Master NEW NotebookLM + Gemini Workflow in 14 Minutes (2026)

Most people think the only way to automate their workflows is to spend weeks learning the latest AI automation software, but there's actually a faster and easier way. By the end of this video, you'll know how to use Notebook LM to turn any complicated process into a step-by-step workflow. But make sure you stay until the end because once this system is installed, I'm going to show you just how far you can actually push it. One of the things that really makes Notebook LM excel with a Gentic Workflows is the fact that it's a rag system. It retrieves information and so it simply retrieves and follows instructions. When you combine that with Gemini and its probabilistic nature, you have the perfect storm. The combination between Gemini and Notebook LM is like a train on tracks. Gemini is obviously the train. It's powerful. It can do a lot of heavy lifting, but without the tracks, you don't necessarily know where you're going to end up. And when you have a workflow, when you have something that you're doing and you know where you want to go, it's best if you build your own tracks. We've known for a very long time that chat bots like chat GPT and Claude and Gemini it excel when you give them one task at a time. And that's why it's so important that we don't try to create one single prompt to handle everything. We want to take our workflow and we want to break it down into the smallest details possible. And then we want to create mega prompts. And these mega prompts are nothing more than several small prompts linked together in logical or sequential order. And so naturally when you begin to take things like research, strategy, writing, design, review, assembly, and even more, you're going to have several small prompts that need to stay in the right place and in the right order. And the best way to do that I found consistently is by using what I call a router prompt. This prompt is going to tell the AI when to enter each of these individual states and also when to exit, when to return, etc. if need be, depending on your particular workflow. So you might have a prompt, a mega prompt that is focused purely on writing, another one that's focused purely on design, another one focused purely on research review or research or strategy, whatever the case may be. And once you build these, you can assemble them with a router prompt. And this is something that I've been doing well before Claude introduced Claude skills. if you go back and check the release date of that feature in my videos and I'm pretty certain that other people have been doing this as well because it's my firm belief that the companies are watching how people use the platforms and they're taking the most popularly used cases and just converting them into features. And I used to love chat GPT projects and claw projects for this and I always wished that Gemini had projects or that there was this Gemini and notebook LM connection. And now that we have it, I'm switching all of my projects over from ChatGpt and Claw because they just don't compare with Notebook LM and Google Gemini for the same integration and the power and the quality of the work and the outputs that I'm getting. And to be 100% transparent, since Gemini 3 released a few weeks ago and this Notebook LM Gemini integration released not too long ago as well, my channel has been growing three times faster because the workflow has given me the space to be so much more creative and to flesh out my ideas without the constraints that I was running into with the other platforms. And that's not to say that those platforms aren't good. They still serve their purpose. But this is where I'm seeing the results and going to be for 2026 with all of my projects and my workflows. And that's what I'm doing in this video. So, when you set up an individual notebook for a single workflow, you just place the Gemini Train on a particular set of tracks that is going to take you to your desired destination. And so, let's just say I have this brain dump and I say whenever I get a sponsored offer email, I have this whole mental process. First, I have to check if they're legit. Like, if it comes from a generic@gmail. com or atyahoo. com address, I usually just ignore it because it's spam. Also, if it's about crypto or gambling, I autoreject it. But, if it looks like a real company that fits my tech niche, I have to figure out the price. If they want a full dedicated video, I charge $5,000. But if they want a 60-second integration, I charge $1,500. Then I have to actually write the email. If it's a no, I need a polite rejection script. If it's a yes, I need to reply with the sitement and quote the price I just calculated. Turn this into a step-by-step process. And so now Gemini has taken everything that I shared with it and turned it into a step-by-step

process. We have phase one, phase 2, three, and four. We also have starts and stops so that it can function like a state machine. And it's going to increase the probability that it follows the directions the way that it's supposed to. And so I might look at this and I might say for phase one verification that instead of just looking at the subject line and email address, I might also wanted to do deep research as far as the company, who owns the company, and how much money they're generating annually. I want to do a deep dive on phase one verification. And instead of just looking at the email and other things that we've listed, I also want to do research about the company, who owns the company, who's the CEO, how much money they're generating annually, and what their reviews are online from Trust Pilot, if applicable. And I'll do this for each of the prompts to make certain that my entire workflow is captured in the right section. Now, what I want to do is tell Gemini, please expand phase one into a mega prompt that gives the AI strict instructions and rules to follow so that it can automate this entire process for me step by step. Make sure that you include checkpoints for approval by the operator and then we'll follow up. Write separate mega prompts for the remaining phases of this process as well. Now we have mega prompt two, three, four, and we're ready for the final step. I need you to write a router prompt that controls how an AI system operates, manages, and executes each of these prompts. It needs to execute each of the prompts in the right order of operations. And they are numerically accurate right now. So go ahead and write a simple prompt for me that can get this done. And I want you to notice that I didn't use any technical language. I didn't over complicate it. It's just a very simple prompt telling the AI exactly what I want. And now it's going to create it because it understands. Now, because we're in Gemini, what we have to do is copy each of these, open Google Docs, and paste it, and then download it as a markdown file. Now, it's important that we download the router prompt as well as the other system prompts that we need, but we also want to make sure that we copy this. Now, we want to come over to Notebook LM and create a new notebook. And then, right here, we're going to choose files. We're going to drop in our four different files that contain our process. We're going to configure notebook, choose custom, and then we're going to paste in our router prompt, and click save. And now that you've set that up inside of notebook LM, the next thing you want to do is come back to Gemini, add that particular notebook, and then just ask Gemini something very plain. Search through my email for any brand sponsorship opportunities that I may have, and then verify them and help me see which ones I should work with, which ones I shouldn't, etc. And because this is a Google AI account, it's not going to connect to my business account. And the Notebook LM feature hasn't rolled out for workspace accounts yet. And of course, it's going to keep searching, but it's not going to find anything. And this is the beautiful thing about using Notebook LM as an agent because it's a rag system. If it doesn't know something, it doesn't know. It is not going to fabricate it or hallucinate. And this is exactly where Chad GPT and Claw probably would have spun something up. But instead, Gemini said, "I've searched your recent emails for keywords like sponsorship, collaboration, but I haven't found a direct inbound sponsorship offer from a brand yet. Most of the results are newsletters, the information, Robin Hood snacks, or personal notifications, school updates, receipts to help me find the specific opportunity, blah blah blah. " So, we know it's working, but I don't have any sponsorships here. But, let's start a new chat. And then, let's click the plus button again. Choose Notebook LM. And this time I'm going to choose YouTube made simple 2. 0. I have an idea for a YouTube video about using notebook LM for automation and agentic workflows very similar to what I've done with chat GPT projects and claw projects in the past. Can you help me build this video out? Now what's important about this is not the actual workflow for how I create my YouTube videos. What's important is this. When I show you the thinking, if you come to this last paragraph right here, you can see that it's analyzing agent instructions. I'm now carefully analyzing the A1 agent instructions to ensure a smooth transition into its role. I'm focusing on crafting an effective response template that acknowledges the YouTube made simple system. The goal is to start with five questions tailored to the user specific context highlighting the shift from content synthesis to automation using their experience but it's focusing on the A1 agent. I want you to notice that this phrase right here. So now if we

come over to notebook LM we open up the YouTube made simple notebook and this is the full A1 agent that it's referencing right here. I can not only run this agent inside of Google Gemini by adding it as a source, just clicking down here and adding a notebook, but if you want to, you can absolutely have this chat right here inside of Notebook LM. So, I'm not simply locked in to using this inside Gemini because I can even automate this entire workflow inside of Notebook LM. And now that I have everything built out right here with the metadata, the idea brief, the series, the title packet, the hooks, thumbnail concept, even the script body, the intro, the promo assertion, recording notes, editing notes. I can now come to the right in the AI studio and start to create different creative assets. I can create a video overview. And that video overview will allow me to listen to this video before I post it to YouTube to hear how it sounds in my own ear. I can also open the slide deck and I can create slides that I can use for B-roll in this video. If I want, I can create a mind map so that I can actually understand the content better to make sure I'm explaining it in a way that people understand it. And as you can see, it's a very simple system. And if we open up each of these, there's the role, the angle arc types, and the output brief. We can expand the role, the psychological engineering, the transformation promise. We can open the angle archetypes, the fortune teller, the experimentter, teacher, investigator. And so it's taking the prompts and it's breaking them down. And so this is just a great way to look at it if I was explaining this particular system, which I'm not, but I'm just saying it's a great way to actually understand what you've created and what you're automating as far as your work goes. And look at this YouTube made simple slide that it created for me. This is the perfect representation of the prompt like the idea engineering, the series architecture, title, hook, thumbnail, script, intro, promo. I absolutely love this. The golden rule of packaging, the tension trinity. This is pretty good. It even created a graph right here like for YouTube and how you want to keep people watching for the first 30 seconds. This absolutely covers everything and it gives us a great visual representation of the final production packet. I couldn't be happier with this slide deck that Notebook LM created and it did it based off of this system. So, if I ever wanted to teach people how to use this system, I have a slide deck that perfectly and clearly walks people through it. And then I have a content creators factory. I have a video overview. And so, I could actually listen to this video and I could say, man, you know what? That sounds like a pretty good idea, but I'm going to tweak it just a little bit. And so, all I would have to do is download this video, come back to Gemini. I'm going to click to add a notebook. I'm going to add my YouTube made simple notebook. And then I'm going to add a file or I could just drag this video in just like so. And now I'm going to ask Gemini using the YouTube made simple workflow optimize this video that I've uploaded so that I can record this tomorrow. And now I'm using my YouTube made simple notebook LM automation agent to process the actual video that it created with the power of Gemini. So now we have the agent, we have the asset, and we also have the intelligence of Gemini 3 Pro all in a single conversation to create an optimized video. So, if you got value out of this video or you learned something new, hit the like button, hype the video if you have the opportunity, and as always, take care, have a great day, and I'll see you in the next