WEBVTT 00:00:00.000 --> 00:00:03.140 Okay, so let's unpack this. Welcome back to the 00:00:03.140 --> 00:00:05.759 Deep Dive. Most professionals have their paid 00:00:05.759 --> 00:00:08.679 AI subscriptions, right? They know the features. 00:00:08.740 --> 00:00:11.480 They're using the tools. We call them AI literate. 00:00:11.859 --> 00:00:15.099 They're efficient, yes. But they're missing out 00:00:15.099 --> 00:00:17.339 on the biggest gains. The real transformation 00:00:17.339 --> 00:00:20.199 happens when you stop seeing AI as just a tool 00:00:20.199 --> 00:00:23.420 and, you know, start treating it like a 247 collaborator. 00:00:23.739 --> 00:00:25.980 Today, we're analyzing a really crucial guide 00:00:25.980 --> 00:00:29.160 that breaks down exactly what separates the top 00:00:29.160 --> 00:00:31.440 performers from everyone else when it comes to 00:00:31.440 --> 00:00:34.679 using these new systems. The mission for us and 00:00:34.679 --> 00:00:36.700 for you is to understand the core difference 00:00:36.700 --> 00:00:40.659 between just efficiency making old work a little 00:00:40.659 --> 00:00:44.479 faster and true transformation, designing fundamentally 00:00:44.479 --> 00:00:47.520 better work with AI at the center. We'll look 00:00:47.520 --> 00:00:50.200 at these specific habits from AI breadcrumbs 00:00:50.200 --> 00:00:52.460 to building swipe files that move you to that 00:00:52.460 --> 00:00:54.850 highest level of mastery. And it's important 00:00:54.850 --> 00:00:56.729 we define these terms because I think everyone 00:00:56.729 --> 00:00:58.609 feels like they're already using AI pretty well. 00:00:58.710 --> 00:01:00.590 Absolutely. And they might be, but maybe not 00:01:00.590 --> 00:01:03.710 at the level for real compounding leverage. Exactly. 00:01:04.109 --> 00:01:06.430 The source material suggests that working with 00:01:06.430 --> 00:01:10.590 AI falls into roughly three clear levels and 00:01:10.590 --> 00:01:14.409 most smart, busy professionals are kind of stuck 00:01:14.409 --> 00:01:17.980 right at level two. So level one is the AI curious. 00:01:18.359 --> 00:01:20.840 This is the experimental user. They use the free 00:01:20.840 --> 00:01:22.599 tools. They play around with them maybe once 00:01:22.599 --> 00:01:24.599 or twice a week. Right. It's optional. It's totally 00:01:24.599 --> 00:01:26.980 optional. There's no compounding benefit to their 00:01:26.980 --> 00:01:28.500 work because they haven't really committed to 00:01:28.500 --> 00:01:31.659 a system. Then you get to level two, the AI literate. 00:01:31.939 --> 00:01:35.060 Now, this user is genuinely skilled. They pay 00:01:35.060 --> 00:01:36.959 for the subscriptions. They keep prompt libraries. 00:01:37.239 --> 00:01:39.480 They know which model to use for what task. They're 00:01:39.480 --> 00:01:41.920 invested. They're invested. But here's the kicker, 00:01:41.920 --> 00:01:45.700 the fatal flaw. AI is still just a tool applied 00:01:45.700 --> 00:01:49.480 inside their old existing workflows. So they're 00:01:49.480 --> 00:01:51.459 doing the same things they always did, just maybe 00:01:51.459 --> 00:01:55.200 20 % faster. They get efficiency, sure, but they're 00:01:55.200 --> 00:01:58.219 missing the transformation that comes from redesigning 00:01:58.219 --> 00:02:00.599 the process itself. And that leads us to level 00:02:00.599 --> 00:02:04.140 three, the AI native. This user is a system builder. 00:02:04.799 --> 00:02:07.099 workflow architect they approach a new project 00:02:07.099 --> 00:02:09.680 and they redesign the process from day one starting 00:02:09.680 --> 00:02:12.199 with the assumption that this intelligent collaborator 00:02:12.199 --> 00:02:16.500 just exists 24 7. ai becomes a genuine force 00:02:16.500 --> 00:02:19.620 multiplier their work compounds faster with less 00:02:19.620 --> 00:02:21.960 effort because the process itself is just fundamentally 00:02:21.960 --> 00:02:24.360 different it's built for collaboration what's 00:02:24.360 --> 00:02:27.960 so fascinating here is that the jump from literate 00:02:27.960 --> 00:02:31.530 to native It's not about writing better prompts. 00:02:31.629 --> 00:02:33.930 It's not about getting more features. It is about 00:02:33.930 --> 00:02:36.550 fundamentally rebuilding how work actually happens. 00:02:36.770 --> 00:02:38.710 That's the real shift, isn't it? If we connect 00:02:38.710 --> 00:02:41.310 this to the bigger picture, what is the core 00:02:41.310 --> 00:02:44.689 difference in the mindset between level two and 00:02:44.689 --> 00:02:47.530 level three? Level two asks how AI can help an 00:02:47.530 --> 00:02:50.479 existing process. Level 3 asks how to design 00:02:50.479 --> 00:02:53.599 a new process, assuming a 247 collaborator. That 00:02:53.599 --> 00:02:56.139 reframing sets the stage perfectly for the first 00:02:56.139 --> 00:02:58.080 habit. Yeah, this is the easiest habit to adopt, 00:02:58.199 --> 00:03:01.039 but it has this outsized impact long term. The 00:03:01.039 --> 00:03:03.819 main problem is that we treat AI chats as disposable, 00:03:03.979 --> 00:03:05.919 one -off things. Right, you use it and you forget 00:03:05.919 --> 00:03:08.340 it. Exactly. And I'll admit, I still wrestle 00:03:08.340 --> 00:03:11.069 with prompt drift myself. Yeah. You spend 20 00:03:11.069 --> 00:03:13.330 minutes getting the perfect sequence, the perfect 00:03:13.330 --> 00:03:16.330 tone. And then three weeks later, you can't find 00:03:16.330 --> 00:03:18.469 that exact thread. Oh, it's the worst feeling. 00:03:18.629 --> 00:03:21.169 It's so frustrating. And that frustration is 00:03:21.169 --> 00:03:22.930 the sign you're leaving performance on the table. 00:03:23.240 --> 00:03:25.159 So the solution is actually pretty simple. We 00:03:25.159 --> 00:03:28.419 call it leaving AI breadcrumbs. You have to anchor 00:03:28.419 --> 00:03:31.759 your AI conversations directly to the work where 00:03:31.759 --> 00:03:34.460 you're using the output. So instead of letting 00:03:34.460 --> 00:03:36.979 those chat threads just live in isolation in 00:03:36.979 --> 00:03:39.479 the app history, which is organized by time, 00:03:39.599 --> 00:03:42.000 you create a unique URL for the conversation 00:03:42.000 --> 00:03:44.520 and paste it right into your document. The core 00:03:44.520 --> 00:03:46.979 productivity principle here is so clear it should 00:03:46.979 --> 00:03:49.759 be obvious, really. Yeah. Organize your information 00:03:49.759 --> 00:03:51.780 by where you will use it, not where you found 00:03:51.780 --> 00:03:54.580 it. Okay, think about preparing a big presentation. 00:03:54.919 --> 00:03:56.860 You've got a Google Doc for the project, right? 00:03:56.979 --> 00:04:00.159 Sure. You just create two tabs. One is the final 00:04:00.159 --> 00:04:02.599 outline, and the other is called AI breadcrumbs 00:04:02.599 --> 00:04:05.620 or helpful hints. That hints tab is where you 00:04:05.620 --> 00:04:08.060 anchor all those chat links. And the process 00:04:08.060 --> 00:04:11.759 is so clean and quick. You ask the AI to optimize 00:04:11.759 --> 00:04:14.819 a prompt or analyze something. The moment you 00:04:14.819 --> 00:04:17.680 get a great result, you immediately copy that 00:04:17.680 --> 00:04:20.509 unique URL. And paste it right in. And paste 00:04:20.509 --> 00:04:22.629 it in your document. And you repeat this for 00:04:22.629 --> 00:04:25.509 any refinement chat you do later. This just makes 00:04:25.509 --> 00:04:28.009 sure you can always return to that exact conversational 00:04:28.009 --> 00:04:31.490 context months later if you need to. And a pro 00:04:31.490 --> 00:04:35.649 tip. Don't just paste the raw links. Add a little 00:04:35.649 --> 00:04:38.050 bit of context next to it, something like Gemini 00:04:38.050 --> 00:04:41.470 chat on brainstorming the Q4 outline or chat 00:04:41.470 --> 00:04:44.189 GPT refining the talking points. Yeah, that context 00:04:44.189 --> 00:04:46.449 is key. The rule of thumb is if the conversation 00:04:46.449 --> 00:04:48.930 took you more than 10 minutes or you know you'll 00:04:48.930 --> 00:04:51.509 need to reference it again, just anchor it immediately. 00:04:51.850 --> 00:04:53.370 So this brings up a really important question 00:04:53.370 --> 00:04:56.050 about the value of that context. Why is linking 00:04:56.050 --> 00:04:58.250 the whole chat better than just copy the final 00:04:58.250 --> 00:05:00.600 output into your document? The link retains the 00:05:00.600 --> 00:05:02.819 full conversation so you can easily pick up the 00:05:02.819 --> 00:05:04.839 thread where you left off, even weeks later. 00:05:05.019 --> 00:05:08.620 That context is often the key. Okay, so breadcrumbs 00:05:08.620 --> 00:05:10.939 solve the memory problem. But what about the 00:05:10.939 --> 00:05:13.740 quality problem? How do we move beyond getting 00:05:13.740 --> 00:05:17.759 consistently, you know, B or B plus work from 00:05:17.759 --> 00:05:20.439 our collaborator? Exactly. That leads us right 00:05:20.439 --> 00:05:24.459 into habit two. Building an AI swipe file system. 00:05:24.639 --> 00:05:27.040 This one takes more effort than breadcrumbs, 00:05:27.040 --> 00:05:30.980 for sure, but the payoff in quality is just... 00:05:31.939 --> 00:05:34.579 Massive. But what exactly is a swipe file? Well, 00:05:34.639 --> 00:05:37.160 historically, it was for like marketing copy. 00:05:37.379 --> 00:05:40.040 But in this context, it's a curated library of 00:05:40.040 --> 00:05:42.980 best in class examples in your field. So emails, 00:05:43.279 --> 00:05:45.600 proposals, summaries, whatever you produce a 00:05:45.600 --> 00:05:48.019 lot of. OK, so most people at level two, the 00:05:48.019 --> 00:05:51.199 AI literate folks, they prompt the AI with basic 00:05:51.199 --> 00:05:52.699 instructions. Yeah, like write me a business 00:05:52.699 --> 00:05:55.620 proposal. Exactly. And they get a generic, competent 00:05:55.620 --> 00:05:58.180 result. An AI native user, on the other hand, 00:05:58.220 --> 00:06:00.740 opens their file system, finds the two or three. 00:06:00.839 --> 00:06:02.759 best proposals they've ever made or seen and 00:06:02.759 --> 00:06:05.300 gives those to the AI first. This is the power 00:06:05.300 --> 00:06:07.300 move. They don't just use a generic prompt. They 00:06:07.300 --> 00:06:08.939 use a really focused structure, something like 00:06:08.939 --> 00:06:11.699 analyze these attached proposals, list the key 00:06:11.699 --> 00:06:14.180 patterns, then apply those patterns to my draft 00:06:14.180 --> 00:06:17.560 content below. And because the AI is learning 00:06:17.560 --> 00:06:19.800 from your absolute best examples, the initial 00:06:19.800 --> 00:06:22.670 draft it gives you is just. It's dramatically 00:06:22.670 --> 00:06:25.569 stronger. Right. You save so much time on refinement 00:06:25.569 --> 00:06:27.910 because the starting quality is already 90 % 00:06:27.910 --> 00:06:29.649 of the way there. That's where the leverage is. 00:06:29.750 --> 00:06:32.550 But it does require some discipline. The new 00:06:32.550 --> 00:06:34.629 habit you have to develop is saving excellent 00:06:34.629 --> 00:06:37.389 work immediately to the system whenever you find 00:06:37.389 --> 00:06:39.189 it. Don't wait until you need it. Don't wait. 00:06:39.329 --> 00:06:42.069 And you can start narrow. Maybe just two or three 00:06:42.069 --> 00:06:44.709 repetitive things like project update emails 00:06:44.709 --> 00:06:47.430 or strategy docs. Just build that initial repository. 00:06:47.829 --> 00:06:49.970 And structure matters a ton for this to scale. 00:06:50.240 --> 00:06:52.639 You have to organize your folders by use case, 00:06:52.800 --> 00:06:57.019 so internal comms, client proposals, not by source 00:06:57.019 --> 00:06:59.839 or date. This is the critical first step to get 00:06:59.839 --> 00:07:02.740 your work library ready so the AI can learn from 00:07:02.740 --> 00:07:05.560 your best stuff and replicate that quality over 00:07:05.560 --> 00:07:08.139 and over. So what is the main outcome of training 00:07:08.139 --> 00:07:11.800 your AI collaborator using this kind of structured 00:07:11.800 --> 00:07:14.980 external swipe file library? It dramatically 00:07:14.980 --> 00:07:17.899 improves output quality and saves time by letting 00:07:17.899 --> 00:07:21.060 the AI replicate proven high standards consistently. 00:07:22.000 --> 00:07:24.500 All right, now we move to habit three. And this 00:07:24.500 --> 00:07:28.550 one is... arguably the hardest to maintain consistently. 00:07:29.870 --> 00:07:32.550 A bit like going to the gym. The long -term impact 00:07:32.550 --> 00:07:35.550 on your output is just massive. It's called AI 00:07:35.550 --> 00:07:37.910 -first task planning. AI -first planning. So 00:07:37.910 --> 00:07:40.310 that means mapping out exactly how and when you're 00:07:40.310 --> 00:07:42.829 going to use AI before you even start a big project. 00:07:43.050 --> 00:07:45.230 Yes. It gets rid of that decision fatigue that 00:07:45.230 --> 00:07:48.050 stalls so many people mid -task. Exactly. You're 00:07:48.050 --> 00:07:50.110 deliberately breaking down a big project into 00:07:50.110 --> 00:07:52.769 small concrete tasks first. Then you go back 00:07:52.769 --> 00:07:54.949 and you mark which tasks AI should help with. 00:07:55.110 --> 00:07:57.649 And this is the key. You specify the best specialized 00:07:57.649 --> 00:08:00.069 tool for that specific task. job. Okay. Let's 00:08:00.069 --> 00:08:01.970 use newsletter creation as a real world example. 00:08:02.189 --> 00:08:05.069 You start mapping. Task 1 .1, brain dump key 00:08:05.069 --> 00:08:07.529 concepts is marked as manual. Right. Because 00:08:07.529 --> 00:08:09.730 you need your human point of view. You have to 00:08:09.730 --> 00:08:11.910 inject your own unique perspective. But then 00:08:11.910 --> 00:08:15.180 task 1 .2, fact check the notes. is marked AI 00:08:15.180 --> 00:08:17.800 -assisted with Notebook LM specifically because 00:08:17.800 --> 00:08:19.680 it's known for better document grounding and 00:08:19.680 --> 00:08:23.120 lower hallucination rates. And task 1 .3, turn 00:08:23.120 --> 00:08:25.939 verified notes into a short brief. That's marked 00:08:25.939 --> 00:08:28.500 as AI -assisted with Gemini because that model 00:08:28.500 --> 00:08:31.220 has a bit of an edge in creative synthesized 00:08:31.220 --> 00:08:33.750 writing. Yeah. This kind of intentional planning 00:08:33.750 --> 00:08:36.929 gives you three big advantages. First, it cuts 00:08:36.929 --> 00:08:39.450 decision fatigue. You never have to stop and 00:08:39.450 --> 00:08:42.090 ask, should I use AI here? It's pre -decided. 00:08:42.549 --> 00:08:45.470 Second, it increases quality and speed because 00:08:45.470 --> 00:08:47.509 you're matching the right tool to the right work. 00:08:47.690 --> 00:08:50.210 Right. And third, it creates reusable templates. 00:08:50.490 --> 00:08:52.590 Once you build this map for a newsletter, you 00:08:52.590 --> 00:08:55.169 just reuse it next month. That's compounding 00:08:55.169 --> 00:08:58.210 leverage. The rule of thumb here is, if a project 00:08:58.210 --> 00:09:00.029 is going to take more than an hour, just spend 00:09:00.029 --> 00:09:02.259 five or ten minutes mapping. the steps and tagging 00:09:02.259 --> 00:09:04.539 them first so if we connect this to the bigger 00:09:04.539 --> 00:09:07.299 picture how does that specific tool assignment 00:09:07.299 --> 00:09:10.019 prevent you from stalling mid -project you pre 00:09:10.019 --> 00:09:12.679 -decide the tool and the task It avoids that 00:09:12.679 --> 00:09:15.559 mid -task analysis paralysis about when or how 00:09:15.559 --> 00:09:17.840 to bring AI in so you keep your momentum. This 00:09:17.840 --> 00:09:20.159 last bonus habit really ties everything we've 00:09:20.159 --> 00:09:21.879 talked about together. It solves one of the most 00:09:21.879 --> 00:09:24.320 frustrating feelings for level 2 users. That 00:09:24.320 --> 00:09:27.179 awful feeling where you wrote a perfect prompt 00:09:27.179 --> 00:09:29.799 three weeks ago. It generated perfect output. 00:09:29.980 --> 00:09:32.039 But today you try to recreate it and the result 00:09:32.039 --> 00:09:35.250 is just... Nah. Yeah, it's just, okay, you lost 00:09:35.250 --> 00:09:38.029 the magic. You lost the magic. The core concept 00:09:38.029 --> 00:09:41.730 here is maintaining a prompts database. You save 00:09:41.730 --> 00:09:43.990 those battle -tested prompts, the ones that leverage 00:09:43.990 --> 00:09:45.610 your swipe files and give you great results, 00:09:45.809 --> 00:09:48.409 into a central library. And it's not about quantity. 00:09:48.610 --> 00:09:50.730 No, you don't need a thousand random prompts. 00:09:50.990 --> 00:09:55.549 Yeah. You need maybe... 10 to 15 really reliable, 00:09:55.730 --> 00:09:58.769 high leverage ones that work every single time. 00:09:58.909 --> 00:10:01.669 And the key, again, is organization, the structure. 00:10:01.870 --> 00:10:04.429 Your database shouldn't just be a list of text. 00:10:04.529 --> 00:10:06.710 Right. It should have the category, like presentation 00:10:06.710 --> 00:10:09.289 outlines, the exact prompt text that worked, 00:10:09.389 --> 00:10:12.549 the context, so when and why to use it, and an 00:10:12.549 --> 00:10:15.370 example output. Maybe even an AI breadcrumb, 00:10:15.389 --> 00:10:17.509 a link back to that original conversation. Whoa. 00:10:17.909 --> 00:10:20.470 That process of refining a perfect prompt, that 00:10:20.470 --> 00:10:22.470 often takes, what, 15 minutes of back and forth? 00:10:22.610 --> 00:10:25.090 At least. And imagine scaling that 15 -minute 00:10:25.090 --> 00:10:28.029 time savings across your whole company or, you 00:10:28.029 --> 00:10:30.929 know, across a billion queries a year globally. 00:10:31.110 --> 00:10:34.850 That compounding value, that leverage, that's 00:10:34.850 --> 00:10:37.350 what being AI native delivers. So for someone 00:10:37.350 --> 00:10:39.690 listening who is ready to make this shift, what's 00:10:39.690 --> 00:10:42.289 the simplest, lowest friction step to start building 00:10:42.289 --> 00:10:45.929 this database this week? Just begin by saving 00:10:45.929 --> 00:10:48.909 the prompts you use repeatedly. Things like rewriting 00:10:48.909 --> 00:10:52.090 rough drafts or analyzing long documents. The 00:10:52.090 --> 00:10:54.129 moment it gives you an excellent output, save 00:10:54.129 --> 00:10:56.629 it. Use a breadcrumb to link to the success story. 00:10:57.000 --> 00:10:59.299 So what does this all mean, really? The key difference 00:10:59.299 --> 00:11:03.259 between AI literate and AI native is this profound 00:11:03.259 --> 00:11:06.820 but subtle mindset shift. AI literate users try 00:11:06.820 --> 00:11:09.220 to fit AI into their existing processes. They're 00:11:09.220 --> 00:11:12.080 aiming for marginal efficiency. AI native users 00:11:12.080 --> 00:11:14.519 redesign the entire process from the ground up 00:11:14.519 --> 00:11:18.039 to leverage a 247 intelligent collaborator. And 00:11:18.039 --> 00:11:20.159 this shift needs momentum, not perfection. You 00:11:20.159 --> 00:11:22.259 have to start small. Right. This week, just commit 00:11:22.259 --> 00:11:24.559 to leaving those AI breadcrumbs. Hyperlink your 00:11:24.559 --> 00:11:26.740 chats into your documents. Next week, create 00:11:26.740 --> 00:11:29.039 a simple swipe file. Just three folders with 00:11:29.039 --> 00:11:31.500 a few high -quality examples. Then map out one 00:11:31.500 --> 00:11:34.220 small project, AI first. The goal isn't immediate, 00:11:34.419 --> 00:11:37.460 flawless implementation. It's just meaningful, 00:11:37.600 --> 00:11:40.720 deliberate progress toward a truly AI -native 00:11:40.720 --> 00:11:43.820 way of working. And this approach, it doesn't 00:11:43.820 --> 00:11:46.279 just save time. It dramatically improves the 00:11:46.279 --> 00:11:48.799 quality ceiling of your output. It helps you 00:11:48.799 --> 00:11:51.159 produce better work than you ever could alone. 00:11:51.740 --> 00:11:54.320 So if you want to achieve true professional transformation, 00:11:54.559 --> 00:11:56.539 not just a little bit of efficiency, start asking 00:11:56.539 --> 00:11:58.960 this question next time you plan a task. How 00:11:58.960 --> 00:12:01.360 would I design this workflow differently if I 00:12:01.360 --> 00:12:03.330 assumed an intelligent Intelligent Collaborator 00:12:03.330 --> 00:12:05.809 was available and ready to help me 24 -7. That's 00:12:05.809 --> 00:12:06.610 something to mull over.