View the slides and provide detailed speaker notes for all slides."

Simply copy and paste the generated text into the speaker notes section of each slide. Your script is now ready for production!

Google Slides interface with the Gemini side panel open, highlighting the generated speaker notes ready to be copied

Step 4: Execute AI Video Creation in Google Vids

Head to Google Vids, choose Convert Slides, and import your newly created presentation. Be sure to uncheck the automatic voiceover and background generation, so you have full manual control.

AI Avatars vs. AI Voiceovers

When adding your script, you have two main options, and you must manage character limitations:

• AI Avatars: These provide a great visual of a presenter with synchronized lip movement, but they have a strict 500-character limit per scene.

• Voiceovers: If your script is longer, use the Voiceover tool, which allows up to 2,500 characters.

Pro-Tip: You can mix and match these! If you use the "Educator" voice for a voiceover scene, you can select an Avatar in the next scene that uses that exact same "Educator" voice profile, keeping the audio perfectly consistent throughout the lecture.

 Google Vids interface showing the menu for selecting a profile for an Avatar.

Step 5: Animate Images and Fix Audio Overlap

To keep the audience engaged, convert static images on your slides into 8-second AI videos. Select an image, click Convert to Video, choose "From Scratch," and paste the relevant sentence from your speaker notes as the prompt.

When you insert these new video elements, you might run into two quick issues:

1. Overlapping Audio: Both your background video and your main voiceover might have audio. Simply select the background video element, click the sound icon, and mute it.

2. Length Mismatches: If your voiceover is 15 seconds but your generated video is only 8 seconds, click the video's Playback Options and set it to Loop. This ensures the visual plays seamlessly while the narrator finishes speaking.

The Google Vids timeline highlighting the loop function enabled

Ready to Publish

Once you have adjusted your timelines and added any slide transition animations, your video is complete! You can share it directly to OneDrive, download the MP4, and upload it to your students.

This workflow takes you from raw data to an engaging, polished lecture while saving hours of traditional production time. Be sure to check out the rest of the resources on the tutorial and course main website to scale your own educational content!