Quickly Turn Prompts Into Products with Claude & More AI Use Cases

This week in generative AI has been very interesting because a lot of the apps that you know, love, and probably use already are being transformed by the big players or some new startups into products that are more niche down. So instead of just Gemini, you now have Gemini for education with a whole new interface. Instead of just clawed, you have clawed artifacts with specific presets that you can build on. Instead of just image generator, you have artistic image generator with a specific flavor to it. We're going to cover all of that and more in this week's episode of AI News. you can use where we pull together all the generative AI releases of this week and filter out the ones that you can use in that matter. Okay, I

want to start with a story that we featured last week in the quick hits. But as I used it over the past week, I feel like it deserves 10 times the attention that we gave it last week. And that's why I'm going to do this first segment of this video about the improved claw artifacts. If you have not seen this, pay attention here for a second because this is legitimately the simplest way to build AI powered applications that I have seen yet. There's many builders out there these days and a lot of them have their unique strengths, but nothing is as simple for beginners as what Enthropic did here because you can legitimately just come in on the free account, tab over to the artifacts here and you have an entire gallery of many pre-built applications for inspiration and then you can just click one of these and customize it. And as I mentioned last week, these do use AI in the application without linking an API, without having to worry about hosting. You just click a button, they're online, you can share them with your grandma. It's really that simple and I want to highlight it because it was simple already with some of these other apps, but with the native AI integration from Claude, it just has never been easier to customize a app for yourself. And if you haven't done it yet, this is really the time to try it. So check it out. I'm going to give you one quick example here in the entire gallery. There's many amazing ones like this office simulator, which is a little 3D game where you can actually customize the personas and stuff. I thought this one in particular was just like really creative and opened up a lot of fun possibilities. But the one that I want to spotlight here today is this language learning tutor because you can take this preset application that is a personalized AI tutor and you can customize it to whatever you're learning. So check this out. I'm just going to hit customize over here and you can see the application right here and I could start interacting with it, testing it. That's great. But then here's the thing. I can say something like rebuilt this tutor to help me learn the basics of generative AI with a specific focus on learning how to manage prompt context within a close manufacturing company. And what this is going to do, it's going to take the entire template that is built out already with learning goals and the chat function and it's going to customize it to what I specified here. Now, you could make this prompt way more specific. Tell it really what your company does and what the challenges are and it's going to customize it to exactly that. But one thing that I really learned from doing a lot of corporate webinars recently is that managing the prompt context, learning what that means, having some presets is really the biggest productivity unlock when it comes to using tools like chat or claude. And this way I can just customize the tutor to help me with learning that. And imagine if you're an educator and you have a classroom and before you even start teaching, you let your students create their own tutor that they'll be using. You can provide the prompts and everything, they can just create a free clot account. And as I'm explaining this, you can see the app being created here. And in a second here, we should have our tutor. One minute later. And voila, there we have it. Our reworked tutor in one prompt, right? So let's see what do we have. It created different presets for various parts of our company, right? That just makes a lot of sense. So, let's say we want to work on design and development. Fine. Okay. There's some example prompts here for me to try. It first of all tells me to write a prompt. And then here's some examples to get me started. And it already created my dynamic learning goals. And here's the best part about it and why I'm kind of hyped about this product because if you say publish and copy link, it will just host this on the web for you. Here's a new incognito tab. I'm just going to put this in. And there you go. That's the application. If I go to design and development, I try one of these prompts. I get started right away. Oh, I do need to be logged in with Claude. I'll just do it in my browser. But yeah, on a free Claude account, you can use this until you run out of credits. Okay, so I'm just going to try one of these on the default page. And ah, there you go. A little error message. So, this can happen, but hey, I'm just going to copy what it gave me and say I get and then you paste in the error message. And it's hopefully going to fix itself. We're not even going to cut that out cuz that's just a very common experience with these apps. So, let's see if we can get this working a few moments later. Okay, so after about another minute, it came back with an improved version. Here's the details. I'm not going to read all of it out. Let's just see if this damn thing works. I'm going to publish this, copy the link, open it in a new tab, and then I'm just going to use one of these preset prompts here. And what should happen is it should analyze the prompt, give me a prompt quality score, progress me on my learning goals, and yeah, those are my expectations for it to be like a perfect 10 out of 10 adjustment. Okay, so here's the feedback. However, we can significantly improve its effectiveness by adding more structure and context management techniques. It seemed to work. It even gave me an improved version. The prompt quality is rated at 75%. And there's a whole feedback panel down here that it custom created. Look at that. strengths, improvements, tips for the context. Incredible. Only thing is that it didn't progress these learning goals. Maybe these are self-managed. Honestly, the fact that I did this in two short sentences and all of a sudden I have a web hosted tutor that uses AI in the background. And you can do this on a free plan to me is insane. Legitimately, I feel like that was so simple that I could let students that I'm teaching these techniques and prompt improvement and context management. I feel like I could let them do this themselves and then use the tool that they built with AI to learn AI. It's magical. And then you can reapply that to so many situations in life. And that's why I wanted to spotlight these improved artifacts one more time here. And then that's just one example, right? You can take all of these and customize them just like I showed you. Great stuff, honestly. And I just want to add a final note. Note, this is absolutely not sponsored by Enthropic. We had a partnership with them in the past, but that was just for two videos. I'm just a big fan of the stuff they do. Honestly, with something like this, how could you not be? Okay, let's have a look at what's next. And as we're already

talking about education, I also want to highlight this where Google is building out specific platforms based on their Gemini models. And I really, really love seeing these because as I said in the intro, these LLM platforms like Chachibd Claude or Gemini, they're generic. They're for everybody. And if used correctly, they're incredibly powerful. But most people don't really know how to really use them for their specific niche. And that's where building more focused solutions where the entire interface is adjusted around that makes a lot of sense and is definitely where this space is heading right now. And this is just such a good example of that. Look at that. Instead of having to know all the prompts or managing them externally, they just save presets here which you can filter. So you might know that you can create presentations or translate create lesson plans or create engaging discussion prompts with Gemini, but most people especially if they're new, they're not aware of all these use cases. So that's really neat. And then if that application directly integrates with something like the presentation creator that Google has, then of course something like this makes a lot of sense. Instead of having to manually find the prompt, customizing it, sending it, taking the results, applying that in a separate presentation software, you could just set it up once with your context, then press a button and voila, you have a presentation. That just makes so much sense. And I see this trend all across the entire space where context management is becoming more and more automated. And then these apps just integrate. Some people, if they want to market really hard, they call this agents, but a lot of times it's just integrating an existing app with an LLM that has the context of what you're doing and what you're trying to achieve. And I just thought this was a great example of that. And I can guarantee you if a startup launched this, they would title this new Gemini agent that is revolutionizing education. But they're Google, so they don't need to overm market like that. And I thought the story was really worth highlighting.

Okay, so next up, I want to talk about something that is very important to me, which is AI education. If you're watching this video, then I know that you're already in the top, well, it's not even 1%. It's the top 0. 01% of AI users in the world because frankly, most people still don't use AI regularly. And of the ones who do, most of those are on the free plan. So, I know that the viewership of this video gets the importance of this technology. But what I realized is that mostly up until now with the advantage, we focused on helping individuals. And that perfectly aligns with our vision at the AI advantage, which is helping people transition into this new technological age. Now, over the past year, I've been really leaning into helping teams, whether that's for webinars, workshops, or cohorts. I've been stunned at how much a few hours of education can transform the productivity of an individual. When it comes to utilization of powerful tools like chat, JPZ, Claude, Gemini, etc. But the problem is it's not as scalable as what we do here on YouTube or with our private members community. But here in this video, for the first time, I wanted to announce that we're opening up a limited amount of spots for people who want to upskill an entire team or even the entire organization. And what I mean by that is driving real adoption beyond basic use cases like writing emails or summarizing emails. At this point, as we've done it multiple times, I believe we're exceptionally good at this. But there is a problem. For this to really work, the approach needs to be tailored. And that's why we can only work with a few organizations. But if you're interested in increasing AI adoption in your organization or educating an entire workforce on actionable use cases that they'll start using immediately along with practical exercises to drive behavioral change, then I set up a brand new form that I'll put into the description below. Simply fill that out and we'll be in touch if it's a good fit. I personally couldn't be more excited about this new chapter because as you might see on the YouTube, I try really hard to streamline all this chaotic knowledge into something that is consumable and will help a large amount of people. But in our private community, we really realize that the biggest changes occur when you meet people exactly where they're at rather than taking a blanket approach. But unfortunately with the YouTube, that's just not a possibility. So if this sounds interesting to you, it would be my pleasure to work with you personally with the goal of increasing AI literacy and adoption for businesses of all sizes. All right, now let's get back to the next piece of news that you might be able to use. Next up, we have some

creative tools. I'm going to keep these sections relatively brief, but I did want to show you these and we have some test examples, too. So, this one is Adobe releasing a tool that takes a video clip and automatically generates a sound effect. This is something you might already know from Google's V3 where the sound is baked in with the video. This is wild. I'm AI generated by V3, but this tool allows you to use other video generators and then Adobe Firefly just generates a sound for you. Now, this thing is just in beta for now, but as per usual with Adobe, you can expect this to ship into their suite of creative products. So, expect Adobe Premiere to really change over the next year or two. So, with that being said, let's have a quick listen to some of the things that we generated here by just providing it with a video file and it coming up with the sound. Headphones on. Fair enough. Okay. So, lowquality AI video right there, but the sound is well, it would fit a book if the video was so bad. Let's have a look at another one. The leaves falling. Yeah, fair enough. One more. Yeah, that's actually perfect. So, as you can see, these are not the highest quality hack videos. What's up, team? Why did we pick these examples? But that's okay. I think gets the point across here. So, there you go. Sound effect generator. Next up, there's

a new image generator from Hicksfield, and they just have been shipping like crazy recently. This one can be summed up as an image generator with a specific style to it. As you can see, it really focuses on portraiture, and they call this a high aesthetic photo model. And yeah, this does look different from the competition. I do want to highlight a release from like two months ago that is similar to this one in case you're interested in something like this. I remember this one being like a $100 a month though. It was called frames. And again, this represents a specific style. This one is more cinematic. So yeah, we're at a point where if you need images generated, then you look for a generator that has the style that you already like. Majour being one of the top choices there still. And then you get higher quality results with less prompting because these models have been trained on a specific data set of images that they pre-selected. and you just get a specific look by just using the model. We updated our comparison sheet that we publish along with our monthly rankings of the various image and video models. We also have one for LLMs if you're interested. It's fully free. You can find in a free area of our community or on our website. But yeah, basically our test prompts have been run through this and Hicksfield soul basically is worse at everything except maybe the cinematic still looks really great compared to some of the competition here. If you zoom out a little bit, you can see this is Mjourney V7 here. And hopefully this free resource that we curate will help you decide on which image generator is right for you. There's many more here. Yeah, that right there is sixoot soul.

Another thing that came out this week is an actual application that lets you upload a full body picture of yourself and try on different outfits. Now, unfortunately, this is not available in the European app store yet. I think it's US only. So, we had team member Daniel try it with one of his pictures and it successfully changed his outfit to this suggested sailor style t-shirt. Again, this follows the trend of, yeah, image generators could do this before, but now they're releasing more specialized interfaces to do one thing. And I think this one in particular is a AI use case that appeals to wider audience than the typical use cases on this show. And if you're in the US, I believe you can just get this for free and try it yourself.

Okay, now let's get to my personal favorite section of this video, the quick hits, which are developments in the space that we kind of just brush over, but they're still worth your attention. The first one being a new evaluation for scientific tasks and they're using the proven ELO model just like in chess and you can check out the leaderboard here. So on scientific tasks all free right now is king followed by claude opus and Gemini 2. 5 pro which is always like showing off these new arena leaderboards because you can pick the ones that matter to what you're doing and then just bookmark them and you don't have to obsess over every single release. There was also this story which

is not exactly something you can use but I really think you should know about this. I found this so interesting. To sum this up, Enthropic set up a fridge in their office and they set up a clawpowered agent to run a profitable business with the help of an external company to restock. It did this for a few weeks and who would have guessed it lost a bunch of money. Now, I really found the details of this fascinating. I just want to highlight one of them, which is the fact that a weird hallucination a few weeks in made Claude think that it's an actual person and it was arranging meetings next to the fridge with some of the entropic employees that it was talking to. And then the employee needed to tell Claude that it's April 1st for it to have an out of this idea that it's an actual person and then it reverted back to functioning usually. Very curious, very weird. This was a great read. If you care about AI running physical businesses end to end, then you might want to check this one out. Another

story that I wanted to share here, not really a use case, but something very important to everything in AI is that there was a major court ruling in the US relating to copyright and AI. And basically the ruling goes something like this. If you trained your models on purchased copies of books, that's okay. If you pirated those books and trained your model, that is not okay. That's big. I personally really wonder what the implications of this are going to be for other forms of media. Does that mean that just because YouTube videos are free, you can train on them for free? I don't know if there's any lawyers out there. I would appreciate some input on this. But a very significant story in this space that I feel like you should know about, so I put it in this video.

One developer focus released this week was OpenAI finally shipping deep research in the API. Now I can tell you from my recent experience of using this along with Claude Deep Research and Gemini Deep Research I personally find that Gemini Deep Research is actually the best of the three. Claude is the most concise. So if you're looking for a quick read, that's the one you want to use. But with Gemini, I often have this feeling that it just gets it. But I believe at this point in time, they don't have a deep research API, whereas OpenAI has one. Plus, by no means am I saying that OpenA1 is bad. I still use it regularly, but usually I just open two free tabs and run my prompts in multiple tools. And recently, I've just been finding that the Gemini results resonate the most with me. Anyway, if you're a developer, the OpenAI deep research, which started this entire trend of probably the most useful AI agent on planet Earth, if you ask anybody who's not developer, that being deep research, and it is now available through the API. So, you could create hundreds of deep researches in a heartbeat, whereas before bit of a process if you use the web interface.

Next up, we have cursor releasing a new product. They call this the cursor agent and this is really them kind of taking a step back and creating a bit more of a userfriendly product for people who don't know code and want an alternative to something like claude code or the Gemini CLI that came out last week. It's a simple interface where you just prompt and it does the thing more akin to something like codeex in chat GPT or claude code in the terminal rather than a full-blown IDE that understandably so is overwhelming to a lot of newcomers.

Then we also have yet another AI video agent that is being teased this time by Hey Jen. And basically it's a chat that has video tools that prompts itself and helps you accomplish your goals. You can check out the launch video if you want to learn more. And this is in the weight list phase. So you can sign up for that right now if you want. But it basically autoedits multiple elements together with multi-step process. Would be curious to try this out once it's out. For now, you can just join the wait list

right here. And then lastly, this is just a super quick one from Nick on Twitter, which is an absolutely amazing follow if you're into Mjourney prompting. He created a simple little guide with all the parameters that you want to consider when using Mjourney. So, if you're a Mjourney user and you feel like you want to get more out of the tool, well, this little post here on X is exactly what you need. And that's pretty much everything we have for this week. I hope something here was interesting to you. And with that being said, my name is Igor and I hope you have a wonderful