WEBVTT 00:00:00.000 --> 00:00:04.160 Imagine you need a digital tool, maybe a custom 00:00:04.160 --> 00:00:07.099 flashcard app for learning, or a really specific 00:00:07.099 --> 00:00:10.220 calculator for your work. Or even like a simulation. 00:00:11.140 --> 00:00:13.419 A little world with rules like an ecosystem. 00:00:13.640 --> 00:00:15.939 And instead of coding, you just describe it in 00:00:15.939 --> 00:00:19.039 plain English. You hit enter. And bam, seconds 00:00:19.039 --> 00:00:21.899 later, it's built. It works right there on your 00:00:21.899 --> 00:00:24.760 screen. And here's the really wild part. You 00:00:24.760 --> 00:00:27.379 don't write any code, and it costs you nothing. 00:00:27.660 --> 00:00:32.600 $0 welcome everyone to the deep dive and Yeah, 00:00:32.679 --> 00:00:35.060 that capability. It's not science fiction. It's 00:00:35.060 --> 00:00:37.219 actually hiding in plain sight inside Google 00:00:37.219 --> 00:00:39.619 Gemini right now It feels like it fundamentally 00:00:39.619 --> 00:00:41.979 changes what building something digital even 00:00:41.979 --> 00:00:44.240 means it really does It's turning text prompts 00:00:44.240 --> 00:00:46.780 directly into working software instantly. Okay, 00:00:46.780 --> 00:00:48.420 so let's really dig into this We want to make 00:00:48.420 --> 00:00:50.979 sure you listening right now can walk away and 00:00:50.979 --> 00:00:52.920 actually use this thing Our mission today is 00:00:52.920 --> 00:00:55.140 pretty clear. We're handing you the keys, right? 00:00:55.200 --> 00:00:57.200 So first, we got to understand what this actually 00:00:57.200 --> 00:00:59.560 is. How is it different, maybe more powerful 00:00:59.560 --> 00:01:02.020 than just a regular chat bot spitting out code? 00:01:03.100 --> 00:01:05.319 Then we'll walk through four examples. And these 00:01:05.319 --> 00:01:07.239 aren't trivial things. They're genuinely impressive. 00:01:07.719 --> 00:01:11.420 Learning tools, simulations with memory, real 00:01:11.420 --> 00:01:14.079 proof. Yeah, got to show it works. And finally, 00:01:14.400 --> 00:01:17.579 the main event, the golden formula. This is the 00:01:17.579 --> 00:01:20.280 step -by -step way to write prompts that actually 00:01:20.280 --> 00:01:22.799 get you useful working tools. You could build 00:01:22.799 --> 00:01:24.689 your first one today. Okay, let's start with 00:01:24.689 --> 00:01:27.150 that core idea. When we say Gemini builds a tool, 00:01:27.590 --> 00:01:29.689 we're not talking about it just giving you, like, 00:01:29.890 --> 00:01:31.769 a chunk of Python code you have to figure out 00:01:31.769 --> 00:01:33.909 how to run. Exactly. That's the old way. This 00:01:33.909 --> 00:01:35.969 is different. It builds the complete application. 00:01:36.349 --> 00:01:38.709 User interface, buttons, everything. Right there 00:01:38.709 --> 00:01:40.750 in the chat window, it is the final product. 00:01:41.069 --> 00:01:43.359 Ready to go. Oh, how? What's the magic trick 00:01:43.359 --> 00:01:45.159 under the hood? What's technically different? 00:01:45.400 --> 00:01:47.920 OK, so this is the cool part. Instead of just 00:01:47.920 --> 00:01:50.680 generating code text, Gemini generates the actual 00:01:50.680 --> 00:01:53.680 web stuff, you know, HTML for structure, CSS 00:01:53.680 --> 00:01:56.560 for looks, JavaScript for the logic. Right. But 00:01:56.560 --> 00:01:59.040 then, and this is the key, it actually runs that 00:01:59.040 --> 00:02:02.420 code. It executes it and renders the live application 00:02:02.420 --> 00:02:05.269 right there inside the Gemini interface. Ah, 00:02:05.489 --> 00:02:08.250 OK. So it's not just the architect handing you 00:02:08.250 --> 00:02:10.550 blueprints. It's the architect, the construction 00:02:10.550 --> 00:02:12.870 crew, and the finished building all presented 00:02:12.870 --> 00:02:15.530 at once. Perfect analogy. Yeah. Think of it like 00:02:15.530 --> 00:02:19.030 having this super fast, perfectly coordinated 00:02:19.030 --> 00:02:22.430 team of programmers on standby. You give them 00:02:22.430 --> 00:02:26.090 the idea. They instantly translate it, build 00:02:26.090 --> 00:02:28.110 it. and show you the working thing. It's like 00:02:28.110 --> 00:02:30.370 an execution sandbox built right into the chat. 00:02:31.069 --> 00:02:33.110 So, practically speaking, what does this unlock 00:02:33.110 --> 00:02:35.310 for people listening? Where can you use this 00:02:35.310 --> 00:02:38.150 power? Well, it spans a few key areas, I think. 00:02:38.610 --> 00:02:41.129 For learning, it's huge. Imagine creating your 00:02:41.129 --> 00:02:44.250 own quiz games from lecture notes or specific 00:02:44.250 --> 00:02:46.509 flashcard sets, maybe even little simulators 00:02:46.509 --> 00:02:48.949 for, like, physics concepts. Then there's work. 00:02:49.330 --> 00:02:51.330 The speed boost here is kind of nuts. You need 00:02:51.330 --> 00:02:53.689 a quick calculator for pricing quotes based on, 00:02:53.689 --> 00:02:56.219 like, five different inputs. Done. Or a simple 00:02:56.219 --> 00:02:57.979 project board just for your team's weird way 00:02:57.979 --> 00:03:00.000 of doing things. You can build it. Right. Tailored 00:03:00.000 --> 00:03:02.740 tools without waiting for IT. Exactly. And, you 00:03:02.740 --> 00:03:06.060 know, just for fun and creativity. Building little 00:03:06.060 --> 00:03:09.919 text games or maybe a random story prompt generator 00:03:09.919 --> 00:03:13.139 if you're a writer. Custom digital toys basically 00:03:13.139 --> 00:03:15.780 made in minutes. So fundamentally, how does this 00:03:15.780 --> 00:03:18.539 shift the barrier to entry for creating digital 00:03:18.539 --> 00:03:21.479 things? It makes building apps free and incredibly 00:03:21.479 --> 00:03:24.580 fast for anyone with an idea. OK, theory is one 00:03:24.580 --> 00:03:26.460 thing. Let's get into the proof. We need to see 00:03:26.460 --> 00:03:28.780 the complexity it can handle. What was test number 00:03:28.780 --> 00:03:32.539 one? Test one, we aimed for a fairly complex 00:03:32.539 --> 00:03:35.939 English vocabulary learning tool. We were pretty 00:03:35.939 --> 00:03:37.479 demanding with the features. Yeah, we didn't 00:03:37.479 --> 00:03:40.919 just ask for flashcards. We specified two input 00:03:40.919 --> 00:03:42.939 boxes, right? One for English, one for another 00:03:42.939 --> 00:03:45.979 language, say Vietnamese. And a working add card 00:03:45.979 --> 00:03:48.240 button that actually saved the pairs locally 00:03:48.240 --> 00:03:50.039 in the browser session. The cards needed to flip 00:03:50.039 --> 00:03:52.000 smoothly when you click them. And the tricky 00:03:52.000 --> 00:03:55.590 part? a next card button, but it couldn't just 00:03:55.590 --> 00:03:57.810 go sequentially. It had to pull a random word 00:03:57.810 --> 00:04:00.689 from the deck each time. That requires some real 00:04:00.689 --> 00:04:03.430 logic behind the scenes. Yeah, state management 00:04:03.430 --> 00:04:06.250 and randomization. And you know what? It nailed 00:04:06.250 --> 00:04:09.569 it. Clean interface, nice soft blue theme we 00:04:09.569 --> 00:04:12.030 asked for, flipping worked, random button worked. 00:04:12.270 --> 00:04:14.050 It wasn't just a picture of an app. It was an 00:04:14.050 --> 00:04:17.569 app. Impressive. OK, moving to more utility. 00:04:18.170 --> 00:04:22.399 Test two, the weekly meal planner. This sounds 00:04:22.399 --> 00:04:25.000 practical super practical the first ask was just 00:04:25.000 --> 00:04:28.259 the interface Yeah, an editable table seven columns 00:04:28.259 --> 00:04:31.199 for days three rows for meals breakfast lunch 00:04:31.199 --> 00:04:34.079 dinner Already a bit tricky for layout, but the 00:04:34.079 --> 00:04:36.000 real test was that create shopping list button 00:04:36.000 --> 00:04:38.019 wasn't it? Oh, yeah, this is where it got really 00:04:38.019 --> 00:04:41.420 interesting this needed some intelligence, some 00:04:41.420 --> 00:04:43.899 semantic understanding. So the idea was if you 00:04:43.899 --> 00:04:47.319 type in meals like braised pork with eggs or 00:04:47.319 --> 00:04:49.439 stuffed bitter melon soup. Right, not just random 00:04:49.439 --> 00:04:51.339 text. The tool had to understand, those aren't 00:04:51.339 --> 00:04:53.759 just words. Yeah. It had to figure out the likely 00:04:53.759 --> 00:04:56.300 ingredients needed for those specific recipes. 00:04:56.459 --> 00:04:58.839 And it worked, like surprisingly well. It looked 00:04:58.839 --> 00:05:01.420 at braised pork with eggs, pulled out pork and 00:05:01.420 --> 00:05:04.199 eggs, saw a stuffed bitter melon soup, added 00:05:04.199 --> 00:05:06.839 bitter melon, and presented them as a nice checklist 00:05:06.839 --> 00:05:09.459 with check boxes. It understood the meaning. 00:05:09.560 --> 00:05:11.939 The context, not just shuffling text around. 00:05:12.300 --> 00:05:14.300 It even added an export to text button, which 00:05:14.300 --> 00:05:16.500 was handy. So what does that meal planner intelligence 00:05:16.500 --> 00:05:19.459 really tell us about its capabilities? The AI 00:05:19.459 --> 00:05:22.480 can process semantic context to actually execute 00:05:22.480 --> 00:05:25.040 logical tasks. It's not just following simple 00:05:25.040 --> 00:05:27.259 commands. OK, let's push harder. Complexity. 00:05:27.420 --> 00:05:30.720 Rules, real -time changes. Test 3 was an ecosystem 00:05:30.720 --> 00:05:33.439 simulation. Yeah, we went full Sunant, but simpler. 00:05:33.620 --> 00:05:36.160 We wanted to see if it could handle multivariable 00:05:36.160 --> 00:05:39.699 logic and crucially visualize changes over time 00:05:39.699 --> 00:05:41.759 based on rules we set. What were the rules? Three 00:05:41.759 --> 00:05:45.500 things. Grass, which just grew randomly. Rabbits, 00:05:45.639 --> 00:05:48.300 which ate grass, but if there was no grass they'd 00:05:48.300 --> 00:05:51.439 disappear. And foxes, who ate rabbits and they'd 00:05:51.439 --> 00:05:53.600 disappear if there were no rabbits. Interconnected 00:05:53.600 --> 00:05:56.339 dependencies. Okay, that's already complex logic, 00:05:56.379 --> 00:05:59.379 but you added controls, too. Yes. Three working 00:05:59.379 --> 00:06:01.720 sliders right on the interface. You could adjust 00:06:01.720 --> 00:06:04.420 the starting number of rabbits, foxes, and how 00:06:04.420 --> 00:06:07.079 fast the grass grew. In real time. And it worked! 00:06:07.259 --> 00:06:08.920 You could move a slider and see the simulation 00:06:08.920 --> 00:06:11.360 change. It was actually kind of stunning. You 00:06:11.360 --> 00:06:13.259 got this little box with colored dots moving 00:06:13.259 --> 00:06:15.660 around, obeying the rules. You'd crank up the 00:06:15.660 --> 00:06:17.860 grass growth speed, and suddenly the rabbit population 00:06:17.860 --> 00:06:20.300 would BOOM! Followed by the foxes. You could 00:06:20.300 --> 00:06:22.860 see the dynamics play out. Two secs silence. 00:06:23.579 --> 00:06:26.980 Whoa, yeah, seriously imagine scaling that kind 00:06:26.980 --> 00:06:29.879 of system interconnected rules object states 00:06:29.879 --> 00:06:33.879 Yeah, you could model factory lines traffic flow. 00:06:33.879 --> 00:06:36.040 Yeah, that's more than just displaying data. 00:06:36.040 --> 00:06:39.459 That's modeling a dynamic system Okay, that tested 00:06:39.459 --> 00:06:41.680 simulation. How about memory persistence test 00:06:41.680 --> 00:06:43.939 four was a game a classic text adventure game 00:06:43.939 --> 00:06:46.540 We call it journey to the ancient temple standard 00:06:46.540 --> 00:06:48.839 stuff, but with a key memory requirement. So 00:06:48.839 --> 00:06:51.819 multiple locations, like a village, forest. Yeah, 00:06:51.819 --> 00:06:54.240 we asked for at least five. Village, forest, 00:06:54.399 --> 00:06:57.120 river, cave, temple, the usual setup. But the 00:06:57.120 --> 00:06:59.139 critical test was the memory part, right? An 00:06:59.139 --> 00:07:02.079 inventory system. Exactly. We specified that 00:07:02.079 --> 00:07:04.680 the player had to find a key item, let's say 00:07:04.680 --> 00:07:07.819 in the cave location. Okay. And only after finding 00:07:07.819 --> 00:07:10.339 the key could they open the final temple door 00:07:10.339 --> 00:07:12.839 if they tried to open the temple without the 00:07:12.839 --> 00:07:14.579 key in their virtual pocket. And the game had 00:07:14.579 --> 00:07:17.199 to stop them. Yep, lock them. Give a message 00:07:17.199 --> 00:07:19.360 like, the door is locked. Perhaps you need a 00:07:19.360 --> 00:07:22.100 key. It needed to remember the player's state 00:07:22.100 --> 00:07:25.160 specifically, whether they possess the key object 00:07:25.160 --> 00:07:27.660 or not across different locations and commands. 00:07:27.819 --> 00:07:30.399 And from just a paragraph description. One paragraph. 00:07:30.560 --> 00:07:33.060 And it built a solid little logic game. It understood 00:07:33.060 --> 00:07:35.899 object persistence, state management. Pretty 00:07:35.899 --> 00:07:39.139 cool. OK, so that's simulation test. Is that 00:07:39.139 --> 00:07:41.480 really pushing the limits of what we'd expect 00:07:41.480 --> 00:07:45.019 from just typing text? Yeah, I think so. It confirms 00:07:45.019 --> 00:07:48.399 the AI handles complex nested logic and can even 00:07:48.399 --> 00:07:51.519 visualize changing variables in real time. So 00:07:51.519 --> 00:07:54.079 bringing this back to the real world, if the 00:07:54.079 --> 00:07:57.060 cost to test a digital tool idea basically drops 00:07:57.060 --> 00:08:00.040 to zero, What does that mean for businesses, 00:08:00.240 --> 00:08:02.500 for projects like right now? It means your innovation 00:08:02.500 --> 00:08:05.339 speed just got turbocharged, like 100 times faster 00:08:05.339 --> 00:08:07.279 maybe. You can test an idea, see if customers 00:08:07.279 --> 00:08:09.439 bite in an afternoon, something that might have 00:08:09.439 --> 00:08:11.839 taken weeks in budget before. But hang on, if 00:08:11.839 --> 00:08:14.740 it's this easy, why would anyone learn to code? 00:08:14.920 --> 00:08:17.240 Does this just kill entry -level developer jobs? 00:08:17.620 --> 00:08:20.399 Hmm, good question. I don't think it kills them, 00:08:20.519 --> 00:08:22.839 but it changes them. Like maybe you won't hire 00:08:22.839 --> 00:08:25.379 someone just to build simple landing page calculators 00:08:25.379 --> 00:08:27.600 anymore. Right. Goaters will focus on the really 00:08:27.600 --> 00:08:30.600 complex stuff, big integrations, secure APIs, 00:08:31.300 --> 00:08:34.159 unique high -performance systems. But for maybe 00:08:34.159 --> 00:08:37.440 80 % of the simple tools people need, yeah, this 00:08:37.440 --> 00:08:39.379 changes the game. It frees up human developers 00:08:39.379 --> 00:08:42.000 for harder problems. So think about specific 00:08:42.000 --> 00:08:45.019 industries, e -commerce. What could you build? 00:08:45.220 --> 00:08:48.179 Oh, easily. A custom sizing helper tool based 00:08:48.179 --> 00:08:50.860 on different country charts. Or a quick profit 00:08:50.860 --> 00:08:54.159 margin calculator for marketing. Maybe an interactive 00:08:54.159 --> 00:08:57.360 quiz for lead gen or a catchy headline A -B testing 00:08:57.360 --> 00:08:59.600 tool. Customer service. Interactive troubleshooting 00:08:59.600 --> 00:09:01.980 guides. Simple support ticket forms tailored 00:09:01.980 --> 00:09:04.919 exactly to your product. All doable, like now, 00:09:04.960 --> 00:09:06.700 without needing a dev team budget. Which leads 00:09:06.700 --> 00:09:09.700 to the big question. Why? Why is Google giving 00:09:09.700 --> 00:09:12.740 away this much power for free? It seems generous. 00:09:13.559 --> 00:09:16.460 Well, it's strategic generosity. They're in a 00:09:16.460 --> 00:09:20.299 massive AI race, right? Against OpenAI with ChatGPT, 00:09:20.419 --> 00:09:23.460 Anthropic with Claude, others. Powerful, free 00:09:23.460 --> 00:09:25.639 features like this are a way to attract users 00:09:25.639 --> 00:09:28.899 and developers, build an ecosystem. It's a competitive 00:09:28.899 --> 00:09:31.419 mode. And we, the users, get the benefit of that 00:09:31.419 --> 00:09:33.879 race. Exactly. Now, we should be clear on the 00:09:33.879 --> 00:09:36.000 limits. You're probably not building the next 00:09:36.000 --> 00:09:38.200 Uber or a super complex database system with 00:09:38.200 --> 00:09:41.629 this yet. But for that huge middle ground of 00:09:41.629 --> 00:09:44.730 useful utilities, yeah, Gemini can handle a surprising 00:09:44.730 --> 00:09:47.129 amount. Probably, like you said, 80 % of what 00:09:47.129 --> 00:09:49.490 most individuals or small teams might need day 00:09:49.490 --> 00:09:51.909 to day. So drilling down, what's the main reason 00:09:51.909 --> 00:09:54.389 Google's offering this for free right now? It's 00:09:54.389 --> 00:09:56.590 a strategic weapon in the hypercompetitive AI 00:09:56.590 --> 00:10:00.090 race. Market share and user lock -in. Mid -roll 00:10:00.090 --> 00:10:02.070 sponsor read. OK, so how do you actually start? 00:10:02.169 --> 00:10:04.190 It's pretty simple. Head over to gemini .google 00:10:04.190 --> 00:10:06.669 .com. Log in with your Google account. Yep. But 00:10:06.669 --> 00:10:09.639 here's a tip. Don't try to build your magnum 00:10:09.639 --> 00:10:13.120 opus on the first go. Start really small, like 00:10:13.120 --> 00:10:15.240 seriously small. Ask it to make a button that 00:10:15.240 --> 00:10:17.120 just changes its color when you click it. Why 00:10:17.120 --> 00:10:19.779 start so basic? Because you learn by watching 00:10:19.779 --> 00:10:22.059 how it builds it. You see the structure it creates, 00:10:22.259 --> 00:10:23.679 the code it uses, even though you don't need 00:10:23.679 --> 00:10:26.039 to touch it. It helps you understand how to ask 00:10:26.039 --> 00:10:28.639 for more complex things later. Gotcha. Okay, 00:10:28.639 --> 00:10:31.139 now let's get to the core takeaway. the part 00:10:31.139 --> 00:10:33.659 everyone's waiting for, the golden formula for 00:10:33.659 --> 00:10:36.200 getting good results. How do you prompt this 00:10:36.200 --> 00:10:38.159 thing effectively? All right. This is crucial. 00:10:39.019 --> 00:10:42.419 This separates a useful app from like a frustrating 00:10:42.419 --> 00:10:47.279 failure. Four main rules. Rule one, be extremely 00:10:47.279 --> 00:10:50.679 specific. No vague requests. No. Don't say make 00:10:50.679 --> 00:10:52.240 me a budget app. That's useless. You got to spell 00:10:52.240 --> 00:10:54.440 it out. Make an app with an input box labeled 00:10:54.440 --> 00:10:57.019 expense name, another labeled amount, a drop 00:10:57.019 --> 00:11:00.049 down menu for category with options, food. transport, 00:11:00.330 --> 00:11:03.549 rent. Below these, display a total expenses line 00:11:03.549 --> 00:11:05.190 that updates automatically. See the difference? 00:11:05.370 --> 00:11:07.970 Huge difference. Detail. Okay, rule two. Rule 00:11:07.970 --> 00:11:12.070 two. Describe the user experience, the UX. How 00:11:12.070 --> 00:11:14.509 should it feel to use? What happens when someone 00:11:14.509 --> 00:11:16.769 clicks a button? Give an example. Okay, so for 00:11:16.769 --> 00:11:19.750 that, budget app. Use a clean, minimalist design 00:11:19.750 --> 00:11:22.129 with a gray background and green accent colors 00:11:22.129 --> 00:11:24.970 for buttons. When the user clicks the Add Expense 00:11:24.970 --> 00:11:27.789 button, the expense should be saved, and importantly, 00:11:28.070 --> 00:11:30.409 the name and amount boxes must immediately clear 00:11:30.409 --> 00:11:33.509 themselves, ready for the next entry. That kind 00:11:33.509 --> 00:11:35.529 of flow detail matters. Right, thinking about 00:11:35.529 --> 00:11:38.350 the interaction. Rule 3. List the main features 00:11:38.350 --> 00:11:40.730 and functional requirements. Think beyond the 00:11:40.730 --> 00:11:43.490 basic inputs. What else does it do? For the budget 00:11:43.490 --> 00:11:45.929 app, maybe you need... Include a button to export 00:11:45.929 --> 00:11:49.049 all data to a CSV file. Add a feature to filter 00:11:49.049 --> 00:11:51.370 the displayed expenses by category. You gotta 00:11:51.370 --> 00:11:53.649 ask for the functions explicitly. Okay, specific 00:11:53.649 --> 00:11:56.330 inputs. Described UX, listed features. What's 00:11:56.330 --> 00:11:59.350 rule four? Rule four. Ask for professional design 00:11:59.350 --> 00:12:02.149 parameters. This is about polish. Don't just 00:12:02.149 --> 00:12:04.830 hope it looks good. Tell it. Say use a clean, 00:12:05.070 --> 00:12:08.330 modern design style. And absolute killer tip. 00:12:08.549 --> 00:12:11.110 Let me guess. Responsive design. You got it. 00:12:11.230 --> 00:12:13.789 Always ask for responsive design. Seriously. 00:12:13.950 --> 00:12:15.690 I still mess this up sometimes myself. I'm rushing. 00:12:16.389 --> 00:12:18.429 I wrestle with prompt drift like anyone else. 00:12:18.750 --> 00:12:20.850 And forgetting responsive is a classic mistake. 00:12:21.070 --> 00:12:23.710 And just to clarify for everyone, responsive 00:12:23.710 --> 00:12:26.409 design means? It means the app automatically 00:12:26.409 --> 00:12:28.850 adjusts its layout to look good and work properly 00:12:28.850 --> 00:12:32.549 on any screen size. Desktops, tablets, and especially 00:12:32.549 --> 00:12:34.570 phones. Because if it doesn't work on a phone? 00:12:34.929 --> 00:12:37.870 It's basically useless for most people most of 00:12:37.870 --> 00:12:40.309 the time. Yeah. Skipping that single phrase, 00:12:40.730 --> 00:12:42.769 responsive design, is probably the number one 00:12:42.769 --> 00:12:45.549 reason a generated app feels broken. It's that 00:12:45.549 --> 00:12:48.450 important. OK, so those are the four rules. Specificity, 00:12:48.570 --> 00:12:52.029 UX, features, design responsive. What about the 00:12:52.029 --> 00:12:55.190 process itself? Is it just one shot? No, definitely 00:12:55.190 --> 00:12:57.649 not. Think of it like a conversation, an iteration. 00:12:57.889 --> 00:12:59.750 Your first prompt probably won't be perfect. 00:12:59.929 --> 00:13:03.269 That's OK. Treat the AI like a junior developer 00:13:03.269 --> 00:13:05.889 you're guiding. So you refine it. Yeah. After 00:13:05.889 --> 00:13:07.529 it builds the first version, you give feedback. 00:13:07.750 --> 00:13:10.500 OK. Good start, but can you make the add button 00:13:10.500 --> 00:13:14.000 20 % larger? Or now let's add a pie chart below 00:13:14.000 --> 00:13:16.080 the list that visualizes spending by category. 00:13:16.460 --> 00:13:18.980 Iterate. Talk to it. And thinking like a professional 00:13:18.980 --> 00:13:22.100 pompter helps too. Absolutely. Break down really 00:13:22.100 --> 00:13:25.539 big ideas into smaller logical chunks. Give context. 00:13:26.000 --> 00:13:28.980 Telling Gemini the tool is, for a science teacher 00:13:28.980 --> 00:13:31.940 explaining photosynthesis, will yield a very 00:13:31.940 --> 00:13:34.059 different result than just asking for a generic 00:13:34.059 --> 00:13:37.419 diagram tool. Context shapes the output. Good 00:13:37.419 --> 00:13:40.139 point. Let's quickly recap the biggest mistakes 00:13:40.139 --> 00:13:42.740 to avoid then. Number one seems to be vagueness 00:13:42.740 --> 00:13:45.980 not being specific. For sure. Number two, absolutely 00:13:45.980 --> 00:13:48.899 forgetting to ask for responsive design. Game 00:13:48.899 --> 00:13:51.100 over if you forget that. Then what else? Not 00:13:51.100 --> 00:13:53.259 testing it. Don't just build it in Gemini and 00:13:53.259 --> 00:13:55.419 assume it's perfect. Send the link. Yeah, you 00:13:55.419 --> 00:13:57.379 can often share them to a friend or a colleague. 00:13:57.899 --> 00:14:00.279 Watch them try to use it. Their confusion is 00:14:00.279 --> 00:14:02.440 gold. It tells you exactly what needs fixing. 00:14:03.059 --> 00:14:05.669 Real user feedback is critical. So if you had 00:14:05.669 --> 00:14:08.149 to pick one single habit that leads to success 00:14:08.149 --> 00:14:10.669 with this? Being ruthlessly specific in your 00:14:10.669 --> 00:14:12.769 requests and treating the whole thing as a back 00:14:12.769 --> 00:14:14.929 and forth conversation, a refinement process. 00:14:15.110 --> 00:14:17.269 OK, let's zoom out. The big picture here. What's 00:14:17.269 --> 00:14:19.350 the ultimate conclusion from all this? I think 00:14:19.350 --> 00:14:21.889 it's pretty profound. The traditional barriers 00:14:21.889 --> 00:14:26.700 to creating digital tools. The time, the money, 00:14:27.059 --> 00:14:29.259 needing specialized coding skills they've basically 00:14:29.259 --> 00:14:32.600 evaporated for, like a huge chunk of common applications. 00:14:32.960 --> 00:14:35.220 That 80 % we talked about. The speed of innovation 00:14:35.220 --> 00:14:38.019 just leaped forward. Dramatically. Individuals, 00:14:38.019 --> 00:14:40.840 small teams, entrepreneurs, smart people are 00:14:40.840 --> 00:14:42.820 already using this. They're building tiny software 00:14:42.820 --> 00:14:46.080 businesses, microsays, testing product ideas 00:14:46.080 --> 00:14:48.659 overnight, creating custom tools to fix their 00:14:48.659 --> 00:14:51.100 own workflow annoyances in minutes. And this 00:14:51.100 --> 00:14:53.240 is just the beginning, presumably. Where does 00:14:53.240 --> 00:14:55.700 this go? It's only going to get more powerful, 00:14:55.779 --> 00:14:58.360 right? In six months, who knows what complexity 00:14:58.360 --> 00:15:01.039 it'll handle in a year or two. It's not crazy 00:15:01.039 --> 00:15:02.779 to think it could replace entire development 00:15:02.779 --> 00:15:05.059 teams for certain kinds of small or medium -sized 00:15:05.059 --> 00:15:08.500 projects. Wow. The trajectory is steep. So the 00:15:08.500 --> 00:15:10.639 call to action for everyone listening is pretty 00:15:10.639 --> 00:15:13.759 straightforward. Dead simple. Go to Gemini .Google 00:15:13.759 --> 00:15:16.720 .com. Log in. Start playing. Build something 00:15:16.720 --> 00:15:19.860 small today. Seriously, the question isn't... 00:15:19.870 --> 00:15:22.629 If AI tools like this will change your job or 00:15:22.629 --> 00:15:25.230 your industry, they will. The only question is 00:15:25.230 --> 00:15:27.370 whether you'll be riding the wave, using this 00:15:27.370 --> 00:15:29.889 advantage, or getting swamped by it because your 00:15:29.889 --> 00:15:32.110 competitors figured it out first. Build something. 00:15:32.250 --> 00:15:35.710 Test an idea today. Use this power. It's a competitive 00:15:35.710 --> 00:15:37.889 edge sitting right there. Use it wisely.