WEBVTT 00:00:00.000 --> 00:00:02.240 I was looking at my credit card bill the other 00:00:02.240 --> 00:00:05.320 day. Oh, no. And the numbers were just, well, 00:00:05.379 --> 00:00:07.980 they were telling a grim story. I was paying 00:00:07.980 --> 00:00:12.359 for five separate AI writing tools, three different 00:00:12.359 --> 00:00:16.379 image generators. It felt less like a strategic 00:00:16.379 --> 00:00:19.120 toolkit and more like, I don't know, a digital 00:00:19.120 --> 00:00:21.980 mess. That is the classic trap. The problem isn't 00:00:21.980 --> 00:00:24.300 that you need more tools. The real struggle is 00:00:24.300 --> 00:00:26.559 that bouncing between tabs, hoping the next shiny 00:00:26.559 --> 00:00:28.640 thing is going to fix it. Yeah, and all it really 00:00:28.640 --> 00:00:30.879 does is create more friction. Exactly. I thought 00:00:30.879 --> 00:00:33.259 I needed this huge arsenal, but the core idea 00:00:33.259 --> 00:00:36.229 we took from the source material is that The 00:00:36.229 --> 00:00:39.670 big players, you know, Chad GPT, Claude, Gemini, 00:00:40.229 --> 00:00:42.570 they're actually enough. They are, but only if 00:00:42.570 --> 00:00:44.729 you have a strategic system for how they talk 00:00:44.729 --> 00:00:46.409 to each other. And that's our mission for this 00:00:46.409 --> 00:00:49.250 deep dive. We've pulled out eight specific high 00:00:49.250 --> 00:00:51.310 leverage workflows. These are practical methods. 00:00:51.429 --> 00:00:53.350 They're designed to stop you from working randomly 00:00:53.350 --> 00:00:56.049 and instead start integrating your AI tools like 00:00:56.049 --> 00:00:58.829 a real pro. We're looking for structure, efficiency, 00:00:59.429 --> 00:01:03.670 and those genuine aha moments. And the core philosophy 00:01:03.670 --> 00:01:06.909 here is really crucial. Stop asking one single 00:01:06.909 --> 00:01:09.409 AI to do everything. That's where quality just 00:01:09.409 --> 00:01:11.930 collapses. Right. You need to use them like a 00:01:11.930 --> 00:01:13.969 specialized team. I mean, if you're building 00:01:13.969 --> 00:01:16.590 a house, you need a hammer, a saw, and a drill. 00:01:17.030 --> 00:01:19.290 You don't need 50 different hammers, right? So 00:01:19.290 --> 00:01:21.989 let's define the five core tools these workflows 00:01:21.989 --> 00:01:24.840 rely on. They each have a very specific job. 00:01:25.340 --> 00:01:28.099 Right. So first up is ChatGPT. It's your general 00:01:28.099 --> 00:01:31.540 helper. Great for writing, generating text, anything 00:01:31.540 --> 00:01:34.379 that needs some personalized context. Then there's 00:01:34.379 --> 00:01:36.959 Perplexity. This is our fact checker, our link 00:01:36.959 --> 00:01:39.239 finder. Think of it like a super fast librarian. 00:01:40.560 --> 00:01:43.159 And Notebook LM is the deep document reader. 00:01:43.379 --> 00:01:45.640 This one is key because it studies your files 00:01:45.640 --> 00:01:48.480 and sources and structurally it just cannot make 00:01:48.480 --> 00:01:51.099 things up. Okay, and Gemini. Gemini is the builder. 00:01:51.420 --> 00:01:53.780 It's for creating tangible things like charts, 00:01:53.959 --> 00:01:55.900 slides, and code, and it often has these really 00:01:55.900 --> 00:01:59.079 essential file exports. And lastly Claude, the 00:01:59.079 --> 00:02:01.599 artist. The aesthetic expert, yeah. It makes 00:02:01.599 --> 00:02:04.500 things look polished, beautiful. It ensures that 00:02:04.500 --> 00:02:07.019 visual consistency. If you can keep those five 00:02:07.019 --> 00:02:09.860 roles straight, these workflows become incredibly 00:02:09.860 --> 00:02:12.719 intuitive. OK, let's unpack this. We can start 00:02:12.719 --> 00:02:15.580 with research and analysis. So focusing on context 00:02:15.580 --> 00:02:17.759 and truth. The first method is what they call 00:02:17.759 --> 00:02:20.199 the smart brain method using chat GPT. Right. 00:02:20.280 --> 00:02:22.240 So the biggest mistake people make is they open 00:02:22.240 --> 00:02:25.479 chat GPT, ask a question, and then close the 00:02:25.479 --> 00:02:28.259 tab. And next time you come back. The AI has 00:02:28.259 --> 00:02:29.860 forgotten everything. You're starting from zero. 00:02:30.520 --> 00:02:33.360 every single time. And that locks you into a 00:02:33.360 --> 00:02:36.439 cycle of just generic low -quality answers. So 00:02:36.439 --> 00:02:38.900 what's the solution? The solution is to leverage 00:02:38.900 --> 00:02:41.680 the project's feature. A project is kind of like 00:02:41.680 --> 00:02:44.439 a dedicated physical folder for one objective. 00:02:44.580 --> 00:02:47.280 You create it, say, my Q3 marketing plan, and 00:02:47.280 --> 00:02:49.439 then you preload it with knowledge. So you're 00:02:49.439 --> 00:02:51.780 uploading your brand guidelines, maybe last quarter's 00:02:51.780 --> 00:02:54.400 performance data, your budget notes, all the 00:02:54.400 --> 00:02:56.840 necessary background. Yes. And now, when you 00:02:56.840 --> 00:02:59.699 prompt it, act as an agency copywriter and draft 00:02:59.699 --> 00:03:01.680 a social media campaign passed on my budget, 00:03:01.979 --> 00:03:04.340 it's immediately smarter. Because it's grounded 00:03:04.340 --> 00:03:07.300 in your specific reality, not just the general 00:03:07.300 --> 00:03:10.139 internet. Exactly. And what's really fascinating 00:03:10.139 --> 00:03:12.960 here is the loop effect. This is the mechanism 00:03:12.960 --> 00:03:16.240 that stops that prompt drift. Ah, that's the 00:03:16.240 --> 00:03:19.460 secret sauce for compounding the knowledge. When 00:03:19.460 --> 00:03:22.240 ChatGPT produces something good, like a competitive 00:03:22.240 --> 00:03:24.800 analysis, you download that report. And then 00:03:24.800 --> 00:03:27.020 you upload it back into the project folder. Yes. 00:03:27.310 --> 00:03:30.530 So now the AI's own research becomes part of 00:03:30.530 --> 00:03:33.090 its permanent memory for that project. So you're 00:03:33.090 --> 00:03:35.449 constantly reinforcing its knowledge base. You 00:03:35.449 --> 00:03:38.030 are. And when you later ask for five Instagram 00:03:38.030 --> 00:03:40.969 captions, the tone and the content are guaranteed 00:03:40.969 --> 00:03:43.610 to match the brand document you fed it. It's 00:03:43.610 --> 00:03:45.710 locked into your world. That makes perfect sense 00:03:45.710 --> 00:03:48.169 for context, but what about facts? When you need 00:03:48.169 --> 00:03:51.740 something that's 100 % verifiable, Context isn't 00:03:51.740 --> 00:03:53.840 enough. And that brings us to workflow 2, the 00:03:53.840 --> 00:03:56.020 trustworthy researcher. This gets at that core 00:03:56.020 --> 00:03:59.159 fear, AI just hallucinating. Making things up. 00:03:59.340 --> 00:04:02.439 Right. So for high stakes facts, you need a combination. 00:04:02.900 --> 00:04:05.639 We use perplexity as the hunter, but, and this 00:04:05.639 --> 00:04:08.479 is important, we restrain it. You don't ask for 00:04:08.479 --> 00:04:11.560 a summary. You only ask for the links. Ah, so 00:04:11.560 --> 00:04:13.900 you'd prompt something like, find the top five 00:04:13.900 --> 00:04:16.680 scientific articles on sustainable concrete and 00:04:16.680 --> 00:04:20.639 give me the URL links only. You want the source, 00:04:21.079 --> 00:04:23.980 not the AI's interpretation. Correct. You take 00:04:23.980 --> 00:04:27.019 those links, download the actual articles, and 00:04:27.019 --> 00:04:29.399 you feed them directly into Notebook LM. Which 00:04:29.399 --> 00:04:31.560 is the Google tool designed for deep reading. 00:04:31.660 --> 00:04:34.439 And it's structurally incapable of making stuff 00:04:34.439 --> 00:04:36.980 up because it only uses the documents you provide. 00:04:37.160 --> 00:04:39.120 You can verify every single output. If it gives 00:04:39.120 --> 00:04:41.420 you a summary, you can instantly see which document 00:04:41.420 --> 00:04:43.680 and page number that fact came from. And we add 00:04:43.680 --> 00:04:46.100 the persona trick here, too. In the Notebook 00:04:46.100 --> 00:04:48.360 LM settings, you can tell it to act as, I don't 00:04:48.360 --> 00:04:51.899 know, a skeptical engineering peer who demands 00:04:51.899 --> 00:04:54.199 empirical proof. So the analysis it generates 00:04:54.199 --> 00:04:57.720 is critical. honest, and fact -grounded, based 00:04:57.720 --> 00:05:00.100 only on the sources you gave it. That combination 00:05:00.100 --> 00:05:02.819 is how you guarantee verifiable truth. OK, let's 00:05:02.819 --> 00:05:05.899 shift gears. Segment two, strategic visualization. 00:05:06.459 --> 00:05:08.660 We've done the deep thinking. Now we need to 00:05:08.660 --> 00:05:10.920 actually show the results to people. And quickly. 00:05:11.779 --> 00:05:14.779 So workflow three, turning long reports into 00:05:14.779 --> 00:05:17.639 infographics using perplexity and Gemini Canvas. 00:05:18.259 --> 00:05:21.019 The pain point here is always dense PDFs. Right. 00:05:21.160 --> 00:05:23.600 So perplexity finds the detailed report, let's 00:05:23.600 --> 00:05:27.029 say, global Q4 sales data. You download that 00:05:27.029 --> 00:05:29.730 PDF. Then you go over to Gemini and its visualization 00:05:29.730 --> 00:05:32.290 space, which is called Canvas. Think of Canvas 00:05:32.290 --> 00:05:35.009 as Gemini's workspace for visual outputs and 00:05:35.009 --> 00:05:37.709 code previews. You upload the PDF there. And 00:05:37.709 --> 00:05:39.850 instead of asking for a text summary, you prompt 00:05:39.850 --> 00:05:42.329 it for a structure. Something like, create a 00:05:42.329 --> 00:05:44.430 visual flowchart showing the revenue breakdown, 00:05:44.709 --> 00:05:47.490 highlighting regions with over 5 % growth. Exactly. 00:05:47.910 --> 00:05:50.410 And Gemini Canvas will generate a visual diagram 00:05:50.410 --> 00:05:53.430 based on code, not just static text. It jumps 00:05:53.430 --> 00:05:56.490 you from raw text analysis straight to a presentation 00:05:56.490 --> 00:05:59.110 -ready visual. That's speed for sure, but sometimes 00:05:59.110 --> 00:06:02.569 those generated visuals lack polish. If we need 00:06:02.569 --> 00:06:04.389 professional aesthetics, the source suggests 00:06:04.389 --> 00:06:07.540 moving to Claude for workflow 4. making beautiful 00:06:07.540 --> 00:06:09.379 dashboards. This is where we need to understand 00:06:09.379 --> 00:06:11.980 Claude's unique advantage. Claude excels because 00:06:11.980 --> 00:06:14.660 of its exceptionally large context window. What 00:06:14.660 --> 00:06:16.879 does that mean exactly? It means it can hold 00:06:16.879 --> 00:06:19.579 complex style guides, like your entire company 00:06:19.579 --> 00:06:22.339 brand guide, in its memory way better than other 00:06:22.339 --> 00:06:26.139 models. It ensures visual consistency. So the 00:06:26.139 --> 00:06:28.560 key step here is teaching it your style first. 00:06:28.879 --> 00:06:31.720 That's it. You set up a Claude project. You upload 00:06:31.720 --> 00:06:34.500 an image or a PDF that defines your color palette, 00:06:34.779 --> 00:06:37.759 say, rich blue and silver, and a specific font. 00:06:38.019 --> 00:06:40.899 And you literally tell Claude, use these specific 00:06:40.899 --> 00:06:43.279 hex codes and fonts for all charts you make for 00:06:43.279 --> 00:06:46.019 me. Then you upload your data, like a CSV file 00:06:46.019 --> 00:06:48.560 of website engagement, and ask it to create an 00:06:48.560 --> 00:06:51.600 interactive dashboard with that exact look and 00:06:51.600 --> 00:06:53.959 feel. And Claude generates what they call an 00:06:53.959 --> 00:06:56.360 artifact. It's a code pop -up that looks and 00:06:56.360 --> 00:06:58.939 acts like a real interactive dashboard. Whoa, 00:06:59.199 --> 00:07:01.180 imagine scaling that. You teach it the style 00:07:01.180 --> 00:07:03.779 once, and it maintains that aesthetic for every 00:07:03.779 --> 00:07:06.300 single monthly report forever. That saves a massive 00:07:06.300 --> 00:07:09.199 amount of design time. It's engineering consistency, 00:07:09.360 --> 00:07:11.199 not just making a picture. Which leads right 00:07:11.199 --> 00:07:13.560 into workflow five, social media images using 00:07:13.560 --> 00:07:17.240 Claude SVG and a tool like Figma. Exactly. Stop 00:07:17.240 --> 00:07:19.639 paying a designer 50 bucks for a simple graphic 00:07:19.639 --> 00:07:22.420 with three bullet points. So first, perplexity 00:07:22.420 --> 00:07:25.079 finds the list. The seven best habits for remote 00:07:25.079 --> 00:07:27.879 workers. Then you take that to Claude. and you 00:07:27.879 --> 00:07:30.259 prompt it to generate a clean modern graphic 00:07:30.259 --> 00:07:33.670 for Instagram, but critically... You demand an 00:07:33.670 --> 00:07:36.470 SVG file. And SVG is a special image file type, 00:07:36.589 --> 00:07:39.110 right? It's made of math, not pixels. It's made 00:07:39.110 --> 00:07:41.129 of mathematical code. And because it's math, 00:07:41.550 --> 00:07:43.529 you can download it, drop it into a free tool 00:07:43.529 --> 00:07:45.949 like Figma or Canva, and suddenly you have total 00:07:45.949 --> 00:07:48.509 control. So if Claude misplaced a text block, 00:07:48.709 --> 00:07:51.129 you just fix it instantly. Change the font, swap 00:07:51.129 --> 00:07:53.550 a color. It's a one -second edit. You get the 00:07:53.550 --> 00:07:56.269 speed of AI design, but the precise control of 00:07:56.269 --> 00:07:58.689 a professional for those final tweaks. It's a 00:07:58.689 --> 00:08:00.959 game changer for content velocity. All right. 00:08:01.019 --> 00:08:03.480 Let's transition from thinking and visualizing 00:08:03.480 --> 00:08:07.160 to our last segment. Building real things. Creating 00:08:07.160 --> 00:08:10.680 high value assets. Workflow six. Instant presentations 00:08:10.680 --> 00:08:13.620 using Gemini gems. A gem is basically Gemini's 00:08:13.620 --> 00:08:16.000 version of a custom GPT. You train it once to 00:08:16.000 --> 00:08:19.300 do one specific repeatable job, like investor 00:08:19.300 --> 00:08:21.660 deckmaker. So you'd instruct this gem to always 00:08:21.660 --> 00:08:24.259 follow a specific structure. Slide title, three 00:08:24.259 --> 00:08:26.810 bullet points, two image suggestions. And when 00:08:26.810 --> 00:08:29.449 you need a deck on the impact of rising interest 00:08:29.449 --> 00:08:32.450 rates, you just talk to the gem. It spits out 00:08:32.450 --> 00:08:34.750 the perfect outline structured exactly how you 00:08:34.750 --> 00:08:37.850 like it. But then comes the magic export. This 00:08:37.850 --> 00:08:40.409 is the feature that really sells it. It is. You 00:08:40.409 --> 00:08:43.169 ask Gemini to create the slides from that outline, 00:08:43.429 --> 00:08:46.289 and it generates a file that opens directly in 00:08:46.289 --> 00:08:49.269 Google Slides. So no more copy pasting. It eliminates 00:08:49.269 --> 00:08:51.929 the single greatest friction point in making 00:08:51.929 --> 00:08:54.929 presentations. No more copy, paste, and reformat. 00:08:55.090 --> 00:08:58.610 That is huge for momentum. Yeah. Okay, now. Workflow 00:08:58.610 --> 00:09:02.070 7. Creating a full training course with Notebook 00:09:02.070 --> 00:09:05.230 LM and Gemini. This sounds like it turns weeks 00:09:05.230 --> 00:09:08.129 of work into an afternoon. It can. You start 00:09:08.129 --> 00:09:09.970 by gathering all your training documents, your 00:09:09.970 --> 00:09:12.669 technical manuals, into Notebook LM. That's your 00:09:12.669 --> 00:09:14.350 authoritative source. And then you use a feature 00:09:14.350 --> 00:09:17.100 called Notebook LM's audio overview. Yeah, this 00:09:17.100 --> 00:09:19.480 is cool. It turns your documents into a synthesized 00:09:19.480 --> 00:09:22.159 two -person discussion podcast, an mp3 lesson 00:09:22.159 --> 00:09:24.860 that's an instant piece of training content generated 00:09:24.860 --> 00:09:27.399 directly from your own documents. That is immense 00:09:27.399 --> 00:09:30.759 leverage. And because Notebook LM is fact -grounded, 00:09:30.779 --> 00:09:33.279 when you ask it to create a 10 -question quiz 00:09:33.279 --> 00:09:36.159 or a one -page cheat sheet for the course, you 00:09:36.159 --> 00:09:38.539 are guaranteed that the output sticks strictly 00:09:38.539 --> 00:09:40.820 to the facts in your original source materials. 00:09:41.139 --> 00:09:43.360 That consistency is essential for educational 00:09:43.360 --> 00:09:46.960 tools. Finally, workflow eight. building a landing 00:09:46.960 --> 00:09:49.600 page website. We start with strategic research. 00:09:49.779 --> 00:09:51.980 I love this example. Scraping real complaints 00:09:51.980 --> 00:09:54.860 people have about a service, say, dog walking 00:09:54.860 --> 00:09:57.419 from a form like Reddit, and pasting them into 00:09:57.419 --> 00:09:59.840 Notebook LM. So you know the exact emotional 00:09:59.840 --> 00:10:01.899 pain points of the market before you write a 00:10:01.899 --> 00:10:04.639 single line of copy. Precisely. Notebook LM generates 00:10:04.639 --> 00:10:06.960 the strategy brief from those complaints. It 00:10:06.960 --> 00:10:09.460 identifies that people hate lateness, so it suggests 00:10:09.460 --> 00:10:12.200 the headline, the only dog walker who is always 00:10:12.200 --> 00:10:14.860 on time. You take that high leverage brief, copy 00:10:14.860 --> 00:10:17.360 it, and take it to Gemini. Prompt it to write 00:10:17.360 --> 00:10:20.500 the basic HTML and CSS code for a landing page 00:10:20.500 --> 00:10:23.600 based on that strategy. And you immediately see 00:10:23.600 --> 00:10:26.200 a live preview of the website in the Canvas view. 00:10:26.480 --> 00:10:28.779 I mean, I still wrestle with prompt drift myself, 00:10:29.019 --> 00:10:31.320 you know, especially asking for code. I was asking 00:10:31.320 --> 00:10:33.240 for some Python code the other day, and it started 00:10:33.240 --> 00:10:35.980 trying to write me a limerick. Right. But seeing 00:10:35.980 --> 00:10:38.460 a code preview instantly just takes away all 00:10:38.460 --> 00:10:40.139 that friction. You can deploy it right away. 00:10:40.480 --> 00:10:43.159 OK. We just covered eight intense workflows. 00:10:44.299 --> 00:10:47.100 The big idea here remains. You don't need more 00:10:47.100 --> 00:10:50.259 tools. You need a strategic system to connect 00:10:50.259 --> 00:10:52.519 the five major ones you probably already have. 00:10:52.700 --> 00:10:55.100 And to recap their roles one last time for you, 00:10:55.340 --> 00:10:58.720 perplexity is for reliable fact finding. Notebook 00:10:58.720 --> 00:11:01.840 LM is for trustworthy, deep reading of your documents. 00:11:02.080 --> 00:11:04.480 ChatGPT gives you that personalized context and 00:11:04.480 --> 00:11:06.840 custom content creation. Claude handles the aesthetics, 00:11:07.120 --> 00:11:09.120 giving you professional polish on visuals and 00:11:09.120 --> 00:11:11.539 dashboards. And Gemini builds the final structures, 00:11:11.740 --> 00:11:14.240 the charts, the code, with those essential file 00:11:14.240 --> 00:11:17.379 exports. But start small. Please do not try all 00:11:17.379 --> 00:11:19.620 eight of these today. That's just a recipe for 00:11:19.620 --> 00:11:21.820 overwhelm. The practical advice from the source 00:11:21.820 --> 00:11:25.139 is this. Pick one problem you have right now. 00:11:25.580 --> 00:11:29.100 Maybe you hate writing emails or you waste time 00:11:29.100 --> 00:11:32.500 formatting slides. Choose the workflow that fixes 00:11:32.500 --> 00:11:34.960 that one thing. And commit to trying it five 00:11:34.960 --> 00:11:37.559 times. The first time will feel slow, it'll feel 00:11:37.559 --> 00:11:40.320 clunky, guaranteed. But by the fifth time, it'll 00:11:40.320 --> 00:11:42.460 be fast, it'll be automatic. And that's when 00:11:42.460 --> 00:11:44.659 you stop paying for the tools and you start leveraging 00:11:44.659 --> 00:11:47.120 the system. The true power here isn't just the 00:11:47.120 --> 00:11:49.460 output. It's not the quiz or the code you made. 00:11:49.919 --> 00:11:52.399 It's the fact that you've engineered a predictable, 00:11:52.759 --> 00:11:54.940 repeatable system that consistently understands 00:11:54.940 --> 00:11:57.059 your context. You've stopped managing a collection 00:11:57.059 --> 00:11:59.279 of apps and started directing a high -performing 00:11:59.279 --> 00:12:01.779 team. And that repeatable system is what scales 00:12:01.779 --> 00:12:04.679 your knowledge across any future idea, any future 00:12:04.679 --> 00:12:07.700 business need. It's an infrastructure of automated 00:12:07.700 --> 00:12:09.779 intelligence. Go build something powerful. We'll 00:12:09.779 --> 00:12:10.299 see you next time.