Hello, I'm Mike with this week's new and trending AI tools you can't afford to miss. coming up this week, Google allow you to create mini AI apps using just natural language. ChatGPT agent is actually insane. and Adobe invite all AI companies to come play in their image, video and sound effects generator that's coming up later in the video. We need to talk about Google Opal. Oh my goodness. Take a look at this. Because this is like literally bang on for the stuff I love doing. Describe create and share your AI. Many apps essentially you can have an input you can generate using Google models under the hood, and you get an output. That means Imagen Veo 3, and all of Google's AI models can be chained together to make a mini app, as they call it. Let's try. Here's the challenge I'm always trying to solve. Take a YouTube video URL. Generate four thumbnail ideas for it. The thumbnail should be clickable and use YouTube best practices to get high click through rates and text, but a maximum of two short words relevant to the video in bold. Modern font. that was easy. There it is. Now I use Wispr Flow to talk to my computer, but note that Google actually has a microphone icon too. Let's send and see what gets created. Wow. That didn't take long. Maybe just a few seconds. And now we can see we've got a little flow here. You get the YouTube video URL as input, and you generate thumbnail images. Look at this. We can actually see there's Gemini 2.5 flash under the hood and it's taking that URL. This is step by step instructions generating the images here. And we can see flash image generation is being used. I'm not sure how good that will be at putting text, but we'll give it a go and then render the thumbnails out as HTML. Now obviously that's all techie. We don't need to worry about that because our natural language prompts basically made this mini app. Let's test it. So now over here in the preview tab I'll just click start and look at this enter YouTube video URL to analyze. I really can't believe this is going to work, but let's test. Over here on my YouTube channel, I'm just going to grab one of my latest videos, go back to my Opal app here and paste in the URL. Hit enter and see what comes up. Okay. We can see that it's jumped up. It's clearly gone to the next step. And we can see it's actually generating those thumbnail images. When we click over to console. So we'll wait just a little bit longer. Boom. Insane. High CTR YouTube thumbnail concepts. Wow. With Anime AI and Groq fast. Indeed I did talk about these things. This is very, very interesting considering I made that up with one single sentence. But look at this. I've got access to all these different models from Google. So if I want to go even more intense I can go for Imagen 4. Let's swap that out and we'll run that app again. Okay, that was really quick. I'm not sure I like these as much as I like the 2.0 flash image generation thumbnails, but there you go. Some refinements of models. It's just a case of clicking into each thing, dropping down and choosing what I want to use to do each step. This really is amazing. You can save stuff to Google Docs. Really, the whole Google ecosystem seems to be open to you via Google Opal. now. Of course I'll share the apps I create in the description down below. And if you like what you see so far, do throw a like and subscribe as I do videos like this weekly. take a YouTube URL and turn it into a blog post. Generate two relevant images for that blog post, and insert them somewhere appropriate. Okay, pretty easy prompts. Let's get it rolling. Boom! It's done it. It's chained together. All of this stuff. And look, we can actually see it. Ask for the YouTube URL. It then generates a blog post from YouTube using 2.5 flash. Let's change that to Pro. Then it generates an image prompt using 2.5 flash. Then the images. Let's use imagine for. That's their most advanced model, and then it creates that blog post with the generated images. If this works as well as the last app, I'll be impressed. I'll title the app YouTube to blog, and then we'll go to the app view so we can actually try this out as an end user would use it. So start again. I'll need a YouTube URL. That's fine. I'm just going to use an old video of mine, paste it in here and hit enter. it. Now look at this. It's gone straight to work. Creating the blog post. If we like, we can look under the hood and actually go to our console and see at what stage it is in the Google ecosystem. So it's using Gemini 2.5 Pro Preview, and we can see all the parts of our prompt if we like, by opening it up. It actually has a system prompt here saying today is and then today's date you're working as part of an AI system. So no chit chat and no explaining what you're doing and why. Interesting to see under the hood of a prompt made by Google Opal. Okay. It seems to be wrapping up. And look at this. Automate your YouTube comments. How I built an AI employee. Minutes. Indeed I did. And look at this. This is actually not too bad. It's inserted two images back to back in this. A load of gobbledygook on this one, but we'll ignore that and see how it's actually made a, you know, reasonable blog post. This definitely is not ready to publish, But we can definitely see with a bit of refinement and maybe going back to the editor and changing the system instructions here, we could probably come up with something half decent. There we go to open apps generated. Let's create one more Check the latest Google trends around AI and no code generate a short video and social media post of 140 characters max to help me, an AI influencer, hop on this trend. Now, I'm not sure how well this is going to work, but it's definitely worth pushing Google Opal to see what kind of mini apps you can create. And I've got to say, this workflow is looking pretty crazy. Okay, target trends influence the name as input research the trends, generate video instructions, generate video, combine it to HTML and give us a social media post. If this all works, I'll be impressed. So let's go to the app view. Let's get rid of all that, node or code view. And we'll actually start up our app. So enter the keyword for trend analysis. Let's just type in something basic like AI tools and then the influencer name okay. As much as I don't like the word influencer, let's use Mike Russell and see what kind of magic my new Opal app will do. Under the hood over here, we can see it's taken in trends. The name it's researching those trends. It's using Gemini 2.5 slash. It's searching the web. I didn't even know it has that capability. It's calling Gemini Flash again. It's using all kinds of models all chained together here. And we can see the flow in action right now with this beautiful animation. Oh, by the way, while that's working away. And we're waiting for our result. These are mini apps that you can create in Google Opal. It's pretty epic, but I like to create fully fledged apps that can solve complex problems. And that's exactly what I've done in my four week course, Ship Your App. It just concluded we did the last live clinic yesterday, and I was absolutely blown away where completely unscripted. Three of the students on the cohort broke into phrase so rather me tell you about it, this is what they had to say. You know, I work in technology. This this has opened up, a significant new avenue for me, both personally and professionally. I think there are very few people have the capability that you have in terms of your ability to deliver content at this style. It's been great meeting everyone else in the cohort. It has been incredibly a collaborative experience, and I look forward to to many more courses that you will bring out, and I certainly would love to be a part of those as you go forward. Big thank you. It's been, you know, very insightful, you know, and just going through the whole process with you the last four weeks, I've enjoyed it immensely. Mike, thanks for your infectious enthusiasm. And that's why I gravitated towards the course. I was like, who's this guy with the fancy shirts and the voice and everything? So you have a way of disseminating that information and encouraging the knowledge. Okay, it's on to the last step. Video generation here. So while we wait for that to complete Ship Your App in Four Weeks has opened for cohort number two. Earlybird pricing over the weekend only. So if you want to get in, it starts on Monday. It's going to be super exciting. I will hold your hand and help you launch your app in four weeks. The link to join is down below. Go and get your Earlybird discount and I hope to see you in the second cohort. And in the time it's taken me to tell you that, it seems to be wrapping up now by generating the social media posts, we can see this beautiful animation in Google Opal. It's just combining the outputs. We should have a result soon. And wow. Oh my goodness me. It is generated this. This is insane. So that's the video it made Mike Russell unlock Google Trends power with AI and no code. Well, it hasn't really given me something I can post on my social media, but it did make this nice video. As you can see, it's experimental, and I'm sure with more prompting, more refining, more tweaking, we can make something pretty interesting with Google Opal. but I can actually look through the log and see what happens. Under research trends. We can see it searched the web. It found a bunch of details and sources online with different trends that are happening right now. So it's powerful. You can build an app with Google Opal that can actually access live data on the web. Okay. Next up. And ChatGPT agent has now pretty much fully rolled out. Every plus user should have it. You can activate it by typing slash agent and go into agent mode. So what is ChatGPT agent? Well, essentially it spins up a virtual computer that can browse, read websites, click around, take actions, and also run code if needed. You got suggestions here of what it can do, such as creating a list on Google Maps of the best Copenhagen bakeries you can generate, reports. You can take actions like booking Tesco deliveries. That's a supermarket here in the UK and ordering ingredients, just as I've actually seen someone do on X recently. It can work with spreadsheets, they can make presentations, but hey, I'm going to try it out on Creator Magic. Find out all you can about the Creator Magic YouTube channel and community, and then generate me a list of potential partners that I could collaborate with that would be of interest to my audience. There you go. There is a very simple prompt. Let's see if the ChatGPT agent can handle it, and you'll see it immediately goes to work. It starts thinking about what it's going to do. And then it says it's setting up my desktop. Now, what's cool about this is it spent some time thinking, But now it's actually spun up a web browser and it's looking for my YouTube channel. It's clicking to it. It's reading up about Creator Magic on LinkedIn. It's found my community. It's finding blog posts. It's looking at my premium community. We'll come back to that in just a moment. But I wanted to share a couple of other really cool use cases that I found for ChatGPT agent, and one of them, as many people have already said, is booking travel. But this is the coolest hack and prompt I want to share with you. I'm planning a trip to Dubai for the end of this year, and I'm searching for flights. And I've said here that return flights should be less than £1,000. That's just over $1,000 for all of us. That's a family of four. Happy to do stopovers for 1 or 2 nights each way if it reduces price. Now, I know this is quite a cool travel hack, and let me show you a comparison. If I do my search the old school way and search for a flight from London to Dubai. Around the time frame, I'm looking to go. I need to select the passengers. That's two adults. One young adults and one child. Click on Find Flights. There we go. I clicked on to the cheapest flights available. And you'll see a total of £1,936, which is about $2,600 for a family of four to fly to Dubai in December 2025. Now, if we look at ChatGPT agent and what it discovered, it clicked around a bunch of different airline providers and found London via Istanbul to Dubai or London to Abu Dhabi, then a bus to Dubai or even London to Sharjah and via certain airline providers. Let's find the cheapest one. and following the first suggestion from ChatGPT agent, I looked at Wizz Air and to get to Istanbul from London Gatwick. Is just £305.92. Then we use AJet to get out to Dubai for passengers for £776. 776.80 + 305.92 is just over £1,000. £1822.72. It's nearly half the price of a direct flight to Dubai. At the current exchange rate, that's just under $1,500 for a family of four to fly from London to Dubai, with one quick stopover in Istanbul on the way. This is just insane that ChatGPT agent discovered that. And that means instead of spending $2,600, I'm spending $1,500 and getting extra culture time in Istanbul. Another use case that's really cool for the agent is that I've said I'm going to Chiang Mai, Singapore in Jakarta later this year with a commitment to the conference there. Also to Wroclaw in Poland. And I'd like to put on some AI community meetups. It worked away for nine minutes here, searching across the web, executing terminal commands to find the optimum places for me to host meetups in Chiang Mai, also in Singapore, in Jakarta and in Wroclaw This is really cool stuff. I really can't overstate how amazing ChatGPT agent is. Research. And if you're in any of the locations I've just mentioned, make sure you join my community down below and check out those meetup dates. That's. I hope to see you in a location somewhere in the world before the end of the year. Finally, it looks like my agent has worked away for multiple minutes, connected together a bunch of different ideas, and we can see here with the code it executed it, found a few no code platforms. It's found some coding platforms here. It's found some great AI platforms for video generation, audio generation and more. This is pretty bang on. We can of course replay all of those steps and look at exactly what my agent did along the way, everything that is recorded. But look at this. I've got a comprehensive high level report of all the different places I could be reaching out to and descriptions about those particular companies. This is really, really cool. And agents are definitely the future, although they've probably been overhyped in 2025. I think AI employees, the way to go, and that's the name of my new course you can check out. It's linked up down below. Learn all about creating AI employees to do even more than ChatGPT agent can do, because you can custom code them with natural language. Finally. This week I love Adobe. In fact, my other YouTube channel featured their products for 15 years. Well, Firefly is a big product of theirs. And just this week, they've announced that they've added other video models to use inside the Firefly interface, including Veo 3. Now I did some calculations. And Adobe are presently offering Veo 3 cheaper than any other AI provider out there, almost at half price. So it kind of makes sense to sign up for a Firefly Standard Pro, or even Premium account to get access to all of these generative credits. Now do note that while the Firefly model is pretty much unlimited, external models do come with credit costs. And let's explore those now. So over here in my Firefly control panel, I'm logged in and I can choose to make an image, a video, or even a vector. Let's go for video and we'll click generate. This will take us to the video generator interface. Now you'll see I've got Firefly video selected here by default. But we don't need to use that. We can switch over and we can choose Veo 3 from Google now resolution seems to be locked at 720p, so does the aspect ratio. Frames per second can't really change anything here. All that you can do is add on audio and you can see down over here it says it's a premium feature, and these are the amount of credits that will be used 800 credits without audio and 1200 credits for a generation with audio. So let's give it a try. A magical wizard waving his wand high in the sky and singing creator magic, first on his own and then in harmony with other spirits. There we go. I think that's a pretty decent prompt. Let's hit enter and let Veo 3 go to work via Adobe Firefly. Now this takes a little while to generate. So while Veo 3 works away here, let's try something else. This time I've got image selected from the drop down. We'll click generate. And yes, we do have the latest Firefly image for model here. But look at all of these part in the models image. And for from Google we've got GPT image generator that really cool one. And also all the Flux models here as well. Now you have to add them to your account. But let's go for something like for instance Flux 1.1 Pro will add it. And Adobe do make it clear you're using an external model here. That's fine. Continue. And we can look again at the credits here. And you'll see it's 10 to 60 credits per image generation. Here's my prompt a magical tabby cat made out of stained glass with the word creator magic on a window behind it. That looks decent. Let's run that. And what do you know? That's not too bad, is it? That's not too bad at all. But we can actually close that. And the beauty of working in Adobe's Firefly is we can quickly switch. Maybe I'll go for Imagen, and for my prompt remains there. Click generate. And look at that. Now we've got a Google image for creation which also looks really good. So we can compare and contrast models right here in the Firefly playground. Of course I can go for Firefly Image for ultra and give that one a test. And of course with Adobe's native model, we get a bunch of other suggestions as to effects and styles that we can use. This one has a lot of complex styles, but we'll just leave it on default. And it's not too bad either. This is the Firefly generation. And finally, just for fairness here, we can actually go ahead and add the ChatGPT model from OpenAI and indeed will run exactly the same prompt one more time. And look at that. That is the ChatGPT result. They're not too bad. So, Adobe Firefly really trying to become the aggregator of great generation models to power your creativity. Let's go back over to this video generation. Here's our wizard with sound. Creator Magic. Creator Magic. Okay, that's pretty good. But there is one more feature I want to show you. I'm going to stick with Firefly Video this time and bump the resolution up to 1080p. A magical lion walking around and roaring in a wizards layer. We'll let this generate. And you'll see. On the native Adobe model, we actually get a status tracker here. So we can see when our generation will be complete. And look at this. It's come to life. It looks good. But wouldn't it be good if we could generate sound effects to go with this video? so now, over here in the Adobe Firefly sound effect generator, I can give a prompt as to the sound, or I can use my voice as a reference. So. Oh. Wow. Wow. apologize for the poor performance there. We'll just type in here, lion growls and then we'll click generate. And it's going to say my voice performance and the text prompt here, combine them together to make awesome sound effects. My video. look at this has already done it. Exactly. Based on my voice directions. And then I can go ahead and select an audio track and just drag the one I want onto the timeline, and then mix it down by clicking the download button up here. I can also separately export any version that I like of the generated audio. This is really cool. I hope you've enjoyed what I've covered this week. Google Opal make mini apps using Google's AI ecosystem. Really cool stuff. Then we get into ChatGPT agent, which is an insane start from ChatGPT to make things fully autonomous and Adobe inviting everyone to come play with their video, image and even sound effect generators. Now, as you can tell, I really enjoy doing these videos every week. I hope you do too. And if you want to continue the conversation, the best way to do that is in my community and is linked up down below. Oh, and by the way, YouTube is showing a video on your screen now you should watch next. Thanks.