Google Opal is Too Good to Be Free & More AI Use Cases

Welcome to another week in generative AI. This week is mostly about unique new features that have been added to various top applications that allow you to do things that well partially weren't doable before or now are just a way more efficient way of doing something. So, as you'll see, we'll be talking about notebook alm adding video overviews where it turns all of your documents into a PowerPoint presentation with a voice over. Interesting. And then we have Photoshop integrating a new feature that saves designers the work on probably the most common interaction that most people have with Photoshop. I was really interested in that one. We'll also touch on the GP5 release which everybody's anticipating but it's not here yet and so much more in this week's episode of AI news you can use. And yeah, if you watch this show weekly. Yeah, I suppose I have a new haircut. It's summer. But now it's time to talk AI. Let's begin. And also important to mention today's AI news you can use video is sponsored by Framer. But more on them later in the video. And our

journey starts in what I think is the most significant release of this week. I really did not expect this one. This is called Opal by Google. And it's essentially like one of the many AI app builders that you're probably very familiar with at this point. I mean, we covered most of the big tools on this channel. Some of them sponsored, some of them just covered in these shows and tutorials. And the deal with these is usually relatively similar. They give you a simpler interface to build applications with the power of AI. So now Google also entered this very crowded space but they did it in a unconventional way and both me and the team actually really like how this came out. It's called Opal. It's an experimental AI app that you can use for free. Me sitting in Lisbon I had to use a VPN to tell the browser that I'm in the US to use this. But look at it. The way it works is you can pick one of these preset apps here. So we're just going to go with the business profiler for this little demo here. And then it already has the entire workflow set up and you can just use it here on the right. So it's kind of a different version of Claw's artifacts, but it gives you this dynamic automation type preview where you can actually edit it. So it's not just a standalone application. It's actually a sequence of different actions and the application over here. So in summary, I think this falls somewhere between the no code automation tools and claifacts that I personally am a big fan of because it's probably the easiest way to build an app in the entire AI space. Okay, so let's just have a look at what we have here with the business profiler. So there's a first step which asks for input URL of a business you want to analyze. Then there's a second step which researches it with Gemini Flash and a custom prompt and some tools here. And then there's a third step that summarizes the results. Really simple. But the beauty of this is one you can extend it and two you can just try it here right away. You can just say start. I'm going to put in our website URL and see what it comes up with. And there it is. It took some info from our homepage. LinkedIn. It found some discussion on Reddit that it summarized here. And there you go. So this thing in itself might not be the most useful thing in the world. But I think with this editor here, that's where it gets interesting because I can remix this app and then add extra steps. So this is me doing that for the first time, but as you can see, you could also add more assets just like you use from some of the no code builders. But again, I think the unique thing here is that you have the app here right away, which I suppose does make it comparable to something like Lovable, but you don't really have workflows there, right? It just builds a full application for you and you get the results, you get to link it to other things, but you don't get to control this flow. And especially when it comes to customizing the prompts and using AI in the middle, you really want control over the flow. And yeah, to me, this is just the first time I've seen this mix of very high user friendliness and customizable workflow. And that's why it caught my attention and I wanted to show it to you because I feel like just because right now we have this avalanche of different AI coding tools that build applications for you. I think that is happening because coding is the first big use case that it unlocks next to information retrieval, research, and stuff. We'll actually talk about our research paper later in this episode showing you what the use cases within Microsoft's Copilot are within companies. But yeah, I think so many apps these days do AI empowered development because that just works really well. But next up, we'll have all of these other niches unlocked like the sciences, writing, marketing, sales because all of those they sort of work, but we don't really have these type of applications purposely built for them cuz they just don't work well enough. And I think the interfaces for that are not going to be a command line like clot code because just devs are familiar with this interface. But I think it's going to look more like this. And that's why I wanted to show you. Again, if you want to try this, you can take all of these and try building your own writing app based on the preset they already have. And maybe one last comment that I want to add here is that yeah, if you have been learning the basics of automation or you're doing that right now, I think that's the right path to take right now after understanding the basics of prompting because both prompting and automation skills will carry forward into the next phase of all of these Genai apps where it might look something like this. Yet, you still have to customize the prompts and understand automation flows, right? Okay, let's see what's next. Since you're watching this

video on YouTube, we likely met on YouTube. And you probably think of me primarily as a YouTuber, and that's fair. I make YouTube videos regularly, but I really only consider it one part of my overall career as an entrepreneur, builder, founder, whatever you want to call it. And one thing that I learned over the years that as an entrepreneur, you're only as good as one, your work ethic, two, your systems, and three, the tools you use. So, if you're on a similar journey to me or you have ambitions to start your entrepreneurial journey, I've got something to show you today. And that's a tool that could be valuable on your journey. And that is Framer, the sponsor of today's video. So Framer is a no code website builder that feels like a top tier design platform like Figma, except it creates these beautiful websites that take very little effort on your part. And I think that's what really sets them apart. So when you're creating something new, you usually need a publicfacing website to represent that idea to the world. And you need to bring it to life quickly. That's exactly what you can do here. In fact, companies like Perplexity, Superhum, and other YCbacked startups have been using Framer recently. So, if you want to get on their level, this is the kind of tool you need. But instead of talking about it, let me just show you how easy it is to set up a website for something like a professional agency. In this example, using Framer. So, as you sign up, you quickly arrive in this interface. And here on the right side, we just start by inserting something. How about the AI tool? And I'll just say Eagor's AI consulting. Let's create a website for that. And immediately, you can see this come to life just like you would in Figma. Maybe I can add a little FAQ section in this chat interface here. And yeah, there you go. Couldn't be any simpler. Now, there's a ton of things I could do. I could watch tutorials, try various templates, browse all of their plugins. And as you can already see, this is really heavily design focused. So, if you have a very refined taste, I believe you'll find what you're looking for here. Anyway, returning back to this builder interface, I want to show you one more thing, which is that this has a built-in content management system. So, if you, for example, want to add a blog to your website, it's as easy as going over to collections, clicking add blog, or importing an existing one, and there you go. This is the interface for your entire blog that you can start building out inside of the built-in content management system. And look how beautiful that is. And there you go. A few more images and I have my consulting website up and running and I can focus on what I do best, which is the business component. So, if you're looking to build beautiful professional websites, you should totally check out Framer today. Go to the link at the top of the description and use the code AI advantage to get 25% off your first free months using Framer. All right, now let's look at the next piece of AI news that you can use. Next up, we have a

very interesting update to an application that I know many of you use. Matter of fact, some viewers of the show that I talked to tell me that Notebook LM is their favorite app outside of Chat GBT. That's also the sentiment of the community. So, if you're still sleeping on it, hey, if you have a lot of context, Notebook LM and there's a new feature. It's called video overview, which essentially turns your notebook into a PowerPoint presentation. This sounds like an excellent feature for visual learners like myself. And hey, if you're watching a video, hm, chances are you might be a visual learners. So, let's have a look at this. I just open up Notebook LM here. And this is a notebook I have with a bunch of resources on public speaking. And as you can see, this button on the right is new. That's what they added, video overview. And if I click this, in a few minutes time there should be a video with different slides and a voice over summarizing the content of your notebook. Let's see how that looks once it's done. Okay, so this actually took around 10 minutes. Let's put on the headphones and see what this looks and sounds like. Okay, so just first up, this slide looks beautiful. No, — if that sounds familiar, you are definitely not alone. So today we're going to break down exactly why that happens and more importantly give you the tools to turn that fear into pure confidence. All right, let's get into it. So, if I asked you this. — Oh, that was really well synced. She finished up the intro and then it moved to the second slide. — What would you guess? Spiders. — It's interactive, too. She poses questions. Okay. — Yeah. — Look at these slides. This is just beautiful. Honestly, seeing this, it just makes me realize that this might be a new workflow to generate slides. Give it all your content and then it comes up with a PowerPoint presentation and even does the script writing and speaking for you. Hm. Very interesting new feature. And I'm not exactly sure what to make of this, but one, it certainly seems like a fantastic feature to learn with. I mean, I myself prefer the video format over everything else when it comes to learning new topics. Matter of fact, one fun piece of lore about me is that I started collecting all of these software skills when I was 16 years old. And I remember back in high school, I pirated a Photoshop course and a copy of Photoshop back in the day. That's 15 years ago. I suppose I can admit that now. And as soon as I took this course, I started applying my skills to Photoshop funny pictures of some classmates. And us being one of these laptop classrooms where every student had a computer. There was three other guys in my class that asked for the software and the course. They took it to installed the tool and all of a sudden we were creating pictures of each other. The point being that was the very first software I learned and then I just proceeded to keep learning video editing, visual effects, live streaming, coding, lighting, audio and more. But the point is it all started with video courses and all my consecutive skills also I learned through video courses. And now that AI is getting to this point where it can sort of create one video based on materials you give it to teach you something. I mean I suppose the next step of this would be creating a curriculum and a whole course for you on demand. Can't be too far away if it can do a video well like this. And let me tell you, I'm here for it. Back when I was 16, I love to learn from video instructions. That's what worked best for me. And I don't think that changed or will change. So, honestly, the more I think about this, the more powerful this seems to me. As a visual learner, this is preferable over just audio to me 100%. And feel free to use the comments below to let me know what you make of this feature. All right, let's see what's next. Okay, so next up, we have

Adobe shipping a absolutely massive feature. It's called Photoshop Harmonize, and it's available in the Photoshop beta if you have a Creative Cloud subscription, which I happen to have. As I mentioned, Photoshop was the very first software that I learned back in the day when I was 16. And let me tell you, even back in school, what harmonize does is one of the main things we were doing as kids, and graphic designers do this all the time. It's when you composite two images together, well, they don't naturally fit together because the lighting and the environment is different. Now, what Photoshop did here is ship a feature where you can just press a button for the very first time and it should do this by itself. So, let's just click okay and see how well this does on the example. No way. Okay. So, like if I had this back in the day, this would have saved me hundreds. Yeah, legitimately hundreds of hours of work because I was compositing stuff together all the time. Like, if you've never done this before, this is quite the process. Like, you have to add a color correction layer, then make sure this matches up with the environment. And then what you would do is kind of go in and use these different brushes to apply the same shadows. So, here the shadows are on the left. So, so I would want to bring shadows here and over here, etc. It's a really manual process. And as you can see, we can skip all that by just pressing a button. It even adjusted the texture so that it matches with this. It's kind of an extra credits type of task. So, let me try that on a random image I find here on the internet. Kind of just drop that in here. And then also, let me bring in this bolt. The smart features can already help me create a layer mask for this. Like so. Gosh. And then when I readjust the size and kind of place the bolt, yeah, I think it should be more like this. Well, now I should be able to click harmonize. And let's see how well it adjusts this. No way. Did the shadow too. Look at that. Okay, this is really impressive. They did a great job on this. Before, after. Wow. Impressive. I really do my best not to throw around this word lightly anymore, but this really is a game changer. That's 30 minutes of work right there with one click every time you bring in a new asset. Let's see what's

next. In other news, there's a minor but very interesting chat GPT update. Matter of fact, they did something they've actually never done before, and that is implement a feature for a specific target audience. It's students and people learning things. Okay, so if you open up the tools, all paid plans at this point should have this study and learn feature. And it starts behaving differently when you turn this on. Now, in essence, this feature could be achieved through a few custom prompts. But let me just show you. If I tell it something like give me information about the basics of Python, you can see it follows up with some extra questions here. And then in the end it suggests a quick practice. So if I tell it let's do the practice it will hit me with some questions here. And if I ask what's the answer, this is the most unusual part of its behavior. As you know chatb will usually tell you what the answer is. But here it tells you nice you're eager but let's slow down a tiny bit. So you learn the why not just the what. And then it restates the question not giving me the answer. So if you want to learn with chatt rather than just getting all the answers, yeah, the study feature is for you. And essentially it integrates a bunch of prompts that previously have been done manually into a feature. So when you add an integer and a float, you get a float. And then it comes up with further questions. So nothing groundbreaking, but I actually do think this is going to be very useful whenever you're trying to grapple with a brand

new topic. Okay, so I want to talk about GBD5 for a second. It's not out yet. We were all expecting it every single week of July because they said multiple times that it would come in July. Now, we're already in August and it's still not here. So, I want to just quickly round up what we know so we're all up to date. So, the most recent mention on Twitter was Sam Ultman talking about GPT5 coming soon on July 19th. And before that, he mentioned twice that it would come by the end of July, which obviously didn't happen. And then in this recent the vaugh Sam Alultman podcast which I would actually highly recommend. This was one of the best conversations with Sam because Theo just kept it very real and Sam was talking about of the different capabilities of GPT5 and it really seemed like a pre-launch teaser. He didn't give any specific details on what it will do. But I think the main kind of like teaser was that he tried it on questions that he didn't expect AI to perform well and it completely nailed it. I was testing our new model and I got a question. I got emailed a question that I didn't quite understand and I put it in the model, this GPT5, and it answered it perfectly. — Personally, I'm getting kind of tired of waiting because I think the first timeline they communicated was that it's going to come in early 2024 and right now we're just kind of stuck waiting. If you want my take on what I think it will be, I think it's going to be a mix of the different applications, modalities they have in here. So, I think it's going to be a new model that natively integrates everything from web search to canvas image creation, video generation, agent mode, video editing, and basically it does the routing between the different models and applications for you. Whereas now, you have to go in and pick what you want to use at any given time. And if it does that well, it doesn't even need to be smarter than the current models. It just has to use the tools well. And I think that's what we're going to see. But one thing is for sure, once it comes out, you'll see extensive coverage on GPD5 on this channel. use cases, quick start tutorials, and more. So, yeah, me and the team are very much looking forward to it, but as of now, we're stuck waiting. Okay, next up, we have the

quick hit segment with all the new stories that are worth mentioning, but maybe not worth spending multiple minutes on. Starting with a story that we kind of hear every week. There's another Chinese LLM that is doing really well. It's called GLM 4. 5. As I understand it, the main selling point here is that it's cheaper than DeepSeek while performing better. They also ship the light version of it. looks good on the benchmarks, especially considering how small the model is making it cheap tool. So yeah, if you're building something and looking for models, this is another one to consider. Let's see

what's next. Now be this very interesting paper and these are my personal favorites because everything we do at the AI advantage is related to use cases and upskilling individuals on how to use these tools well for themselves. So whenever I see a paper on how AI is actually implemented and work, I pay extra close attention. And this one in particular is new. It doesn't have anything revolutionary in it, but I wanted to share with you that this data they collected on how companies use co-pilots paints a clear picture. If you look at the top three categories of the activity that the employees had with Copilot, you can see all of it is related to either gathering information or obtaining info or responding to customer problems. Then this one is knowledge related too. This one is customer support related too. customer support and research and information gathering are the top use cases for people who use copilot. And I think this does map. There's a lot of esoteric use cases and amazing things that you can do with these LLMs. But the bulk of it is just getting the right information and manipulating it in ways that is useful to you. And then there's a bunch of ways to that more effectively to get more accurate information like providing context documents, prompting more accurately, equipping it with tools that make sense. But for that some training is required. So, I thought this was interesting and relevant to the show. If you want more details, it's linked below as per usual.

Then we have some quick updates from various Google apps. I think one of them is a complete miss and actually kind of interesting and it's essentially the same feature, turning a image into video. So, this first one essentially is being integrated into Google Photos and you can turn your images of memories into small videos. I personally think this is a complete miss. Like, who would want to turn their old real pictures memories into something fake? I mean, look at this guy. It's a beautiful picture of him hiking and then when you turn it into a video, all of a sudden he's throwing up confetti. Like, what's the point of this? That's just deeply unhuman and I'm not a fan of that. But on the other hand, they implemented the same features in shorts creation. And I think here it actually makes sense cuz you can do cool stuff like, yeah, turn this video into a video of himself sinking into the floor doing visual effects that otherwise wouldn't have been possible. That might be interesting in the context of the short. So, it's not that the technology itself is bad, it's really how it's used. And I don't know, in my humble opinion, applying it to past memories is not a good way, but when you're getting creative with shorts might be a

fantastic use case. One more thing from Google, and this is super quick, and it's just them experimenting with the search interface in public and essentially testing what people actually like. So, as you might know, there's Google search, and then there's the AI search engines, and then there's Google integration of AI search, which is sort of like a separate tab and a separate section at the top. Well, now they're testing a new way which kind of integrates the answers with links and then a paragraph of answers and more links. And I just wanted to highlight this, not because I think that this is the future of search, but because we're witnessing the transformation of what information retrieval on the internet looks like in 2025. I'm really curious to see where this is eventually going to end up. But one thing is for sure, and that's Google is not going to give up its position, and they're going to experiment with every possible modality to maintain their monopoly position for as long as possible. And that's really everything we have for this week. I hope you found something useful here or at the very least feel a bit more inspired to go out there and do more with AI or alternatively just create some space to do more of the stuff that you really enjoy. And with that being said, my name is Igor and I hope you have a wonderful