WEBVTT 00:00:00.000 --> 00:00:02.560 You know, right now, for most people, AI is still 00:00:02.560 --> 00:00:04.919 just a slightly smarter autocomplete. They ask 00:00:04.919 --> 00:00:06.980 a question, they get an answer, maybe they paste 00:00:06.980 --> 00:00:09.220 it somewhere, and then they just move on. It's 00:00:09.220 --> 00:00:11.539 purely transactional. Totally. And that's like 00:00:11.539 --> 00:00:14.539 operating AI as a fancy calculator. But the shift 00:00:14.539 --> 00:00:17.440 that's happening with serious founders is it's 00:00:17.440 --> 00:00:19.600 transformative. How so? They're running entire 00:00:19.600 --> 00:00:22.539 business functions. I'm talking analysis, content 00:00:22.539 --> 00:00:25.820 generation, delegation, all through integrated 00:00:25.820 --> 00:00:29.000 AI systems. We actually have hard data. on this. 00:00:29.640 --> 00:00:32.140 Yeah, groups using the more advanced models like 00:00:32.140 --> 00:00:34.840 GPT -4, they're finishing complex tasks about 00:00:34.840 --> 00:00:39.159 25 % faster, and the quality is noticeably higher. 00:00:39.340 --> 00:00:42.619 Wow. That's a real compounding productivity shift 00:00:42.619 --> 00:00:44.799 that's already happening. So the hard truth is, 00:00:44.880 --> 00:00:46.840 if you're just using AI like a quick helper, 00:00:47.200 --> 00:00:48.619 you're not just missing out, you're actually 00:00:48.619 --> 00:00:51.479 falling behind. This deep dive isn't about shortcuts, 00:00:51.799 --> 00:00:54.350 it's about building the entire system. Welcome 00:00:54.350 --> 00:00:56.530 to the Deep Dive. Our mission today is to cut 00:00:56.530 --> 00:01:00.250 through the noise and really extract the six 00:01:00.250 --> 00:01:03.049 foundational AI skills you need to build what 00:01:03.049 --> 00:01:05.469 we're calling an AI -first operating system for 00:01:05.469 --> 00:01:08.530 your business. Exactly. The whole principle is 00:01:08.530 --> 00:01:11.349 about shifting AI from being this site tool you 00:01:11.349 --> 00:01:14.150 visit sometimes to being the core infrastructure 00:01:14.150 --> 00:01:16.870 that your workflow passes through. It's the difference 00:01:16.870 --> 00:01:20.409 between asking AI for one answer today and building 00:01:20.409 --> 00:01:22.409 a system that automatically repeats your wins 00:01:22.409 --> 00:01:25.200 tomorrow. That's how you scale output without 00:01:25.200 --> 00:01:27.560 scaling chaos. So we'll walk through all six 00:01:27.560 --> 00:01:30.939 skills, right? From just capturing raw ideas 00:01:30.939 --> 00:01:33.579 at thinking speed all the way to turning AI into 00:01:33.579 --> 00:01:36.280 your strategic advisor. Let's start with that 00:01:36.280 --> 00:01:39.540 mindset shift. Ditching what you call the fancy 00:01:39.540 --> 00:01:42.659 calculator trap. A tool helps you once. Right, 00:01:42.879 --> 00:01:45.019 but infrastructure like, you know, plumbing or 00:01:45.019 --> 00:01:47.579 electricity, it compounds. It works for you forever. 00:01:47.859 --> 00:01:50.219 So infrastructure means the AI is sitting inside 00:01:50.219 --> 00:01:52.950 your workflow. Exactly. It learns your standards, 00:01:53.150 --> 00:01:55.450 your tone, your frameworks, and it repeats those 00:01:55.450 --> 00:01:57.609 wins automatically. So you don't start from zero 00:01:57.609 --> 00:01:59.909 every single morning. But here's a critical insight. 00:02:00.409 --> 00:02:03.489 Okay. AI does not fix chaos. It just amplifies 00:02:03.489 --> 00:02:06.510 it. Wait, say that again. If AI amplifies chaos, 00:02:06.689 --> 00:02:08.550 does that mean most founders rushing to adopt 00:02:08.550 --> 00:02:10.250 these tools are actually making their businesses 00:02:10.250 --> 00:02:13.229 worse? Just faster? That is precisely it. They 00:02:13.229 --> 00:02:16.189 are automating a bad habit. So the first step, 00:02:16.330 --> 00:02:18.789 the single most critical step, is to build and 00:02:18.789 --> 00:02:22.030 test the manual process first. Then, and only 00:02:22.030 --> 00:02:24.650 then, do you use AI to scale that structure. 00:02:24.770 --> 00:02:27.530 Okay, that makes sense. Let's move to skill number 00:02:27.530 --> 00:02:29.930 one then, the foundation for this whole system. 00:02:30.569 --> 00:02:32.870 Voice Detects Mastery. Because if you're still 00:02:32.870 --> 00:02:35.129 typing out your first ideas, you're just moving 00:02:35.129 --> 00:02:37.610 too slow. It really comes down to the cognitive 00:02:37.610 --> 00:02:40.210 math. You can think and speak about three times 00:02:40.210 --> 00:02:42.780 faster than you can type. Three times. Yeah. 00:02:43.080 --> 00:02:46.199 And typing, it forces you to edit while you think. 00:02:46.300 --> 00:02:48.680 It just kills your creative flow before the idea 00:02:48.680 --> 00:02:51.060 can even fully form. So it's not just about speed. 00:02:51.159 --> 00:02:53.060 It's about the quality of the input. I'm feeding 00:02:53.060 --> 00:02:57.520 the AI a much richer, a more raw version of my 00:02:57.520 --> 00:02:59.879 thoughts if I'm speaking. Exactly. You stop wasting 00:02:59.879 --> 00:03:02.460 mental energy on grammar and structure at that 00:03:02.460 --> 00:03:05.250 initial stage. There's a tool mentioned, Whisper 00:03:05.250 --> 00:03:07.469 Flow, which is basically an intelligent transcription 00:03:07.469 --> 00:03:09.729 service. And what does that do? It turns that 00:03:09.729 --> 00:03:12.349 messy stream of consciousness speech into clean, 00:03:12.650 --> 00:03:14.909 structured text. And it immediately formats it 00:03:14.909 --> 00:03:17.610 into a really clear, usable prompt. I love that 00:03:17.610 --> 00:03:19.870 example of voice prompting a job description 00:03:19.870 --> 00:03:22.909 for a Chief of Staff. You just ramble about the 00:03:22.909 --> 00:03:26.150 requirements, the personality, everything. And 00:03:26.150 --> 00:03:28.650 it spits out a professional structured prompt 00:03:28.650 --> 00:03:31.270 with bullet points. So you skip the whole first 00:03:31.270 --> 00:03:33.729 draft. Yep. You get better inputs. which naturally 00:03:33.729 --> 00:03:36.270 leads to better outputs. Okay, so we've captured 00:03:36.270 --> 00:03:38.330 the idea at the speed of thought. Skill number 00:03:38.330 --> 00:03:41.469 two is about what to do with it. AI assisted 00:03:41.469 --> 00:03:44.830 content creation. Because let's be honest, most 00:03:44.830 --> 00:03:47.770 AI content sounds really generic. It does. You 00:03:47.770 --> 00:03:50.169 can smell it a mile away because it's just guessing. 00:03:50.349 --> 00:03:53.050 It's averaging out the entire internet. The solution 00:03:53.050 --> 00:03:56.770 is to teach AI your unique voice. You do this 00:03:56.770 --> 00:03:59.610 using dedicated Claude projects. Think of it 00:03:59.610 --> 00:04:02.969 like a content DNA vault. Your content DNA. I 00:04:02.969 --> 00:04:05.870 like that. Yeah. Instead of asking it to create 00:04:05.870 --> 00:04:09.330 from scratch, you feed it your best stuff. Transcripts, 00:04:09.370 --> 00:04:12.330 frameworks, sales calls, anything that's uniquely 00:04:12.330 --> 00:04:15.370 you. And Claude stops guessing and just starts 00:04:15.370 --> 00:04:17.660 pattern matching you. So there's a system to 00:04:17.660 --> 00:04:19.899 it. One, upload your material to build that knowledge 00:04:19.899 --> 00:04:22.740 base. Two, you lock in a really specific prompt 00:04:22.740 --> 00:04:25.379 that defines your role, your tone, even band 00:04:25.379 --> 00:04:28.319 phrases. Right. And then three, you collaborate. 00:04:28.720 --> 00:04:31.220 The AI gives you hooks and structure, but you're 00:04:31.220 --> 00:04:33.779 still the final editor. Precisely. The human 00:04:33.779 --> 00:04:36.540 has to maintain ultimate judgment, you know, 00:04:37.139 --> 00:04:39.839 taste. And I'll admit something here, just to 00:04:39.839 --> 00:04:42.660 be transparent. Yeah. I still wrestle with prompt 00:04:42.660 --> 00:04:45.500 drift myself. It takes constant checking to keep 00:04:45.500 --> 00:04:49.139 the AI aligned with that subtle human personality. 00:04:49.339 --> 00:04:51.620 It's not a set it and forget it thing. That brings 00:04:51.620 --> 00:04:54.060 up a critical point, though. If the AI needs 00:04:54.060 --> 00:04:57.220 a system prompt, what is the human's ultimate 00:04:57.220 --> 00:05:00.339 role in this content process? The human must 00:05:00.339 --> 00:05:03.160 maintain ultimate judgment, taste, and personality 00:05:03.160 --> 00:05:05.800 over the final draft. Skill number three feels 00:05:05.800 --> 00:05:08.319 like the answer to a problem every founder has. 00:05:08.699 --> 00:05:11.000 It's the delegation bottleneck, right? Building 00:05:11.000 --> 00:05:14.100 SOPs with loom AI. Writing those standard operating 00:05:14.100 --> 00:05:16.579 procedures is just so slow. It's agonizing, so 00:05:16.579 --> 00:05:19.069 the fix is just to explain it once. With a tool 00:05:19.069 --> 00:05:21.790 like Loom AI, you record a quick, like, five 00:05:21.790 --> 00:05:24.009 -minute video. You're just talking and showing 00:05:24.009 --> 00:05:26.089 your screen while you do the task. That's it. 00:05:26.189 --> 00:05:29.370 And then the magic happens. Then you click Generate 00:05:29.370 --> 00:05:33.290 SOP. And the AI produces this clean, ready -to 00:05:33.290 --> 00:05:36.050 -use document. It's got numbered steps and even 00:05:36.050 --> 00:05:38.389 pulls screenshots from your video and puts them 00:05:38.389 --> 00:05:40.810 in the right place. So every explanation becomes 00:05:40.810 --> 00:05:44.189 a reusable asset. Yes. And that compounding effect 00:05:44.189 --> 00:05:48.019 is massive. One SOP doesn't change much. but 00:05:48.019 --> 00:05:51.120 50 of them. That turns a founder from an executor 00:05:51.120 --> 00:05:53.860 into an operator. If every explanation becomes 00:05:53.860 --> 00:05:56.180 an asset, how does that fundamentally change 00:05:56.180 --> 00:05:58.660 a founder's job? It turns the founder from a 00:05:58.660 --> 00:06:00.939 continuous executor into a strategic operator 00:06:00.939 --> 00:06:03.300 who builds systems. Okay, let's talk skill number 00:06:03.300 --> 00:06:06.339 four. Sales analysis and voice of customer. You 00:06:06.339 --> 00:06:08.040 call it the copy gold mine, and I think this 00:06:08.040 --> 00:06:10.240 is huge because most marketing just fails. It 00:06:10.240 --> 00:06:12.180 fails because people are guessing what customers 00:06:12.180 --> 00:06:14.220 care about. Right, so you get this copy that's 00:06:14.220 --> 00:06:17.370 polished, but just... Empty. Totally empty. So 00:06:17.370 --> 00:06:19.790 the mining system here is brilliant. You stop 00:06:19.790 --> 00:06:23.610 guessing. You use a tool like Fathom AI on your 00:06:23.610 --> 00:06:26.250 Zoom calls. It records, it transcribes, and it 00:06:26.250 --> 00:06:28.850 captures the exact words from your best customers. 00:06:29.410 --> 00:06:31.610 The ones who bought fast and paid full price. 00:06:32.170 --> 00:06:34.829 They are your gold mines. And then you upload 00:06:34.829 --> 00:06:37.850 maybe 10 or 20 of those transcripts into a dedicated 00:06:37.850 --> 00:06:40.639 Claude project. right into a vault. And then 00:06:40.639 --> 00:06:42.779 you ask it specific questions, like what are 00:06:42.779 --> 00:06:45.680 the top five pain points in their exact words? 00:06:46.000 --> 00:06:48.019 You're looking for those patterns. So you can 00:06:48.019 --> 00:06:51.060 contrast the generic copy, like scale your business 00:06:51.060 --> 00:06:53.839 more efficiently. With what a customer actually 00:06:53.839 --> 00:06:57.019 said on a call, which was, stop being the biggest 00:06:57.019 --> 00:07:00.500 bottleneck in your own company. Wow. That second 00:07:00.500 --> 00:07:03.300 one hits so much harder. It's recognition, and 00:07:03.300 --> 00:07:05.579 recognition drives conversion way better than 00:07:05.579 --> 00:07:08.339 persuasion does. So how does this constant stream 00:07:08.339 --> 00:07:11.000 of customer data prevent marketing from just 00:07:11.000 --> 00:07:13.379 sounding like persuasion? When marketing uses 00:07:13.379 --> 00:07:15.720 their exact words, it sounds like recognition, 00:07:15.779 --> 00:07:18.180 which strongly drives conversion. Alright, skill 00:07:18.180 --> 00:07:21.579 five. Data -driven decision -making. This is 00:07:21.579 --> 00:07:24.480 where we build the AICOO. The problem isn't a 00:07:24.480 --> 00:07:27.879 lack of ideas anymore. It's too many ideas. Decision 00:07:27.879 --> 00:07:31.810 paralysis. Yes. And the solution is kind of counterintuitive. 00:07:32.470 --> 00:07:35.550 You use AI to kill bad ideas. Kill bad ideas. 00:07:35.670 --> 00:07:38.629 Yeah. You set up a chat GPT project to act as 00:07:38.629 --> 00:07:42.089 your strategic advisor, your COO. Its job is 00:07:42.089 --> 00:07:44.490 to find high leverage moves, not just create 00:07:44.490 --> 00:07:47.970 busy work. The framing is so important. The inputs 00:07:47.970 --> 00:07:50.029 have to be real. You have to upload the whole 00:07:50.029 --> 00:07:52.990 picture, your vision, your goals, current projects, 00:07:53.449 --> 00:07:55.610 revenue data. And where your time is actually 00:07:55.610 --> 00:07:57.769 going. That's the critical one. If you skip that, 00:07:57.829 --> 00:07:59.889 the advice is just generic. And then you ask 00:07:59.889 --> 00:08:01.810 it the hard questions. The really hard ones, 00:08:02.189 --> 00:08:04.449 like which projects create the biggest upside 00:08:04.449 --> 00:08:06.889 with the least effort or what should we just 00:08:06.889 --> 00:08:09.709 cut entirely? The AI doesn't have an emotional 00:08:09.709 --> 00:08:12.709 attachment to your pet projects. Whoa. Imagine 00:08:12.709 --> 00:08:15.069 scaling decision making across all that data 00:08:15.069 --> 00:08:17.129 without all the human emotional noise. I mean, 00:08:17.269 --> 00:08:19.680 that's an incredible advantage. It is. Why does 00:08:19.680 --> 00:08:22.360 AI as a COO make better choices than a founder 00:08:22.360 --> 00:08:24.920 sometimes? Because it optimizes for outcomes 00:08:24.920 --> 00:08:27.360 and data, not for excitement or vibes. Which 00:08:27.360 --> 00:08:29.379 brings us to the final piece, the one that ties 00:08:29.379 --> 00:08:32.500 it all together. Skill number six, workflow design. 00:08:32.779 --> 00:08:35.340 This is the operating system itself. Yeah, this 00:08:35.340 --> 00:08:37.360 is the one that quietly controls all the others. 00:08:38.179 --> 00:08:39.840 And the rule is so simple, we keep coming back 00:08:39.840 --> 00:08:42.179 to it. Design the system first, then add the 00:08:42.179 --> 00:08:45.240 AI. On paper first. On paper. Draw the business 00:08:45.240 --> 00:08:48.120 machine. Awareness, sales, delivery, renewal. 00:08:48.460 --> 00:08:51.320 See the structure. AI amplifies that structure. 00:08:51.379 --> 00:08:53.720 It can't create it from confusion. And this is 00:08:53.720 --> 00:08:55.279 where you have to set that boundary rule for 00:08:55.279 --> 00:08:58.039 the whole system. You decide what stays human. 00:08:58.240 --> 00:09:01.039 Right. Taste, trust, the final call. Those are 00:09:01.039 --> 00:09:03.360 human. And then what becomes AI? Repetition, 00:09:03.539 --> 00:09:06.080 analysis, first drafts. And that's why you have 00:09:06.080 --> 00:09:09.740 to assign each tool one specific job. It prevents 00:09:09.740 --> 00:09:11.960 total chaos. So you're not letting every tool 00:09:11.960 --> 00:09:14.659 do everything. God, no. That's a recipe for disaster. 00:09:14.860 --> 00:09:18.059 You need clarity. For this system, chat GPT gets 00:09:18.059 --> 00:09:20.919 strategy and prioritization. Claude gets all 00:09:20.919 --> 00:09:23.399 the writing and editing. Fathom is just for sales 00:09:23.399 --> 00:09:26.700 calls. Loom AI is just for SOPs. And Whisper 00:09:26.700 --> 00:09:28.879 Flow is just for voice to text. And but tools 00:09:28.879 --> 00:09:31.360 change every month. So why is assigning one job 00:09:31.360 --> 00:09:34.100 so essential for consistency? Because mastering 00:09:34.100 --> 00:09:36.580 core skills lets you swap tools without starting 00:09:36.580 --> 00:09:39.139 from zero or creating chaotic overlap. We'll 00:09:39.139 --> 00:09:41.120 be right back after a quick word about something 00:09:41.120 --> 00:09:44.159 we rely on. Midroll sponsor, Read. So when we 00:09:44.159 --> 00:09:46.220 look at how these six skills all work together, 00:09:46.519 --> 00:09:49.039 you start to see this daily flow. The system 00:09:49.039 --> 00:09:50.840 gets smarter without you having to put in more 00:09:50.840 --> 00:09:54.120 manual effort. Give me an example. OK. So you're 00:09:54.120 --> 00:09:57.460 morning. You're checking your strategic pulse 00:09:57.460 --> 00:10:00.519 with chat GPT. That's skill five. And you're 00:10:00.519 --> 00:10:02.740 asking it questions using your voice. That's 00:10:02.740 --> 00:10:06.460 skill one. OK. Then midday, maybe I'm creating 00:10:06.460 --> 00:10:10.279 content. using my Claude content DNA vault. That's 00:10:10.279 --> 00:10:12.399 skill two, and it's based on the concepts you 00:10:12.399 --> 00:10:14.080 voiced out earlier. And then at the end of the 00:10:14.080 --> 00:10:16.960 day. You're delegating a new process with a loom 00:10:16.960 --> 00:10:19.440 video skill three, which automatically builds 00:10:19.440 --> 00:10:21.759 out your operating system skill six. The whole 00:10:21.759 --> 00:10:24.200 thing is this self -refining loop. But founders 00:10:24.200 --> 00:10:26.899 still get stuck. They get stuck at that autocomplete 00:10:26.899 --> 00:10:29.820 level. Let's just quickly hit the four big mistakes 00:10:29.820 --> 00:10:32.340 that keep them there. Number one. Publishing 00:10:32.340 --> 00:10:35.460 raw AI content. It has no personality. It just 00:10:35.460 --> 00:10:38.519 instantly destroys trust. Mistake two, skipping 00:10:38.519 --> 00:10:42.279 the foundation. Automating chaos just makes mess 00:10:42.279 --> 00:10:44.159 faster. You have to build it manually first. 00:10:44.299 --> 00:10:47.379 It's non -negotiable. Third, using AI for everything. 00:10:48.100 --> 00:10:51.590 Strategy and taste. They have to stay human. 00:10:51.870 --> 00:10:53.909 The soul of the business, yeah. And finally, 00:10:54.230 --> 00:10:56.330 chasing tools instead of building skills. If 00:10:56.330 --> 00:10:58.149 you're rebuilding your workflow every time a 00:10:58.149 --> 00:11:00.049 new model comes out, you're always at square 00:11:00.049 --> 00:11:02.149 one. So what's the big takeaway here? The big 00:11:02.149 --> 00:11:04.990 idea? Here's the uncomfortable truth. The gap 00:11:04.990 --> 00:11:07.809 between these AI native businesses and traditional 00:11:07.809 --> 00:11:10.960 ones? It's not closing. It's getting wider. It 00:11:10.960 --> 00:11:14.460 is growing exponentially. Using AI was a cool 00:11:14.460 --> 00:11:18.120 bonus six months ago. Today, building this kind 00:11:18.120 --> 00:11:20.519 of infrastructure is table stakes. It's about 00:11:20.519 --> 00:11:23.759 survival and scale. So since AI is getting so 00:11:23.759 --> 00:11:25.980 good at repetition, analysis, documentation, 00:11:26.899 --> 00:11:29.080 the founder's highest value job has to shift 00:11:29.080 --> 00:11:31.639 away from just doing the work. Completely. So 00:11:31.639 --> 00:11:34.179 here's a thought. If AI handles all the mechanics 00:11:34.179 --> 00:11:37.039 of creation and even strategy analysis, does 00:11:37.039 --> 00:11:39.559 the human founder's highest value job just become 00:11:39.559 --> 00:11:42.500 purely the definition of taste? That is a really 00:11:42.500 --> 00:11:44.419 powerful thought about where all this is heading. 00:11:44.679 --> 00:11:47.019 Our encouragement to you listening is just start. 00:11:47.100 --> 00:11:49.419 Take one skill today. Start building your machine. 00:11:49.759 --> 00:11:51.159 Thank you for joining us for this deep dive. 00:11:51.200 --> 00:11:52.120 We'll see you next time.