0:00 Stop. Before you open chat GPT and ask it another question, watch this. Everybody is treating AI like a chatbot. Type a question, get an answer, copy, paste, repeat. And every single time you start from zero. No memory, no system, [music] no consistency. But there's a completely different way to use it. One that most people don't know about. Instead of a chatbot, you build a factory. A system where each module has one specific job, does it perfectly, and never forgets the rules you gave it, and the tool that makes this possible is completely free. So instead of handing you another list of prompts to copy paste, I'm going to show you how to build the factory itself. In this video, we'll build four modules inside Notebook
0:46 LM. A topic selector that scans AI news and finds your next winning video idea. A thumbnail and title creator that analyzes competitor channels and generates packaging for you. A script writer that learns from trending videos and produces a full script in your niche. And a visual prompt generator that reads tool documentation and builds technically precise image and video prompts. Stay with me through the last module. That's where it gets really interesting. Quick note before we start. This is part of the Notebook LM series on this channel. Each episode explores new use cases and real world examples. If you'd like me to continue this series, leave a comment below. That's honestly how I decide what to create next. Now, let's dive in. Case one, the
1:35 topic selector. Every creator hits the same wall. You sit down to plan your next video and your mind goes completely blank. You could spend an hour reading newsletters and still walk away with nothing useful. Or you let the factory handle it in 5 minutes. Head over to notebook LM. It's free. Click the plus icon and create a new notebook. Before we ask it anything, we need to feed it the right inputs. The quality of your output is directly tied to the quality of your sources. First, TLDDR AI daily AI news in short scannable format. Copy the link, paste it in. Next, the rundown AI tool launches and creator focused updates. Copy it, paste it. Third,
2:22 superhuman AI breaks down how AI is being used in real workflows. Copy it, paste it. Then there's an AI for that. A massive directory with thousands of categorized tools. Copy it, paste it in. As you can see, I've also added additional pages from the Product Hunt website. Now I click next, and within a few seconds, our knowledge base is loaded and live. Now, here's the step that changes everything. Go into the notebook settings to open the chat configuration panel. Click into the custom box and paste in the topic selector prompt, which you can find in the description. This transforms the AI from a simple chatbot into a creative strategist that thinks about what actually gets views. Hit save. Now, give
3:08 this notebook a name. I'm calling it topic selector. Let's put it into action. I type my request. Give me five potentially winning YouTube topics I should cover this week. For each one, explain why it has strong potential right now. Hit enter. Watch what comes back. It's not just listing headlines from the newsletters. It's analyzing which topics have momentum right now, identifying angles that haven't been fully covered yet, [music] and giving me a clear reason for each one. I scroll through all five suggestions. The fourth one immediately grabs me. Four AI tools automating the oneperson business. That's the topic we'll be using in our factory today. Case two, the thumbnail and title creator. Most people write their script first and figure out the
3:54 packaging last. That's backwards. Your title and thumbnail determine who clicks before anyone watches a second. They need to be built first based on data. Open notebook LM and create a fresh notebook. Now, we need to load it with the right inputs. Open YouTube and go to a competitor channel in your niche. Not a specific video, the channel homepage. Copy the URL from the address bar and paste it to notebook LM. Go to the next competitor. Copy it. Paste it. Do that for three channels in your exact niche. Then do the same for three more channels in different niches. Same kind of audience but different topics. Again, copy URLs one by one and paste them to notebook LM. This will help AI to learn
4:40 different winning patterns. After finishing pasting URLs, hit next. Let it load. What we've just built is a library from channels that are already proven to work and it took about 3 minutes. Now, let's configure the engine. Go into the notebook settings. Open the chat configuration panel and click into the custom box. Paste in the thumbnail and title creator prompt. It's in the description. [music] This tells the AI to think like a YouTube packaging strategist, analyzing the visual patterns, emotional triggers, and title structures from the channel pages we loaded. Hit save. Now give this notebook a name, thumbnail and title creator. Let's put this into action. I paste this prompt and enter the topic from the previous example to have three title and
5:26 thumbnail options. Then hit generate. Three options appear, each using a completely different psychological angle. Every option includes a title and matching thumbnail concepts with readyto-use image prompts. I review all three combinations and one instantly stands out for me. I copy the thumbnail prompt. Now, let's see it in the real world. [music] I go into Nano Banana, paste in the prompt exactly as Notebook LM wrote it, and hit generate. Look at this. This is a thumbnail I could put on a video right now. What would normally take a professional editor and few rounds of trial and error in a design tool, notebook LM handled the brief and Nano Banana handled the execution. You can do the same for any niche you like.
6:12 Two modules down, topic selected, title and thumbnail ready. Now we need to fill the middle, the actual script. This is where most AI generated content falls apart. Let's make sure ours doesn't. Case three, the script writer. The biggest problem with AI written scripts is that they all sound the same. Flat openings, generic transitions, no personality. The fix is simple. Stop giving it a blank page. Give it proven references. Open Notebook LM and create a new notebook. Remember those six YouTube channels we used in the previous step? We're going back to them, but this time we're not looking at their homepages. Go to the first channel. Click the popular tab. This shows you
6:58 the content that performed best on that channel, the videos that prove themselves over time. Rightclick the top video and copy the link. Go back to Notebook LM and paste it in. Go back to the second channel, sort by popular, copy the top video, paste it in. Do the same for all six channels, six most popular videos, six proven scripts, loaded in and ready to be analyzed. Now, let's configure it. Go into the notebook settings. Open the chat configuration panel and click into the custom box again. Paste in the script writer prompt. This tells the AI to study the hook structures, pacing, and retention techniques from the videos we just loaded and then apply those exact patterns to a completely original
7:44 script. It's not rewriting anyone's content. It's pattern recognition. Big difference. Hit save. Name this notebook script writer. Let's put it to work. I type write a 1,00word YouTube script for the following topic. Then I copy the topic from our first module and paste it here. Hit enter. Our full script is now ready. It explains four different AI tools in detail designed to keep solopreneurs engaged and watching exactly as we intended. Let me read you the hook it came back with. Running a oneperson business is the ultimate dream of freedom. But let's be honest, it often turns into a nightmare of endless administrative tasks. You are the CEO, the marketing department, the web
8:29 developer, and the accountant. You wear a dozen different hats before lunch. But what if you could hire an entire team of specialized assistants for a fraction of the cost? Hear that? It opens with tension. The transitions flow smoothly, and the pacing feels natural. I'm instantly hooked, which is exactly what we want. Now, [music] let's build our visual prompt generator. You have your script. Now, you need visuals. And if you've ever typed what you wanted into an image or video AI tool and got something completely different back, that's a prompt problem. These models are extremely literal. If you don't know those rules, you're guessing every time. We're building a module that knows them for you. Open notebook LM and create a
9:15 new notebook. Now we load the manuals. First source, the nano banana prompt guide. Copy the link and paste it in. Second source, the VEO guide. Copy the link, paste it in. You can use any source you want. Let it load. These two guides are now our AI's technical documentation. It knows exactly what each tool supports, what syntax to use, and how to structure a prompt for maximum accuracy. [music] Rule book loaded. Let's configure the engine. Go into the notebook settings, open chat configuration, and paste in the visual prompt generator prompt. This tells the AI to operate as a senior prompt engineer to reason from the actual documentation and produce prompts that
10:01 are technically precise. Hit save. Name this notebook visual prompt generator. The engine is ready. Let's test it. Let's start with an image. I type generate a text to image prompt for a cinematic overhead night shot of a vintage wooden desk with an open notebook dim lamp and coffee cup. The notebook reads, "Subscribe to the AI garage for the best AI workflows." It comes back with a full structured prompt and a short explanation of every parameter chosen referencing the guide directly. [music] I copy the prompt. Go into Nano Banana. Paste it in. Generate. Look at this. The warm lamp glow, the
10:47 texture of the paper, the shadows between the objects. It feels like a real photograph. All of that came from a prompt built on documented rules, not from guessing and regenerating 15 times. Now, let's test video. I type, "Create a medium close-up text to video prompt of a creator seated at a desk in a modern home studio speaking directly to camera. I also ask for a short speech that pushes it to the limits. The AI builds the full prompt from the VEO guide. I copy it without changing a single word and paste it into Gemini. After selecting the video option powered by VEO, hit generate. Let's watch. >> Did you know that AI Garage brings you the latest AI tools and systems every
11:34 week? >> Clean [music] lip sync. Natural delivery. The way they talk about the AI garage feels completely authentic. The studio environment looks real. That's what happens when the prompt is built from the tools actual documentation. So that's your personal AI prompt factory. Four modules, all free, all running on the same tool. All four custom prompts are in the description and the pinned comment. Copy them, paste them into your notebooks, and start using AI like a factory. If you want me to continue this series with more modules, drop a comment. I read every single
