WEBVTT 00:00:00.000 --> 00:00:02.200 Welcome to the deep dive. If you're anything 00:00:02.200 --> 00:00:06.679 like me, you've probably felt that exhilarating, 00:00:06.860 --> 00:00:10.060 but sometimes overwhelming pace of AI innovation. 00:00:10.339 --> 00:00:12.980 Totally. It's constant. Right. New breakthroughs 00:00:12.980 --> 00:00:15.019 seem to pop up almost daily, making it pretty 00:00:15.019 --> 00:00:17.679 tough to keep up. But amidst all that noise, 00:00:17.800 --> 00:00:21.100 there's one AI platform that's been quietly becoming 00:00:21.100 --> 00:00:24.800 a real powerhouse. And here's the kicker. Almost 00:00:24.800 --> 00:00:27.660 all its best features are completely free. Yeah. 00:00:27.839 --> 00:00:30.339 That's the amazing part. So today we're taking 00:00:30.339 --> 00:00:33.280 a deep dive into Google Gemini. That's right. 00:00:33.759 --> 00:00:35.859 And our mission here really is to give you a 00:00:35.859 --> 00:00:39.100 shortcut. We're going to unpack 28 incredible 00:00:39.100 --> 00:00:41.359 ways you can actually use Gemini's power without 00:00:41.359 --> 00:00:44.399 spending a penny. Think of it as democratizing 00:00:44.399 --> 00:00:46.979 these advanced AI tools that usually cost a fortune 00:00:46.979 --> 00:00:49.700 somewhere else. Exactly. From creating content, 00:00:49.700 --> 00:00:52.200 doing deep research. We'll show you how to tap 00:00:52.200 --> 00:00:54.600 into this really versatile tool. And that's the 00:00:54.600 --> 00:00:56.340 key, isn't it? Whether you're a student, maybe 00:00:56.340 --> 00:00:58.340 an entrepreneur working in a content creator, 00:00:58.659 --> 00:01:01.179 or honestly just someone curious about AI. Yeah. 00:01:01.179 --> 00:01:03.619 You're about to find features that can genuinely 00:01:03.619 --> 00:01:06.879 compete with expensive paid software. This deep 00:01:06.879 --> 00:01:09.420 dive, it really could change how you work and 00:01:09.420 --> 00:01:12.859 create. So let's jump right in. Let's do it. 00:01:13.099 --> 00:01:16.019 Okay, let's kick things off with Gemini as a 00:01:16.019 --> 00:01:18.239 creative powerhouse. You know, traditionally 00:01:18.239 --> 00:01:20.980 turning an idea into something real, it meant 00:01:20.980 --> 00:01:23.799 meeting special skills, lots of time, or, well, 00:01:23.980 --> 00:01:26.620 money. Yeah, the usual barriers. But Gemini is 00:01:26.620 --> 00:01:28.519 sort of fundamentally changing that equation. 00:01:28.599 --> 00:01:32.230 It really is. Thinking about building games and 00:01:32.230 --> 00:01:35.530 applications, Google has this platform, AI Studio, 00:01:35.750 --> 00:01:37.849 right? That's like the main workspace. Exactly. 00:01:37.989 --> 00:01:40.269 AI Studio is where you'll do a lot of this. It's 00:01:40.269 --> 00:01:42.689 the development hub, basically. And normally, 00:01:42.810 --> 00:01:46.069 making even a basic game is complex design, coding, 00:01:46.450 --> 00:01:48.870 testing, all that. But with Gemini, it kind of 00:01:48.870 --> 00:01:50.769 boils down to just describing your idea. You 00:01:50.769 --> 00:01:53.030 give it a good prompt, figures out what you want, 00:01:53.209 --> 00:01:55.930 writes the code, usually HTML, CSS, JavaScript, 00:01:56.090 --> 00:01:58.430 for web stuff, and boom, you get a playable thing 00:01:58.430 --> 00:02:01.430 pretty much instantly. Wow, so like I could say, 00:02:01.730 --> 00:02:05.450 create a game like Snake, but maybe make it a 00:02:05.450 --> 00:02:09.629 Vietnamese dragon collecting pearls against an 00:02:09.629 --> 00:02:11.919 ink -washed background. Precisely. And it would 00:02:11.919 --> 00:02:14.120 even try to generate images for the dragon and 00:02:14.120 --> 00:02:16.860 pearls. That's wild. Yeah, within minutes, AI 00:02:16.860 --> 00:02:20.099 Studio gives you a working game. It really democratizes 00:02:20.099 --> 00:02:22.300 game development. Anyone with a cool idea can 00:02:22.300 --> 00:02:25.259 try it. Even without coding experience. Exactly. 00:02:25.860 --> 00:02:27.800 And here's where it gets really interesting, 00:02:27.979 --> 00:02:30.879 talking about visual understanding. Gemini can 00:02:30.879 --> 00:02:33.599 actually try to recreate software just from a 00:02:33.599 --> 00:02:35.969 screenshot. Wait, from just a picture? Yeah, 00:02:36.189 --> 00:02:38.150 it shows off its multimodality, understand different 00:02:38.150 --> 00:02:40.090 types of information. You upload a screenshot 00:02:40.090 --> 00:02:42.990 of an app. The AI looks at it, understands the 00:02:42.990 --> 00:02:45.110 layout, the colors, the buttons, and then writes 00:02:45.110 --> 00:02:47.330 the source code to mimic it. OK, give me an example. 00:02:47.729 --> 00:02:50.189 So maybe you upload a picture of a finance app 00:02:50.189 --> 00:02:52.349 you like. You could ask it for specific things. 00:02:52.750 --> 00:02:55.430 I need a plus button here for expenses, categories 00:02:55.430 --> 00:02:58.710 like food, transport, bills, a dashboard showing 00:02:58.710 --> 00:03:01.610 the balance, and maybe a pie chart for spending. 00:03:01.750 --> 00:03:03.960 And it generates the code for that. It generates 00:03:03.960 --> 00:03:06.300 the code to build that interface and functionality. 00:03:06.960 --> 00:03:10.180 Yeah. That's incredibly powerful. I mean, think 00:03:10.180 --> 00:03:13.500 about it. Quick prototypes, internal tools, custom 00:03:13.500 --> 00:03:15.979 software without a huge development cost. Exactly. 00:03:16.099 --> 00:03:18.740 It speeds things up massively and cuts costs. 00:03:18.840 --> 00:03:21.740 Makes it accessible. OK, let's shift gears to 00:03:21.740 --> 00:03:25.219 images. Google's image generation. like Imogen, 00:03:25.439 --> 00:03:27.719 it's baked into Gemini, right? Offering a free 00:03:27.719 --> 00:03:30.060 alternative. Deeply integrated, yeah. And it's 00:03:30.060 --> 00:03:33.259 super flexible, often free, which is a huge plus 00:03:33.259 --> 00:03:35.780 compared to many paid image tools out there. 00:03:35.819 --> 00:03:38.360 Right. So text to image is the main thing. Your 00:03:38.360 --> 00:03:40.620 creativity is pretty much the only limit. You 00:03:40.620 --> 00:03:44.360 describe something detailed, like. create a surreal 00:03:44.360 --> 00:03:47.159 artistic image of a water buffalo grazing on 00:03:47.159 --> 00:03:50.199 fluffy clouds. Below is a Vietnamese terraced 00:03:50.199 --> 00:03:53.139 rice field at sunset, glowing golden orange, 00:03:53.439 --> 00:03:55.960 and it generates it. But it's not just the first 00:03:55.960 --> 00:03:58.419 images that the editing part sounds key. Oh, 00:03:58.419 --> 00:04:00.300 absolutely. The iterative editing is brilliant. 00:04:00.419 --> 00:04:02.280 It's like having a conversation with the AI. 00:04:02.500 --> 00:04:04.400 You don't need the perfect prompt first time. 00:04:04.740 --> 00:04:06.560 So with the water buffalo? Yeah, you get that 00:04:06.560 --> 00:04:08.960 image. Then you could say, OK, now add some traditional 00:04:08.960 --> 00:04:11.900 lanterns, make it feel festive. Then maybe. actually 00:04:11.900 --> 00:04:14.259 change it to sunrise with some morning mist. 00:04:14.580 --> 00:04:17.000 And it understands and adjusts the image. Exactly. 00:04:17.399 --> 00:04:20.240 It refines based on your feedback. It's a dialogue. 00:04:20.399 --> 00:04:22.399 That's very cool. And what's also super powerful 00:04:22.399 --> 00:04:26.899 is modifying your own photos. You can like add 00:04:26.899 --> 00:04:30.379 magic to your moments. How so? Well, things like 00:04:30.379 --> 00:04:33.939 style transfer. Make this photo look like a character 00:04:33.939 --> 00:04:37.560 from a Ghibli anime. Whoa! Or removing objects. 00:04:37.800 --> 00:04:40.019 Take out that coffee table, put a big fern there 00:04:40.019 --> 00:04:42.959 instead, or completely changing the background. 00:04:43.439 --> 00:04:46.439 Keep me, but put me on a busy street in Tokyo 00:04:46.439 --> 00:04:49.040 at night. Okay, that opens up so many possibilities. 00:04:49.160 --> 00:04:52.540 Social media, personal projects. Just having 00:04:52.540 --> 00:04:54.699 fun with photos, it's like pro -editing for everyone. 00:04:55.740 --> 00:04:57.920 But Gemini's creativity isn't just visual. It 00:04:57.920 --> 00:05:00.519 has an audio toolkit, too. That handles things 00:05:00.519 --> 00:05:03.579 like transcription, but also like making a whole 00:05:03.579 --> 00:05:06.740 podcast. It does. A really robust audio toolkit, 00:05:06.920 --> 00:05:09.680 actually. And yeah, it competes well with specialized 00:05:09.680 --> 00:05:12.040 paid services for things like transcription and 00:05:12.040 --> 00:05:14.290 more creative stuff. Tell me about transcription 00:05:14.290 --> 00:05:17.149 first. Okay, so video transcription with accurate 00:05:17.149 --> 00:05:20.449 timestamps, super useful for creators, journalists, 00:05:20.949 --> 00:05:23.769 researchers, anyone dealing with audio or video. 00:05:23.970 --> 00:05:25.970 How does it work? You just upload your file, 00:05:26.029 --> 00:05:28.449 video or audio, ask for a transcript with timestamps 00:05:28.449 --> 00:05:31.810 like HHMFS for each speaker change, and you get 00:05:31.810 --> 00:05:34.449 a full text output. Great for subtitles, pulling 00:05:34.449 --> 00:05:37.329 quotes, analyzing meetings, lectures, really 00:05:37.329 --> 00:05:39.860 speeds things up. And what about AI voices? I 00:05:39.860 --> 00:05:41.579 think some people still imagine them sounding 00:05:41.579 --> 00:05:44.279 robotic. Yeah, that stereotype is fading fast. 00:05:45.040 --> 00:05:47.480 The multi -voice text -to -speech in AI Studio 00:05:47.480 --> 00:05:50.319 is, well, it's seriously good. High quality, 00:05:50.600 --> 00:05:53.560 natural voices, lots of customization. But the 00:05:53.560 --> 00:05:56.079 standout feature is? Creating conversations with 00:05:56.079 --> 00:05:59.300 multiple AI actors. Each can have a totally distinct 00:05:59.300 --> 00:06:01.839 voice, style, tone. So you could make, like, 00:06:01.920 --> 00:06:04.560 a short ad? Easily. You could script it. Speaker 00:06:04.560 --> 00:06:07.399 1. deep professional male voice, speaker two 00:06:07.399 --> 00:06:09.480 upbeat female voice, then back to speaker one, 00:06:09.500 --> 00:06:12.540 but maybe more emphatic this time. Gemini generates 00:06:12.540 --> 00:06:15.319 the whole audio file. No actors, no studio needed. 00:06:15.839 --> 00:06:17.360 OK, that's impressive. But you mentioned something 00:06:17.360 --> 00:06:20.420 about turning any content into a podcast show. 00:06:20.899 --> 00:06:24.360 Using Notebook LM, that sounds huge. It is. Yeah. 00:06:24.639 --> 00:06:26.839 For me, this might be the most groundbreaking 00:06:26.839 --> 00:06:30.259 feature in the audio space. Notebook LM is this 00:06:30.259 --> 00:06:33.639 companion tool in the Gemini world designed for 00:06:34.250 --> 00:06:36.689 really digging into information and generating 00:06:36.689 --> 00:06:38.769 content from it. So what can it handle? What 00:06:38.769 --> 00:06:41.430 kind of content? Pretty much anything. PDFs, 00:06:41.610 --> 00:06:45.569 Google Docs, just raw text, website links, entire 00:06:45.569 --> 00:06:47.949 YouTube videos, just paste the link. Whoa, YouTube 00:06:47.949 --> 00:06:50.470 videos too. Yep, and even your own notes. It 00:06:50.470 --> 00:06:52.589 just ingests it all. Okay, so walk me through 00:06:52.589 --> 00:06:55.529 the process. Let's say I have a PDF about Vietnamese 00:06:55.529 --> 00:06:58.870 faux history. Right, a 20 -page PDF. You upload 00:06:58.870 --> 00:07:00.970 it to Notebook LM. then you give it a creative 00:07:00.970 --> 00:07:02.829 command, like create a deep dive conversation 00:07:02.829 --> 00:07:05.589 about this. The AI reads it, understands the 00:07:05.589 --> 00:07:07.930 key points, then generates a whole discussion 00:07:07.930 --> 00:07:10.089 script, maybe between two AI hosts with different 00:07:10.089 --> 00:07:12.149 perspectives or roles. And then? And then it 00:07:12.149 --> 00:07:14.970 produces the complete professional sounding audio 00:07:14.970 --> 00:07:18.610 file ready to publish in like five minutes. That's 00:07:18.610 --> 00:07:21.029 genuinely game changing. For anyone wanting to 00:07:21.029 --> 00:07:23.490 make audio content without the usual resources 00:07:23.490 --> 00:07:27.620 or skills, Wow, exactly it transforms the possibilities 00:07:27.620 --> 00:07:29.759 for audio creation. Okay, so we've seen the creative 00:07:29.759 --> 00:07:34.100 side Let's switch gears now to Gemini as more 00:07:34.100 --> 00:07:37.920 of a sharp analyst its ability to consume Understand 00:07:37.920 --> 00:07:40.860 and restructure information. Yeah, this is another 00:07:40.860 --> 00:07:43.120 area where it really shines, especially its video 00:07:43.120 --> 00:07:46.250 analysis Well, most AIs, when you ask them to 00:07:46.250 --> 00:07:49.009 watch a video, they're really just processing 00:07:49.009 --> 00:07:51.629 the audio track, the transcript. Yeah. Gemini 00:07:51.629 --> 00:07:53.730 is different. It can actually analyze the video 00:07:53.730 --> 00:07:56.649 frames themselves. It understands what's visually 00:07:56.649 --> 00:07:58.550 happening on screen. So I just give it a YouTube 00:07:58.550 --> 00:08:01.850 link? Pretty much. Pates the link, then ask specific 00:08:01.850 --> 00:08:04.129 questions about the visuals. OK. Like what? All 00:08:04.129 --> 00:08:06.750 right. Say you paste a link to a travel video 00:08:06.750 --> 00:08:10.829 about Hoi, an ancient town in Vietnam. You could 00:08:10.829 --> 00:08:13.339 ask. list all the different types of lanterns, 00:08:13.439 --> 00:08:15.699 shapes, colors that you see, and tell me the 00:08:15.699 --> 00:08:17.879 timestamp when they first show up. And a transcript 00:08:17.879 --> 00:08:20.660 only AI couldn't do that. Nope. You'd have no 00:08:20.660 --> 00:08:22.620 idea. Yep. But Gemini could come back with, you 00:08:22.620 --> 00:08:28.680 know, 0 .32. Round red lanterns, 0 .1 .15, garlic 00:08:28.680 --> 00:08:32.120 -shaped lanterns clustered together, 0 .2 .47, 00:08:32.820 --> 00:08:35.500 cylindrical blue ones. That's incredibly detailed, 00:08:35.740 --> 00:08:37.960 useful for filmmakers, researchers. Absolutely. 00:08:38.200 --> 00:08:40.399 Anyone needing to dissect visual content. That's 00:08:40.399 --> 00:08:43.500 really impressive visual analysis. But what about 00:08:43.500 --> 00:08:46.500 summarizing, say, really long videos? Is there 00:08:46.500 --> 00:08:49.080 a quicker way than having it analyze every single 00:08:49.080 --> 00:08:51.690 frame? Yeah. There's a very clever and efficient 00:08:51.690 --> 00:08:53.970 workaround for that, especially for things like 00:08:53.970 --> 00:08:56.070 long interviews or lectures. OK, how does that 00:08:56.070 --> 00:08:57.590 work? It's pretty straightforward. You go to 00:08:57.590 --> 00:08:59.230 the YouTube video, click the little three dots 00:08:59.230 --> 00:09:02.470 menu, select show transcript, and you just copy 00:09:02.470 --> 00:09:04.690 that whole transcript, paste it into Gemini and 00:09:04.690 --> 00:09:07.809 ask for a summary. Ah, clever. Exactly. So for 00:09:07.809 --> 00:09:09.549 like a two hour interview, you could ask for 00:09:09.549 --> 00:09:11.769 bullet points on the main arguments and any key 00:09:11.769 --> 00:09:14.210 stories the core questions discussed saves you 00:09:14.210 --> 00:09:16.470 hours of listening and note taking. That's a 00:09:16.470 --> 00:09:18.429 great practical tip. Now, moving on to data. 00:09:18.549 --> 00:09:21.330 Raw numbers, spreadsheets, they can be hard to 00:09:21.330 --> 00:09:23.669 grasp, right? Definitely. Just staring at rows 00:09:23.669 --> 00:09:25.889 of data isn't very intuitive for most people. 00:09:26.370 --> 00:09:29.169 So Gemini can act like a data analyst, turning 00:09:29.169 --> 00:09:32.049 numbers into charts and maps. It can. And this 00:09:32.049 --> 00:09:34.750 used to require coding skills, knowing libraries 00:09:34.750 --> 00:09:37.909 like mapplotlib or things like that. Now, you 00:09:37.909 --> 00:09:39.830 can just ask in plain English. Like creating 00:09:39.830 --> 00:09:42.230 interactive maps. Yeah, it can write the code 00:09:42.230 --> 00:09:45.409 HTML, JavaScript, usually for an interactive 00:09:45.409 --> 00:09:48.639 map. You could ask for, say, a world map. colored 00:09:48.639 --> 00:09:51.139 by coffee, export volume, maybe light green to 00:09:51.139 --> 00:09:53.700 dark brown. And when you hover over a country, 00:09:53.840 --> 00:09:56.240 it shows the name and the volume. That's fantastic 00:09:56.240 --> 00:09:58.340 for presentations and charts too. All sorts. 00:09:58.559 --> 00:10:00.899 Bar charts, pie charts, line charts to show trends 00:10:00.899 --> 00:10:03.559 over time. For example? You could ask it to create 00:10:03.559 --> 00:10:06.019 a line chart showing the projected growth of 00:10:06.019 --> 00:10:08.460 the electric vehicle market in, say, Vietnam, 00:10:08.720 --> 00:10:11.840 Thailand, and Indonesia from 2020 to 2025. And 00:10:11.840 --> 00:10:14.039 does it just give you an image? Often it gives 00:10:14.039 --> 00:10:16.539 you the image, and it can also provide the source 00:10:16.539 --> 00:10:19.659 code, maybe in Python, using Plotly or Matplotlib. 00:10:19.820 --> 00:10:21.600 So if you do have some coding skills, you can 00:10:21.600 --> 00:10:24.279 tweak it further or integrate it elsewhere. Nice. 00:10:24.779 --> 00:10:27.519 OK, let's unpack this next one. Because you suggested 00:10:27.519 --> 00:10:29.460 this might be one of the most powerful features 00:10:29.460 --> 00:10:33.259 in the whole Gemini ecosystem, turning it from 00:10:33.259 --> 00:10:36.179 just a chatbot into a real research assistant. 00:10:36.519 --> 00:10:39.139 I think so, yeah. The deep research capability 00:10:39.139 --> 00:10:41.720 is a significant step up from just getting a 00:10:41.720 --> 00:10:44.039 quick summary or search results. It functions 00:10:44.039 --> 00:10:46.220 more like an AI assistant doing actual research 00:10:46.220 --> 00:10:48.620 for you. How does it work? Is it just a different 00:10:48.620 --> 00:10:51.200 prompt? It's a specific feature, often in the 00:10:51.200 --> 00:10:53.179 more advanced versions, maybe with limited free 00:10:53.179 --> 00:10:55.679 access sometimes, when you activate it with a 00:10:55.679 --> 00:10:57.740 complex research question. OK. It doesn't just 00:10:57.740 --> 00:11:01.450 spit out an answer. First, it proposes a research 00:11:01.450 --> 00:11:04.409 plan. A plan. Yeah. So if you ask it to analyze, 00:11:04.950 --> 00:11:07.429 say, why international tourists choose Vietnam 00:11:07.429 --> 00:11:10.309 post -pandemic, it might suggest, OK, I'll analyze 00:11:10.309 --> 00:11:13.009 official tourism reports, synthesize online traveler 00:11:13.009 --> 00:11:15.690 reviews, search academic papers on tourism trends, 00:11:15.809 --> 00:11:18.289 and compare Vietnam with nearby countries. And 00:11:18.289 --> 00:11:20.590 you approve the plan. Exactly. Once you approve 00:11:20.590 --> 00:11:22.889 it, the AI goes out, browses tons of sources, 00:11:23.470 --> 00:11:25.509 weeds out duplicates, identifies key themes, 00:11:25.889 --> 00:11:28.429 and then it synthesizes all of that. And the 00:11:28.429 --> 00:11:30.789 final output isn't just a paragraph? No, no. 00:11:30.870 --> 00:11:34.210 The end result is typically a multi -page structured 00:11:34.210 --> 00:11:36.789 report, something you can export to Google Docs, 00:11:36.830 --> 00:11:39.570 properly formatted. So for that Vietnam tourism 00:11:39.570 --> 00:11:42.289 example, what might the report look like? You 00:11:42.289 --> 00:11:44.509 could get an executive summary, then sections 00:11:44.509 --> 00:11:47.250 on key factors, culture, food, cost, safety, 00:11:48.029 --> 00:11:50.029 maybe emerging trends like sustainable travel. 00:11:50.250 --> 00:11:53.250 analysis by tourist -type backpackers, families, 00:11:53.629 --> 00:11:57.009 luxury, and crucially, it includes sources and 00:11:57.009 --> 00:12:00.110 citations, proper research methodology. The quality 00:12:00.110 --> 00:12:02.830 of synthesis and the time saved there sounds 00:12:02.830 --> 00:12:06.129 immense, like weeks of human work compressed. 00:12:06.409 --> 00:12:08.210 It really is. Now, it's important to remember 00:12:08.210 --> 00:12:10.570 this deep research feature is often premium, 00:12:10.889 --> 00:12:13.529 maybe with limits on free use, but even trying 00:12:13.529 --> 00:12:16.029 it out shows its incredible potential for big 00:12:16.029 --> 00:12:18.269 academic projects, business strategy, creative 00:12:18.269 --> 00:12:20.720 development. Right. Okay, shifting from heavy 00:12:20.720 --> 00:12:22.980 analysis to more everyday help. How does Gemini 00:12:22.980 --> 00:12:25.720 work as an assistant for like real -time learning 00:12:25.720 --> 00:12:27.559 or tech support? Well, one really interesting 00:12:27.559 --> 00:12:30.759 feature is the AI tutor via screen sharing. This 00:12:30.759 --> 00:12:32.700 has the potential to really change how we learn 00:12:32.700 --> 00:12:35.000 practical skills. Screen sharing. So it sees 00:12:35.000 --> 00:12:37.580 what I'm doing. Exactly. Instead of just watching 00:12:37.580 --> 00:12:39.899 a generic tutorial video, you can be working 00:12:39.899 --> 00:12:42.779 on something, share your screen, and get step 00:12:42.779 --> 00:12:45.159 -by -step guidance on your specific problem. 00:12:45.419 --> 00:12:48.970 So say I'm stuck in Excel with a VLOG up. Perfect 00:12:48.970 --> 00:12:51.669 example. You share your spreadsheet, ask Gemini, 00:12:51.970 --> 00:12:55.230 how do I get this VeloCup to work? It sees your 00:12:55.230 --> 00:12:58.169 data, sees your formula, and guides you right 00:12:58.169 --> 00:13:00.850 there. Contextual help is way more effective. 00:13:01.110 --> 00:13:03.610 That sounds incredibly useful. And its sight 00:13:03.610 --> 00:13:06.190 isn't just limited to screens, right? You mentioned 00:13:06.190 --> 00:13:08.850 object analysis via camera. Yeah, this turns 00:13:08.850 --> 00:13:11.870 your phone camera into like an intelligent eye 00:13:11.870 --> 00:13:13.529 that interacts with the real world. What kind 00:13:13.529 --> 00:13:16.000 of things can you do with that? Oh, loads. Point 00:13:16.000 --> 00:13:18.960 it at your kid's handwritten math homework. It 00:13:18.960 --> 00:13:20.980 can help solve the problem. Point it at a menu 00:13:20.980 --> 00:13:23.279 in a foreign language. Get an instant translation. 00:13:23.720 --> 00:13:25.279 See a plant you don't recognize in your garden. 00:13:25.500 --> 00:13:27.600 Ask Gemini what it is and how to care for it. 00:13:27.980 --> 00:13:30.100 Or point it at an old building and ask about 00:13:30.100 --> 00:13:32.759 its history. So it brings AI smarts into your 00:13:32.759 --> 00:13:35.740 physical surroundings. Pretty much. Instant information 00:13:35.740 --> 00:13:37.919 about the world around you. Now for developers 00:13:37.919 --> 00:13:41.070 or even people who dabble in code. How helpful 00:13:41.070 --> 00:13:44.149 is it? Hugely helpful. Debugging and fixing code 00:13:44.149 --> 00:13:47.190 is a massive time saver. We all know how frustrating 00:13:47.190 --> 00:13:50.629 finding that one bug can be. You can paste your 00:13:50.629 --> 00:13:53.690 broken code snippet, say some Python, with a 00:13:53.690 --> 00:13:56.669 list index out of Ranger. Yeah, classic. So Gemini 00:13:56.669 --> 00:13:59.250 can often find the bug in seconds, give you the 00:13:59.250 --> 00:14:01.509 corrected code, and explain why it was wrong. 00:14:01.909 --> 00:14:05.169 Like, you used range plus one, which goes one 00:14:05.169 --> 00:14:07.870 step too far. It should just be range. That explanation 00:14:07.870 --> 00:14:10.330 helps you learn. That's brilliant. And what about 00:14:10.330 --> 00:14:12.570 for non -programmers? Can it help automate tasks? 00:14:12.929 --> 00:14:15.629 Absolutely. Writing automation scripts is surprisingly 00:14:15.629 --> 00:14:18.230 accessible. You can describe a repetitive task 00:14:18.230 --> 00:14:20.789 you do, and Gemini can write a script, often 00:14:20.789 --> 00:14:23.250 in Python, to automate it. Like, what sort of 00:14:23.250 --> 00:14:25.610 task? Okay, imagine you regularly have to go 00:14:25.610 --> 00:14:28.570 through a folder of monthly report CSV files, 00:14:29.110 --> 00:14:31.149 pull out all the unique email addresses from 00:14:31.149 --> 00:14:33.610 a specific column, and put them into one master 00:14:33.610 --> 00:14:36.679 list. Yeah, tedious. You can ask Gemini. Write 00:14:36.679 --> 00:14:38.779 a Python script that opens my monthly reports 00:14:38.779 --> 00:14:42.580 folder, reads every CSV file, finds the email 00:14:42.580 --> 00:14:45.139 column, extracts all the unique emails, and saves 00:14:45.139 --> 00:14:49.820 them to a file called emaillist .txt. It generates 00:14:49.820 --> 00:14:52.559 a script, you run it, saves you hours. Minimizes 00:14:52.559 --> 00:14:54.960 errors too, I bet. Big time. Okay, let's talk 00:14:54.960 --> 00:14:59.039 communication. Good writing, clear emails. That's 00:14:59.039 --> 00:15:00.879 crucial professionally. Can Gemini help there? 00:15:00.940 --> 00:15:02.779 Definitely. It's a very strong assistant for 00:15:02.779 --> 00:15:04.620 drafting all sorts of content, especially things 00:15:04.620 --> 00:15:06.759 like marketing materials. So drafting marketing 00:15:06.759 --> 00:15:08.600 emails? Yeah, it can help you come up with catchy 00:15:08.600 --> 00:15:10.799 subject lines, write persuasive body copy, figure 00:15:10.799 --> 00:15:13.519 out a clear call to action, all tailored to your 00:15:13.519 --> 00:15:15.460 audience. Give me an example. Let's say you run 00:15:15.460 --> 00:15:18.179 the local beans coffee shop. You want to promote 00:15:18.179 --> 00:15:20.940 a new, organic cold brew coffee, you tell Gemini. 00:15:21.320 --> 00:15:23.799 Draft a marketing email targeting busy, health 00:15:23.799 --> 00:15:27.620 -conscious office workers 25 -35 years old. Emphasize 00:15:27.620 --> 00:15:30.299 convenience, great taste, health perks. Need 00:15:30.299 --> 00:15:32.480 a catchy subject line and a call to action with 00:15:32.480 --> 00:15:35.440 the discount code COLDBRU20. It'll generate a 00:15:35.440 --> 00:15:37.840 draft email hitting all those points. That saves 00:15:37.840 --> 00:15:41.500 a lot of brainstorming time. And what about translation? 00:15:41.820 --> 00:15:44.279 You mentioned it goes beyond basic machine translation. 00:15:44.539 --> 00:15:47.039 Yes, this is really fascinating. Translation 00:15:47.039 --> 00:15:49.679 and cultural adaptation. It's not just swapping 00:15:49.679 --> 00:15:53.100 words. How so? Gemini tries to understand and 00:15:53.100 --> 00:15:57.120 convey the nuance, the tone, the cultural context. 00:15:57.799 --> 00:16:00.259 Literal translations can often sound weird or 00:16:00.259 --> 00:16:03.240 miss the mark entirely. Right. So take a Vietnamese 00:16:03.240 --> 00:16:07.240 advertising slogan like tin hồ quà vịt. A literal 00:16:07.240 --> 00:16:10.529 translation might be awkward. Gemini could offer 00:16:10.529 --> 00:16:12.870 options that capture the feeling for an English 00:16:12.870 --> 00:16:15.409 -speaking audience, like the essence of Vietnamese 00:16:15.409 --> 00:16:18.090 gifting, or maybe a taste of Vietnamese heritage 00:16:18.090 --> 00:16:20.950 perfectly gifted, or the finest gifts crafted 00:16:20.950 --> 00:16:23.490 in Vietnam. So it ensures the message actually 00:16:23.490 --> 00:16:26.590 resonates culturally, not just translated. Exactly. 00:16:26.730 --> 00:16:29.710 Trying to avoid that lost in translation problem. 00:16:30.009 --> 00:16:31.850 Okay. Now this is where it gets really powerful, 00:16:32.230 --> 00:16:34.490 right? Combining these individual features into 00:16:34.490 --> 00:16:37.470 like complete workflows and maybe some hitting 00:16:37.470 --> 00:16:39.950 gems. Absolutely. The real magic often happens 00:16:39.950 --> 00:16:41.450 when you start stringing these tools together. 00:16:41.830 --> 00:16:43.809 The synergy is incredible. It's where one plus 00:16:43.809 --> 00:16:45.830 one definitely equals more than two. So walk 00:16:45.830 --> 00:16:49.519 me through an example. Let's say A big project, 00:16:49.580 --> 00:16:53.299 like creating a whole content package about climate 00:16:53.299 --> 00:16:55.440 change impacts on the Mekong Delta. That's a 00:16:55.440 --> 00:16:57.659 complex topic. OK, great example. Let's break 00:16:57.659 --> 00:16:59.580 down how you could use Gemini for that. Step 00:16:59.580 --> 00:17:02.960 one would be research, I guess, using that deep 00:17:02.960 --> 00:17:05.640 research feature. Precisely. Step one, in -depth 00:17:05.640 --> 00:17:07.940 research. You'd prompt it for a detailed study 00:17:07.940 --> 00:17:10.700 on saltwater intrusion, land subsidence in the 00:17:10.700 --> 00:17:13.559 Mekong, asking for data, reports, proposed solutions. 00:17:13.900 --> 00:17:16.079 Get that foundational knowledge. OK, got the 00:17:16.079 --> 00:17:19.180 research. Step two. Step two. Blog content writing. 00:17:19.680 --> 00:17:22.279 Take that research output and ask Gemini to write, 00:17:22.480 --> 00:17:24.960 say, a 1 ,500 -word blog post. Give it a title 00:17:24.960 --> 00:17:28.339 like, The Mekong Delta's Cry for Help. Act now 00:17:28.339 --> 00:17:29.940 before it's too late. Tell it you want a clear 00:17:29.940 --> 00:17:32.099 structure, persuasive tone, call to action. All 00:17:32.099 --> 00:17:33.920 right. Turn the research into an article, then. 00:17:34.079 --> 00:17:36.920 Visuals. Exactly. Step three, data visualization. 00:17:37.259 --> 00:17:39.940 Use the data from the research. Ask Gemini. Create 00:17:39.940 --> 00:17:42.180 an interactive map of Mekong provinces colored 00:17:42.180 --> 00:17:44.980 by saltwater intrusion levels. And maybe create 00:17:44.980 --> 00:17:47.079 a line chart showing sea level rise there over 00:17:47.079 --> 00:17:49.480 the past 10 years. Adding data evidence. Good. 00:17:49.579 --> 00:17:52.640 What about a main image? Step four, thematic 00:17:52.640 --> 00:17:56.000 image creation. Use the text to image feature. 00:17:56.440 --> 00:17:58.720 Prompt something symbolic. Generate an image 00:17:58.720 --> 00:18:01.420 of a cracked dry rice field next to a rising 00:18:01.420 --> 00:18:05.099 river under a gloomy ominous sky. something powerful 00:18:05.099 --> 00:18:07.380 to grab attention. Okay, article, data, viz, 00:18:07.480 --> 00:18:11.339 image, what else? Audio, video. Yep. Step five, 00:18:11.599 --> 00:18:13.980 audio and video. Upload that blog post, text 00:18:13.980 --> 00:18:16.279 to Notebook LM, and ask it to create a five -minute 00:18:16.279 --> 00:18:19.460 podcast discussion between two AI hosts debating 00:18:19.460 --> 00:18:21.779 the issue. Wow. And for video, you could use 00:18:21.779 --> 00:18:24.380 that perplexity trick on Expo we mentioned, feeding 00:18:24.380 --> 00:18:26.559 it a prompt based on your research to generate 00:18:26.559 --> 00:18:29.599 maybe a short, dramatic flycam -style video showing 00:18:29.599 --> 00:18:32.200 the transition from lush delta to dry land. So 00:18:32.200 --> 00:18:34.720 just to recap that workflow. One person using 00:18:34.720 --> 00:18:37.160 Gemini can generate in -depth research, a long 00:18:37.160 --> 00:18:39.200 -form article, interactive data maps and charts, 00:18:39.460 --> 00:18:41.900 a thematic image, a podcast episode, and a short 00:18:41.900 --> 00:18:44.579 video. Exactly. A complete multimedia package 00:18:44.579 --> 00:18:47.019 on a complex topic. That's the power of the ecosystem 00:18:47.019 --> 00:18:49.440 working together. It really changes what individuals 00:18:49.440 --> 00:18:51.940 or small teams can produce. Mind -blowing. OK. 00:18:51.940 --> 00:18:53.700 Besides these big workflows, are there other 00:18:53.700 --> 00:18:56.220 maybe quicker, cool applications, like a rapid 00:18:56.220 --> 00:18:58.460 -fire round of power tips? Sure. There are tons 00:18:58.460 --> 00:19:01.539 of smaller but super useful things, like creating 00:19:01.539 --> 00:19:04.460 those free videos via perplex— Veo, tweet at 00:19:04.460 --> 00:19:07.140 ask perplexity with a prompt like, create a video 00:19:07.140 --> 00:19:09.759 of a small robot watering a plant growing from 00:19:09.759 --> 00:19:13.339 an old book, stop motion animation style. Gemini 00:19:13.339 --> 00:19:15.859 helps you craft the perfect process. Nice. Designing 00:19:15.859 --> 00:19:18.359 infographics. Give it data like, here are the 00:19:18.359 --> 00:19:20.500 six steps for proper hand washing from the Ministry 00:19:20.500 --> 00:19:23.299 of Health. Design a simple, colorful infographic 00:19:23.299 --> 00:19:25.980 for kids. Makes info visual and easy to grasp. 00:19:26.299 --> 00:19:29.099 Handy. Sketching logo ideas. Brainstorm visually. 00:19:29.680 --> 00:19:32.400 Sketch five modern minimalist logo ideas for 00:19:32.400 --> 00:19:34.920 Saigon Brewed Coffee. Hinting at Saigon imagery 00:19:34.920 --> 00:19:37.299 gets the creative juices flowing fast. Fast. 00:19:37.440 --> 00:19:39.960 Writing video scripts. Quick turnaround on concepts. 00:19:40.200 --> 00:19:41.880 Write a one -minute ad script on data backup. 00:19:46.499 --> 00:20:13.940 Very practical. Sorry. your personal tutor. Explain 00:20:13.940 --> 00:20:16.619 blockchain like I'm five using a village ledger 00:20:16.619 --> 00:20:19.460 analogy. Breaks down jargon. I need that one. 00:20:19.680 --> 00:20:21.539 And finally, this is a cool one. Role -playing 00:20:21.539 --> 00:20:24.519 for soft skills. Practice makes perfect. Act 00:20:24.519 --> 00:20:26.819 as a tough tech recruiter. I'm interviewing for 00:20:26.819 --> 00:20:29.059 product manager. Ask me behavioral questions. 00:20:29.700 --> 00:20:32.279 Safe space to rehearse. That's actually brilliant 00:20:32.279 --> 00:20:34.680 for interview prep. Okay, so we've seen a ton 00:20:34.680 --> 00:20:37.250 of features. Where does all this place Gemini 00:20:37.250 --> 00:20:39.730 in the bigger AI picture? How does it stack up 00:20:39.730 --> 00:20:43.150 against, say, chat GPT or quad? And what seems 00:20:43.150 --> 00:20:45.569 to be Google's overall game plan here? Right. 00:20:45.769 --> 00:20:47.750 Context is key. In this sort of three -way race 00:20:47.750 --> 00:20:49.869 among the big language models, each has its niche. 00:20:50.549 --> 00:20:52.910 Gemini's position is really defined by balancing 00:20:52.910 --> 00:20:57.049 serious power with remarkable accessibility, 00:20:57.369 --> 00:20:59.349 mostly through being free. OK. So compared to 00:20:59.349 --> 00:21:02.150 chat GPT, chat GPT was first, has a huge community. 00:21:02.329 --> 00:21:05.289 True, ChatGPT has that first mover advantage 00:21:05.289 --> 00:21:08.950 and a massive user base. But many of its really 00:21:08.950 --> 00:21:12.670 advanced features, the latest models, Dell E3 00:21:12.670 --> 00:21:16.029 image creation, the slick data analysis, often 00:21:16.029 --> 00:21:19.269 sit behind the ChatGPT Plus paywall. Gemini's 00:21:19.269 --> 00:21:21.690 Edge is offering a lot of that multimedia capability, 00:21:21.869 --> 00:21:24.950 images, video analysis, even the coding stuff 00:21:24.950 --> 00:21:28.269 for free. It levels the playing field significantly. 00:21:28.549 --> 00:21:30.430 And versus Claude. Claude's known for handling 00:21:30.430 --> 00:21:32.529 huge amounts of text, right? Exactly. Claude, 00:21:32.609 --> 00:21:34.829 from Anthropic, its strengths are that massive 00:21:34.829 --> 00:21:37.230 context window, it can read huge documents, and 00:21:37.230 --> 00:21:39.630 it's really sharp reasoning and summarization 00:21:39.630 --> 00:21:42.990 skills. Great for deep text analysis. But its 00:21:42.990 --> 00:21:45.529 weaponess is multimedia. It doesn't really do 00:21:45.529 --> 00:21:48.109 images, video, code generation, game creation. 00:21:48.859 --> 00:21:51.599 That's where Gemini pulls way ahead. So Gemini's 00:21:51.599 --> 00:21:53.700 sweet spot is that free all -in -one package, 00:21:53.839 --> 00:21:56.180 especially if you need to work across text, visuals, 00:21:56.319 --> 00:21:58.299 and maybe even code or audio. That's a great 00:21:58.299 --> 00:22:00.319 way to put it. It's the versatility within that 00:22:00.319 --> 00:22:02.180 free ecosystem that makes it stand out for many 00:22:02.180 --> 00:22:04.240 tasks. OK, so if people want to get the most 00:22:04.240 --> 00:22:06.220 out of Gemini, are there some key rules or tips? 00:22:06.720 --> 00:22:09.180 Absolutely. Owning the tool is one thing. Using 00:22:09.180 --> 00:22:11.859 it well is another. I'd say three golden rules. 00:22:12.200 --> 00:22:14.900 Rule number one, master the art of prompting. 00:22:14.970 --> 00:22:17.289 Okay. A good prompt isn't just a question, it's 00:22:17.289 --> 00:22:20.329 a detailed instruction. Use the context task 00:22:20.329 --> 00:22:23.089 format formula. Context. I'm a small coffee shop 00:22:23.089 --> 00:22:26.289 owner. Task. Write a Facebook post promoting 00:22:26.289 --> 00:22:29.490 my new roasted oolong milk tea format. Make it 00:22:29.490 --> 00:22:32.130 short, use a youthful tone, include three relevant 00:22:32.130 --> 00:22:34.650 hashtags, and end with an engaging question. 00:22:34.930 --> 00:22:37.589 More detail equals better results. Precisely. 00:22:37.849 --> 00:22:40.900 Rule two. Think iteratively and refine. Don't 00:22:40.900 --> 00:22:42.779 expect perfection first try. Right. Treat it 00:22:42.779 --> 00:22:44.660 like a draft. Exactly. See, the first output 00:22:44.660 --> 00:22:47.240 is draft one, then talk back to it. Make the 00:22:47.240 --> 00:22:49.480 tone more professional. Can you add another example? 00:22:49.819 --> 00:22:51.579 That sentence sounds a bit clunky. Rewrite it. 00:22:51.740 --> 00:22:54.059 Use the conversational aspect to polish it. Makes 00:22:54.059 --> 00:22:56.400 sense. And rule three. Rule three. Understand 00:22:56.400 --> 00:23:00.640 its limitations. It powerful, but it's still 00:23:00.640 --> 00:23:04.680 a machine. Meaning, always verify critical information. 00:23:05.079 --> 00:23:08.549 AI can hallucinate. make stuff up, so fact check 00:23:08.549 --> 00:23:11.549 in court and details. Also, be aware of token 00:23:11.549 --> 00:23:14.069 limits. You might need to break down super long 00:23:14.069 --> 00:23:17.109 documents and know that some super advanced features, 00:23:17.390 --> 00:23:19.809 like deep research, might have free usage caps. 00:23:20.089 --> 00:23:23.549 Just be realistic. Good advice. So this strategy 00:23:23.549 --> 00:23:25.789 from Google offering so much advanced stuff for 00:23:25.789 --> 00:23:28.589 free, what is that signal about their long -term 00:23:28.589 --> 00:23:31.609 vision? It points to a very distinct AI for everyone 00:23:31.609 --> 00:23:34.710 philosophy, I think. Google seems to want AI 00:23:34.710 --> 00:23:37.869 to become just. Ambient, an indispensable part 00:23:37.869 --> 00:23:40.049 of daily life for billions. How so? By weaving 00:23:40.049 --> 00:23:42.630 Gemini deeply into everything. Search, Android, 00:23:42.789 --> 00:23:45.250 Chrome, Google Workspace. The goal seems to be 00:23:45.250 --> 00:23:47.190 an ambient assistant that's always there, subtly 00:23:47.190 --> 00:23:49.369 helping, maybe even anticipating what you need. 00:23:49.630 --> 00:23:52.210 And making it free helps drive adoption. Exactly. 00:23:52.450 --> 00:23:54.829 And that massive adoption feeds them unbelievable 00:23:54.829 --> 00:23:57.109 amounts of usage data, which lets them improve 00:23:57.109 --> 00:23:59.210 the models incredibly quickly. It's a virtuous 00:23:59.210 --> 00:24:01.410 cycle for them. So looking ahead, what can we 00:24:01.410 --> 00:24:04.920 expect? Deeper integration. Definitely. Expect 00:24:04.920 --> 00:24:07.380 Gemini to soon be doing things like reading your 00:24:07.380 --> 00:24:09.779 emails in Gmail and drafting replies based on 00:24:09.779 --> 00:24:12.240 your style, or creating presentations and slides 00:24:12.240 --> 00:24:14.920 from notes in Keep, or automatically summarizing 00:24:14.920 --> 00:24:17.200 your Google Meet calls without you even asking. 00:24:17.500 --> 00:24:20.140 Seamless integration, what else? More autonomous 00:24:20.140 --> 00:24:24.140 agents. Deep research is just the start. Imagine 00:24:24.140 --> 00:24:27.119 future agents tackling complex multi -step tasks 00:24:27.119 --> 00:24:29.700 like find and book the cheapest flight to Da 00:24:29.700 --> 00:24:32.259 Nang for next weekend plus a four -star hotel 00:24:32.259 --> 00:24:34.480 near the beach with good reviews and put it all 00:24:34.480 --> 00:24:36.880 on my calendar. Whoa handling the whole process. 00:24:37.059 --> 00:24:40.180 Yeah and finally hyper -personalization. The 00:24:40.180 --> 00:24:42.359 AI will learn your habits, your writing style, 00:24:42.519 --> 00:24:44.740 your common workflows to offer assistance that 00:24:44.740 --> 00:24:47.180 feels completely tailor -made. It becomes your 00:24:47.180 --> 00:24:49.359 assistant, not just an assistant. That's quite 00:24:49.359 --> 00:24:51.579 a future unfolding. OK, so we've really covered 00:24:51.579 --> 00:24:54.539 a lot today. We journeyed through, what, 28 incredible 00:24:54.539 --> 00:24:57.039 features. Showing Gemini isn't just a chatbot. 00:24:57.220 --> 00:24:59.960 Not at all. It's a creative studio, a data analyst, 00:25:00.119 --> 00:25:02.440 a coding buddy, a personal tutor, all wrapped 00:25:02.440 --> 00:25:05.259 up in this free, accessible package. And the 00:25:05.259 --> 00:25:08.299 main takeaway, really, is that this age of AI, 00:25:08.559 --> 00:25:10.539 it's not some far -off thing or just for tech 00:25:10.539 --> 00:25:13.759 elites anymore. It's here. It's within your reach 00:25:13.759 --> 00:25:17.740 right now. The biggest barrier isn't cost. It's 00:25:17.740 --> 00:25:19.940 probably just curiosity and being willing to 00:25:19.940 --> 00:25:21.940 play around with it. Whether you're a student, 00:25:22.200 --> 00:25:25.440 a business owner, an artist, it can genuinely 00:25:25.440 --> 00:25:28.359 become a powerful ally. So the best advice is 00:25:28.359 --> 00:25:31.640 just to start using it. Absolutely. That's the 00:25:31.640 --> 00:25:34.099 best way to feel its power. Try something small 00:25:34.099 --> 00:25:37.160 today. Ask it to write a poem. Create a silly 00:25:37.160 --> 00:25:39.700 image. Plan your weekend. Explain something you've 00:25:39.700 --> 00:25:41.519 always wondered about. You'll be amazed. You 00:25:41.519 --> 00:25:43.339 really will be amazed at what you can achieve. 00:25:43.660 --> 00:25:46.599 The future of creativity. Of productivity. Yeah. 00:25:46.819 --> 00:25:49.720 What's happening now. And with these free superpowers, 00:25:50.240 --> 00:25:52.640 really the only limit is your own imagination. 00:25:53.359 --> 00:25:55.259 Which leaves us with the final question for you, 00:25:55.259 --> 00:25:57.559 the listener. What will you create next?