Transcription
Google never stops. It continues to expand its digital empire with constant updates in artificial intelligence. Competitors like Antropic seem to lead at times, but the infrastructure and capabilities behind Yemin allow models like Clot to survive with cutbacks and uses that border on the ridiculous. Google knows this, and that's why it's accelerating, launching a number of updates that are impossible for other companies.
At first, I thought they were minor advances, but they can truly change our workflow completely. To see for yourself, we'll be testing all its new features through eight use cases that will allow you to do things like visualize any concept in seconds to understand it, receive real-time assistance from anywhere, generate voices so realistic they 're already top-tier, or even create your own platforms, among other use cases. The best part is that, as you'd expect from Google, everything will be free. Sounds good? Let's take a look at the first new feature.
In the first case, we'll see how Jimine is quietly releasing small updates and new features that are actually very useful for our daily lives. To discover this, let's jump to Gemini, the platform you'll find the link to below the description. Now, although everything might seem practically the same, if you ask it anything from here, like asking it to show you how a car engine works, clicking "send" will start processing it. While it may seem like it's giving us a normal answer, it's not. If we keep scrolling down, we'll find a new button next to the explanation that says "show me the visualization." Clicking it will start loading the visualization. Notice how we get an explanatory visualization of what we asked it to show us. In this case, we have an internal combustion engine, and from here we can see different steps, for example, the intake. Clicking on it will show us the movement it undergoes. In second place, we have compression, which would be shown here. It also explains that the piston rises, compressing the mixture, and both scales are hermetically sealed. I click on the third step, power. We see how it has created that animation. It also provides the explanation here. And when I click on the last step, called exhaust, we can see how the gases escape, along with the explanation above. In addition to all this, we can also play it continuously simply by selecting this button, and notice how the entire movement appears here, showing each phase, angle, and a visual RPM speed that we can modify. By default, it would be at 200. I could increase this even more, as if it were a real engine, but I can also slow it down considerably so we can see all these steps we just read about and understand everything more visually. And with this, we're no longer talking about it explaining something in an easier way to understand; it will literally create animations tailored to what we need to learn, making everything much simpler.
Knowing this, I'm going to move to a new chat, as I want to show you some more examples of how these visualizations change. If I ask it to explain how gravity works, before sending, keep in mind that in my tests, this has worked when I use the pro option. And then, when we click send, it starts processing. Once we have all the answers written down, we'll again have a visualization button. And notice how it has created a space-time gravity well, from which we can even navigate, since we have a 3D visualization of what's happening, and it will be interactive again. For example, I can increase the mass of the star. Notice how the depth here... It would have increased, and the ball would be even closer at certain points. Conversely, this could also reduce its mass, and we can see how that well is even smaller. Along with this value, we can also change the speed of what we have here, from which we can also zoom in or even remove it. I can increase the speed again so we can see how much faster everything would go. Similarly, I could move it and again, make it much slower. And look, these aren't just simple explanatory movements; it can literally create 3D simulators if the concept we want to learn requires it for easy visualization.
Along with this, I also want to show you another case where we could be talking to Gemini and then later ask for these kinds of visualizations. To do this, from a new chat, I'm going to ask it to show me the complete flow of how Gemini's artificial intelligence models work to convert an input into a response. And by clicking send, I could not only get this response, but even though it didn't give me the option to visualize it at the end, I could ask it to show me the complete flow visually. And if I click send, a little later we have this other flow here where we can see all these steps. On one hand, how it divides the text into manageable units. Then we have the next step here, where the entire embedding section takes place, which we'll again have the explanations for above, and along with this, we can also see how the attention shifts to the generation where it detected that the phrase means the house is more spacious. And if we click next, we get the final response telling us that the house is more spacious. If we go back to the very first step, we can see that the input was "my house is big."
In addition to this, we can also connect it with any information source we can add to Gemini. And so, moving on to a new chat, one of the most important sources we could incorporate would be directly from Notebook LM. If I select, for example, the option for healthy habits and tips, clicking "add" will give me access to over 47 information sources I have in that notebook. So now I can tell it directly that, beyond all the formats Notebook LM can create, it should show me how the key elements of a healthy habit are related and create an interactive visualization. I click "send," and a little later, the result appears here. If I click "show visualization," we can see how we have those five blocks related to the biological system, where we can find sections such as nutrition, physical activity, emotional well-being, rest, and lighting environment.
In the second case, we'll be looking at a new model launched by Yemini, capable of holding video calls, screen sharing, and ultimately, interacting with us through different formats. This is possible thanks to Gemini 3.1 Flash, a model that not only improves upon previous versions in terms of speed, but also in efficiency and power in terms of the responses it can provide. To discover how this works, let's jump to Google AI Studio, another platform we'll be using. From there, when we click in the upper right corner and select the "Live" option, we'll find this new Gemini 3.1 Flash model. So, we'll click on it, and from here, it will enable different options, such as video calls or, what I find even more useful, the ability to share our screen. So, if I click on it now, we'll enter a call, and I'll be asking questions about this raw spreadsheet I have here. To do this, I had to click on "share screen." I'm going to select that spreadsheet. I confirm. I come back here to enable the microphone, and I start asking questions.
"Hello, very Hi. Can you see my screen?"
"Yes, I can see it perfectly. You have a spreadsheet with sales data for different products."
"Right. Correct. And look, I'd really like to see all this in a more organized way because it's all raw, but I'd like to be able to see, for example, the products with the highest margin, so I can sort them from highest to lowest. How could I do that?"
"Sure, it's very easy. To sort by margin from highest to lowest, you can use the sort function. First, select the entire data table, including the headers. Then go to the Data menu at the top and choose the Sort Range option. Make sure you check the box that says 'Data has a header row' and then select the Margin column in the Sort By section of Sort Range."
"Okay, but beyond the settings, I'd like to configure something in the table to make it easier and not have to go into this settings section. Could you do something to the table to make it even easier to make those sorting changes? Understood?"
"You want the sorting to be more accessible? One option is to create filters. Go to the Data menu and select Create a Filter. This will add small arrows to the column headers. By clicking the arrow in the Margin column, you can quickly sort it from highest to lowest with just a couple of clicks, without having to open the Sort Range window each time."
"Okay, I just clicked the Margin filter here. Now how can I sort it from highest to lowest? Because it shows A to Z and Z to A."
"It's right there. Since they're numbers, A-Z will sort from lowest to highest, and Z to A will sort from highest to lowest. What do you want?"
"That's right, I just clicked it, and look, I've already applied it here."
"Perfect, now you have the list sorted by margin, showing the products with the highest profit first. Would you like to make any other changes or analyze any other aspect of the data?"
"No, just that. Thank you very much."
"You're welcome. I'm glad it's more organized now. If you have any other questions or need anything else, please don't hesitate to let me know."
I'm going to end the call here; I'm stopping screen sharing. Until next time. It just said goodbye to us there too. And notice how it now not only responds to us in a much more intelligent way, adapting to the different situations we might encounter. For example, here it told me to sort the table using one method, but then I told it that I wasn't convinced, and how it was able to give me a second option that is perfectly valid. But besides this, the most impressive thing is how little time there is between my question and its answer, since everything you've seen hasn't had any kind of interruption; it's practically as if I were talking to a person in real time.
In this third use case, we'll be looking at another model that was just launched yesterday at the time of recording this video, and it's truly amazing. The specific model is the Yemini 3.1 Flash TTS, and it has literally positioned itself as one of the best models for text-to-speech. I've been testing it since yesterday, and it really does speak like a human. To verify this, let's jump back to Google AI Studio. From here, we'll click again on the models section, but this time, instead of "live," we 'll go to "audio." Notice that the Gemini 3.1 Flash TTS model will also appear as new. So, let's click on it, and we'll see a new visual. If we click here on "Convert from text to natural voice," we'll see that we can even choose between different voices. And along with that, notice how the precision also shows us that we can add emotions within brackets. So, let's put this to the test.
In the context section, I'm going to say a Sevillana song while testing a microphone. And now, where we would normally have the voice, I'm going to put what I want to record. Here, except for the emotions, I'll be saying, "Hello, this is a test. Up to this point, I should be saying it with some hesitation. I would ask if it sounds good to you." "How do I sound? I'd cough and then finish with, 'Hey, this cold is driving me crazy, son,' and I'd have to end with some laughter. Let's see if this can nail it." So, I'm going to press "ran." This would start processing, and in a few seconds, we'd have the audio here. Let's see how it turned out.
"This is a test. Does my sound sound okay to you? 'I've got a cold, son, it's driving me crazy.'"
And look, it really nailed it here. It hesitated perfectly. We even had a question mark here, like a query, and then it could have made it more affirmative, but the way it linked the doubt between the previous and next sentences, it coughed perfectly and finished with that laugh. So now we're going to test it with other specific cases to see if it still works. For this, we're going to play with the scene section. So, from here, I'm going to tell it that a person is arriving at the beach. In the context section, I'm simply going to tell it to use a Canarian accent. I'm going to Let's add the following, where I tell him that in a relaxed way he says, "Dude, this is awesome. It makes you want to stay here all day." He gets scared like he's about to get on a bus and says, "Listen, kid, I almost got hit by a bus in my rush." Let's see how he would say this. First of all, I'm going to change the voice so we can hear another one. In this case, notice how we would have it labeled. And notice that I'm telling you manually how you could add all those styles, since for those of you watching the video in Spanish, unfortunately, from here in the accent section, the preselected ones are only English, although as you can see, we can also emulate it manually. So now from here I'm going to select this third one here called Algen Genip. I already have it selected here, so I'm going to click on "ran" and a few seconds later we'll see what it's done.
"This is awesome. It makes you want to stay here all day. Listen, kid, I almost got hit by a bus in my rush. Bus in a hurry."
And look how it said it, I mean, how well it 's taking into account all the emotions we're putting into it. The thing is, with this, whether we want to create a story, voice a professional ad, or whatever we need with a voice, today we have this model that's completely free and it really does produce very good results. Let's do a third and penultimate test to see if it's capable of other Spanish accents. For this, in the scene section, I'm going to tell it about a person who finished eating. In context, I'm going to say Valencia, that is, a Valencian accent. And from here, I'm going to give it the following phrase. So again, I click on rank and let's see what it did on the first attempt.
"Yeah, it was all delicious, dude, worth repeating, eh, now I'm finishing up, you've never tried this before, mate."
And the truth is, it did quite well. Let's finish with one last use case where we'll be Testing with many emotions. And here we're also going to separate from Spain, since many of you watch me from Latin America. Specifically, the second country after Spain is Mexico, so let's do an example from there. For this, as a final example, we're going to do a kind of story. So, from the scene section, I'm going to say, "A friend tells a story." We're going to give him a Mexican accent, and from here I'm going to give him the following command: "Soon, whisper." Then we're going to add a pause along with a soft laugh, then we're going to tell him to increase the intensity, then from here we tell him to laugh loudly, then we go back to a quick whisper, and then to shout loudly—that is, a whole lot of emotions that even a human would struggle to produce in this sentence. So let's see what he's capable of doing. I click on "rank" again, and from here we could find this.
"You won't believe it. This starts calmly, but suddenly everything spirals out of control. Do you hear it? It's coming... It's coming. Now run."
The truth is, with all the new things that keep coming out, we can really apply it to both We can handle our own projects as well as those of third parties, offering services to them. So, if you want an online presence, I recommend Hostinger as a platform, mainly because they have various products that make this very easy, especially with Horizon. This platform will transform everything we put in the prompt section into a fully functional website or app, and by fully functional, I mean they're even adding packaging. Let's see how this works. I'm going to ask them to create a modern, premium website for a professional voiceover company focused on speed and quality. It will be called Locutavi, and I've added some extra details. So, I'm going to click submit, and a little later, on the first try, look what we have here! We can find that professional website we requested, where we have different sections like use cases, but the most surprising thing would be sections like the demo, where we can see that it has even added different audio files that we can play back to hear each of these voices and allow clients to request that voice for voiceovers. And this is thanks to Horizon, one of the tools that makes everything so easy, because in addition to generating an application using artificial intelligence, it has a whole infrastructure behind it, being hosted by Hostinger, where we have all the hosting services to easily publish it and also add different tools. Among them, we can now add authentication services, databases, and tools to make it 100% functional from start to finish. And thanks to the hosting sponsors of this video, they've given me a special link, which I'll leave in the description below. If you add the coupon code "Javi" to it, you'll get an additional 10% discount on their promotions.
Moving on to the fourth use case, let's quickly see how we can even connect this latest Gemini TTS model we've seen with custom tools we want to develop. To do this, I'm going back to Google Studio, but instead of using the Playground section, I'm going to go to the Build section, since this will allow us to build applications directly by connecting the latest models released by Yemini. So, the first thing I'm going to do is select this text-to-speech model, and in the Prompt section, I'm going to add this extensive prompt, which I'll also include in a document below in the description if you want to use it. I'd paste it here and here. Keep in mind that I'm basically telling it I want to create my own voiceover tool, similar to platforms like Level Labs, so that it has certain features and options suited to my workflow or a platform I want to offer to some clients. So, with all that in mind, from the settings section, I'm going to select the latest Yemini Pro model, the most powerful one they offer. I click on build, and it starts processing from there. But look, another quiet new feature Yemini has launched: now, when you're creating the application, while Yemini handles the backend, it will show you different interfaces you could use for your app. For example, here we have an elegant dark theme, a second option here, a third one we have this one, this other one, and we could also skip these or request a specific design from here, making the waiting times even shorter.
While all this is processing, I want to tell you about a second new feature within the Yemini API. This is very useful, especially if we want to automate on other platforms, connect it with other applications we have, for example, in tools like Hyon, or even develop certain tools directly from here. And it's thanks to the new feature they've just launched, which allows us to pay for the API with a prepaid plan. In other words, you can now set a balance you want to add to your APIs. Here, for example, you could add $25, and once it's loaded, you can disable auto-renewal. So, with that, we can say goodbye to the fear of an API going haywire and spending all the money we have, since this way we can add credits in advance without auto-renewal, giving us much better control. As I've said, notice how I already have the application created here, so to see it better, I'm going to click this button to enlarge it. And from here, we have a super cool application. With a "prom" section here, we can add different emotions, as we saw before. It also lets us switch between different voices, although we could modify all of this as well. It also lets us choose between different regions and accents. Along with this, we can also choose the narration style and the main emotion. We also have a bunch of details settings. We also see the library of all the voices we've added, the history of audio we could be generating, the different templates we could use depending on the use case, and a settings section. Also, keep in mind that if I enter something like, for example, "Hi, this is a test to see if it works," and then add "excited" at the end and change the voice to, for example, Valentina, and click on "Generate Audio," you'll see how it starts generating, and in about 3 seconds, we'd have this result here:
"Hi, this is a test to see if it works."
And look, we can not only use these great templates, but we can also adapt them to any project. I just created my own tool for voiceover development, but we could also have applied it to other use cases, such as adding conversational chatbots to existing websites and applications.
With the fifth case, we'll also see how Google provides tools where we can use its technology in an unlimited and completely private way. This seems unbelievable, but they've released a new model called Yenma 4, and they've paired it with a new app called Google AIH Edge Gallery. This app, as you can see, is available for both iPhone and Android mobile devices. It's also completely free, so when you download it and click "open," you'll see an interface like this one. From there, clicking on any option, such as the chat option, will display all the open-source models you can install on your mobile device. I already have Yenma 4 installed, which has 4 billion parameters. We were discussing this in this video; we were looking at all the open-source tools available for free, with limitations and no censorship. I'll leave that video on the card if you're interested. But now, to see how to download another model, it's as easy as clicking this download button. So, I'll click it. Notice that the download will start. The speed will depend on your internet connection. And that's it, a little later it should be reaching 100%, in fact, it's already there. So I'm going to click the "Tri Now" button. And now that's done, notice how the model starts loading so that I can begin writing from here in the prompt section. But the most interesting thing is that I can do it without any connection. I, for example, right now, as you can see above, have airplane mode with Wi-Fi. So now I've disabled the Wi-Fi connection, and notice that this will continue to work exactly the same. To do this, I'm going to ask it, "Tell me, what's the best thing about the 'A' in Google?" Click send, and with that, look, practically instantly it starts writing everything super fast, and here we have that whole answer telling us, for example, about all the deep integrations in the Google ecosystem. It has more information, and if you're interested, I'll leave it on screen for you to pause and read. Besides this answer, which took about 30 seconds, we'll also have the option to use several other tools. To do this, I'm going to click back, return to the home section, and you'll see how we have agent modes with different abilities to perform various more complex tasks. They can also analyze images, transcribe audio, and offer other options. If you'd like to see them, I have them in the previous video I recommended.
With this next use case, we'll see a new feature that Gemine is starting to incorporate, which will completely change how we use the platform. They're now integrating Notebook LM directly into the chat to organize all messages. From here, we can move messages to any existing notebook or add files from existing notebooks to enrich the information. Once we have all this, we can choose any notebook we want to work with. And request everything we need from Gemini. Along with this, you can also create your own LM notebooks directly from GNI. We could give them a title like this. And once we have this notebook, we can start adding all the information sources we want, such as documents, spreadsheets, and PDFs. Once we've added them, we can start working in a unified way from Gemini.
Regarding the next use case, let me show you a couple of testimonials.
"When I was going to take the course, I called Javi, I wrote to him, and he said, 'No, I want to talk to the teacher because I understand online courses, and that's all there is to it.' In the end, right? And when he told me that you said you were going to do it in person with a small group, I thought, 'This guy's going to rip me off, he's going to rip me off.' And well, man, I hate being wrong, but in this case, I'm glad. Look, I do little things that are light years away for others who haven't taken the course, and I'm really happy about it. Thank you so much. Keep going, but be very careful with what you do, because if you do it this way, you'll get wherever you want. I'm telling you this from business experience, but this is the system. This is the system. Stick to it, because not many will. There are tons of courses being sold out there, but you're going to tackle this one. Well, I won't take up any more of your time. Congratulations."
These testimonials come from some students of the intensive Artificial Intelligence course I've been running, and by the way, this is the last enrollment I'll ever open. It's a training program I really love, but of course, doing it in such small groups makes it very difficult to scale the project. So, if you were interested, take advantage now because you won't be disappointed. There's no other training program like it, so close-knit with a group of 25 people to give you real support. And believe me, you won't regret it. I'll leave the link in the description below, and the spots will close this month, so if you're interested, take advantage now because you won't be disappointed.
With this seventh and final use case, we'll be looking at different Google Chrome extensions that we can all use today, since the sixth use case, Notebook LM with Gemini, is an option that's starting to be added to accounts, but not everyone has it. So now let's see how we can enhance Gemini starting today. For this, one of the first applications I'm going to recommend is this Google Chrome extension called Chat Architect, which will allow you to create folders similar to a Notebook LM notebook, so you can organize all your chats until that update arrives. So from here, if we click on "add to Chrome," we would simply have to add the extension, and once it's added, if I go into the Yemini application, notice that on the left side I would see both the folder we could open and all the chats we have. over here. And now we can move these too. So if I click the plus sign and name the folder something like "learning," then clicking "create" will create the folder here. Now, if I click the plus sign again, I can also add different subfolders, like, for example, I'll name this one "test," click "create," and see how this other little folder appears here. You can even customize the icons, but the most interesting thing is that I can now move any chat to a folder like "learning." In fact, I'm going to add these three chats, and it'll be super organized here. I can even compress and decompress this, and I can also add other chats to the subfolder so we really have everything organized, something Gemini really needed. It's true that some accounts are already starting to see this with Notebook LM, but it's also worth noting that it doesn't have the same level of functionality that we can directly achieve with this free extension.
Along with this, I also want to introduce you to a second extension, this one here, which might seem trivial, but it completely enlarges all the unnecessary margins in Gemini so that the text stretches from side to side and takes full advantage of the screen. To show you what I mean, let's go into any chat, like this one here. Notice all the margins on the left and right that make it look a bit worse. But if we add this Google Chrome extension, it will immediately show that it's been added. And if we now reload in Yemini, we can see how this becomes even wider and we can even customize it, since we can click on that extension and add even more pixels, for example, 10,000. And notice how this is getting even wider. In the case of my computer, it would be roughly 10,000 pixels, and that way I'd be using 100% of the screen. These are the little details we notice every day, and now we can have them with a simple click.
Let's continue with the last use case and look at the official Yemini desktop application, which has just been quietly launched again. Here we have a desktop version for Max, and I'll tell you about a small alternative later so that Yemini can continue to work with you if you have Windows, although keep in mind that this will eventually come to Windows devices as well. This has several advantages, so to see it, I'm going to click on download on Mac. I would have already done that, which is why I have the application open here. And the most interesting thing is that beyond being able to use it from here—as you can see, we have exactly the same options as in the web version— we can also share our screen, and if I click a shortcut, which would be the "more space" option, the shortcut will appear here. It's true that I see two because I have practically all the AI tools installed, but the one for Yemini would be this one here, from which we can click on "add more." I can share my screen, like this example we're seeing of Gemini on Macs. So I'm going to click it, and with that, I'll be capturing the screen where we left off, and from here we can ask it any question, which we can even expand to have a faster workflow and direct access. The most curious thing about all this is that this application was developed in just a few days using VIE coding with Google Antigravity. This is n't just my opinion; it's Google's own CEO who explained how a small team launched this initial phase of a desktop application in just a few days using Google Antigravity. So, if you thought artificial intelligence was already moving fast, just wait, because companies are starting to accelerate even more with all the... tools that they already have.
So, I want to give special mention again to the intensive artificial intelligence program that I've linked in the description for those of you with Windows. I also want to share this extension with you, as it will allow Gemini to track you wherever you are. I already have it installed. Notice that if I go back to the application's page, which is currently only available for Mac, I can click on the extension I just downloaded. And look how we would have Gemini following us wherever we go, in this case through the browser. The truth is, something new comes out in the world of AI every week, or even every day. It can be overwhelming or beneficial, but that's up to you.