WEBVTT 00:00:00.000 --> 00:00:02.140 When you think about traditional business growth, 00:00:02.319 --> 00:00:05.200 the answer was always just hiring more people. 00:00:05.299 --> 00:00:07.379 You needed more sales reps, more accountants. 00:00:07.500 --> 00:00:09.859 You just threw bodies at the problem. It was 00:00:09.859 --> 00:00:11.980 the only way to scale, really. It was the bottleneck. 00:00:12.039 --> 00:00:14.460 But that's changing. It's changing in a pretty 00:00:14.460 --> 00:00:17.519 astonishing way. We're seeing data now where 00:00:17.519 --> 00:00:20.800 a single role, what some are calling AI ops, 00:00:20.940 --> 00:00:25.120 in a small, like 20 -person company, can recover 00:00:25.120 --> 00:00:30.039 between 400 and 800 hours of team capacity. every 00:00:30.039 --> 00:00:32.719 single week. Wow. Okay. So that just reframes 00:00:32.719 --> 00:00:34.539 the entire cost of scale. That's like getting 00:00:34.539 --> 00:00:38.159 five to 10 full -time employees from one strategic 00:00:38.159 --> 00:00:41.549 person directing automation. Exactly. Welcome 00:00:41.549 --> 00:00:43.810 to the Deep Dive. Today we're going way beyond 00:00:43.810 --> 00:00:46.570 just productivity hacks. We're unpacking a framework 00:00:46.570 --> 00:00:50.570 that reveals five really seismic shifts in how 00:00:50.570 --> 00:00:52.570 businesses are creating value right now. And 00:00:52.570 --> 00:00:54.429 these insights, they're coming straight from 00:00:54.429 --> 00:00:56.329 the founders who are actually building these 00:00:56.329 --> 00:00:59.469 new AI native companies. And our mission here 00:00:59.469 --> 00:01:01.369 is to get to the bottom of this new paradox. 00:01:01.630 --> 00:01:04.349 I mean, when anyone, and I really do mean anyone, 00:01:04.469 --> 00:01:07.530 even a smart 12 -year -old, can build a sophisticated 00:01:07.530 --> 00:01:10.450 app. Using plain language. Then what actually 00:01:10.450 --> 00:01:13.069 becomes valuable? If building is easy, what's 00:01:13.069 --> 00:01:15.629 hard? That's the question. That's the question. 00:01:15.790 --> 00:01:19.269 So our roadmap today will cover how teams are 00:01:19.269 --> 00:01:21.189 being restructured, how you defend yourself when 00:01:21.189 --> 00:01:23.670 anyone can copy your features instantly, and 00:01:23.670 --> 00:01:26.290 where your focus really needs to be. So we're 00:01:26.290 --> 00:01:28.969 moving from a world that valued execution. To 00:01:28.969 --> 00:01:31.569 one that values direction and distribution. Okay, 00:01:31.590 --> 00:01:34.200 let's get into it. Shift one. Moving from org 00:01:34.200 --> 00:01:36.519 charts to what you're calling leverage charts. 00:01:36.700 --> 00:01:39.519 Yeah. The old model is all hierarchy departments. 00:01:39.920 --> 00:01:42.340 And like we said, the default solution was always 00:01:42.340 --> 00:01:45.379 hire more people. Right. Throwing bodies at the 00:01:45.379 --> 00:01:47.939 problem. Yeah. That model is just, it's obsolete 00:01:47.939 --> 00:01:50.799 now. So what's the new paradigm? It's nonlinear. 00:01:51.219 --> 00:01:53.620 Instead of a whole department doing routine work, 00:01:53.840 --> 00:01:57.349 you have one strategic leader. And that person 00:01:57.349 --> 00:02:00.250 directs a team of AI agents and tools to deliver 00:02:00.250 --> 00:02:03.189 the outcome. So finance isn't a team of six people 00:02:03.189 --> 00:02:06.629 anymore. No. It's one person overseeing systems 00:02:06.629 --> 00:02:11.289 that handle 98 % of the routine operations. The 00:02:11.289 --> 00:02:14.030 human focus shifts entirely to strategy. Let's 00:02:14.030 --> 00:02:15.830 make that concrete. What does this look like 00:02:15.830 --> 00:02:18.069 for a sales team? Okay, so a traditional sales 00:02:18.069 --> 00:02:22.360 team might need, what, 10 to 14 people? SDRs, 00:02:22.360 --> 00:02:25.580 AEs, engineers. To hit their targets, yeah. With 00:02:25.580 --> 00:02:29.560 this new leverage model, one elite closer, just 00:02:29.560 --> 00:02:32.240 one person, backed by the right AI infrastructure, 00:02:32.560 --> 00:02:35.319 can handle the revenue of that entire team. This 00:02:35.319 --> 00:02:37.979 is that one -person sales machine idea, the infinite 00:02:37.979 --> 00:02:41.180 SDR. Precisely. The AI agent does all the repetitive 00:02:41.180 --> 00:02:43.840 tiring work, the outbound prospecting, personalizing 00:02:43.840 --> 00:02:46.460 messages, qualifying leads. So the human closer 00:02:46.460 --> 00:02:48.819 only steps in when a deal is actually ready. 00:02:48.919 --> 00:02:51.069 When it's hot. And the AI is doing more than 00:02:51.069 --> 00:02:53.310 just prospecting. It's optimizing their calendar. 00:02:53.430 --> 00:02:55.569 It's drafting follow -ups based on the call. 00:02:55.689 --> 00:02:57.490 And it can even provide real -time competitive 00:02:57.490 --> 00:02:59.969 intel during the call. Exactly. The human brings 00:02:59.969 --> 00:03:02.550 judgment, relationship skills. The AI brings 00:03:02.550 --> 00:03:05.469 perfect, tireless execution. Which brings us 00:03:05.469 --> 00:03:08.270 back to that AI ops role. Their whole job is 00:03:08.270 --> 00:03:10.409 to find the repetitive work. And eliminate it. 00:03:10.509 --> 00:03:12.389 And then train the rest of the team to become 00:03:12.389 --> 00:03:14.969 these high -level directors of AI systems. Whoa. 00:03:15.210 --> 00:03:18.090 I mean, just imagine scaling that. Taking that 00:03:18.090 --> 00:03:21.789 400 to 800 hours a week of recovered time across 00:03:21.789 --> 00:03:24.750 a company of, say, 100 people. It completely 00:03:24.750 --> 00:03:27.449 changes what's possible. It redefines growth 00:03:27.449 --> 00:03:29.750 itself overnight. So you have to stop asking, 00:03:29.889 --> 00:03:32.520 who should I hire? and start asking, what can 00:03:32.520 --> 00:03:36.240 we build to create leverage? So if AI is handling 00:03:36.240 --> 00:03:38.740 all the execution, what is the new core skill? 00:03:38.979 --> 00:03:42.159 What separates success from, you know, disappointment? 00:03:42.719 --> 00:03:44.960 Directing the systems. It's having the vision 00:03:44.960 --> 00:03:47.400 and judgment to guide the AI. And that leads 00:03:47.400 --> 00:03:50.020 right into shift two, from being a doer to being 00:03:50.020 --> 00:03:52.120 a director. It does, because this is where a 00:03:52.120 --> 00:03:54.789 lot of people get stuck. They try chat GPT. you 00:03:54.789 --> 00:03:57.090 know, for some generic blog post, and they walk 00:03:57.090 --> 00:03:59.430 away thinking it's overhyped. Right. But the 00:03:59.430 --> 00:04:01.889 problem isn't the AI. The problem is they're 00:04:01.889 --> 00:04:04.509 using it like a simple tool, not directing it 00:04:04.509 --> 00:04:06.389 like a team. They're treating it like a fancy 00:04:06.389 --> 00:04:08.509 calculator instead of a thinking partner. You've 00:04:08.509 --> 00:04:11.650 got it, which points to this massive shift in 00:04:11.650 --> 00:04:13.849 how we need to spend our time. The time flip. 00:04:14.009 --> 00:04:18.610 Yeah. The old split was maybe 90 % execution. 00:04:19.579 --> 00:04:22.560 Doing the actual work in 10 % direction. Strategy, 00:04:22.560 --> 00:04:24.620 planning. Right. Now, you have to flip that. 00:04:24.720 --> 00:04:27.920 It has to be 80 % or 90 % direction. Designing 00:04:27.920 --> 00:04:30.540 the system, setting strategy, quality control, 00:04:30.819 --> 00:04:33.899 only 10 % to 20 % is execution. That's a huge 00:04:33.899 --> 00:04:35.699 reversal. You mentioned a film director analogy. 00:04:35.980 --> 00:04:38.540 I think it's the perfect analogy. A film director's 00:04:38.540 --> 00:04:41.620 value isn't in holding the camera. No. It's their 00:04:41.620 --> 00:04:45.279 vision. orchestrating all the resources, judging 00:04:45.279 --> 00:04:47.560 the quality. The human's job is now to design 00:04:47.560 --> 00:04:49.779 the world in which the AI can execute perfectly. 00:04:50.000 --> 00:04:52.100 And honestly, I still wrestle with prompt drift 00:04:52.100 --> 00:04:53.879 myself. Oh, yeah. You know, where you give the 00:04:53.879 --> 00:04:56.540 AI a complex task and halfway through it just 00:04:56.540 --> 00:04:58.660 kind of forgets the most important instruction. 00:04:58.699 --> 00:05:01.500 It's a real thing. It's tough to consistently 00:05:01.500 --> 00:05:04.800 provide all the right context, but that struggle, 00:05:05.000 --> 00:05:07.829 that iteration. That is the new job. So how do 00:05:07.829 --> 00:05:09.970 we get better at directing? You had a three -step 00:05:09.970 --> 00:05:12.089 framework for this. Yeah, it's pretty straightforward. 00:05:12.410 --> 00:05:14.889 First, you have to audit your time. For two weeks, 00:05:15.009 --> 00:05:17.790 just track your actual execution versus direction 00:05:17.790 --> 00:05:20.589 ratio. Be honest with yourself. Be ruthless. 00:05:21.589 --> 00:05:24.970 Second, identify the opportunities. Find that 00:05:24.970 --> 00:05:27.550 20 % of repetitive work that's eating 80 % of 00:05:27.550 --> 00:05:30.149 your time. That's your first target. Okay. And 00:05:30.149 --> 00:05:33.300 finally, learn to direct. Pick a couple of those 00:05:33.300 --> 00:05:36.519 tasks, write really clear system prompts, give 00:05:36.519 --> 00:05:39.819 the AI good examples, and just iterate. The goal 00:05:39.819 --> 00:05:42.560 is to flip your ratio over, say, the next year 00:05:42.560 --> 00:05:45.199 or 18 months. Okay, so we can build things instantly 00:05:45.199 --> 00:05:48.120 with good direction. But if everyone can build 00:05:48.120 --> 00:05:49.939 and copy things that quickly, how do you build 00:05:49.939 --> 00:05:51.879 any kind of long -term competitive advantage? 00:05:52.379 --> 00:05:55.459 By using your own unique data to build a moat 00:05:55.459 --> 00:05:57.860 that makes your product smarter over time. Which 00:05:57.860 --> 00:06:00.720 is shift three. From feature defenses to data 00:06:00.720 --> 00:06:03.040 defenses. Exactly. The old feature race is just, 00:06:03.220 --> 00:06:06.060 it's over. Competing on features is suicidal 00:06:06.060 --> 00:06:08.459 now. It is. You used to get maybe a three to 00:06:08.459 --> 00:06:10.480 six month head start with a new feature. But 00:06:10.480 --> 00:06:14.180 now with AI coding tools like Cursor or Copilot, 00:06:14.279 --> 00:06:16.939 a competitor can copy your feature in weeks. 00:06:17.100 --> 00:06:20.319 The moat is gone. Instantly. So what actually 00:06:20.319 --> 00:06:22.720 protects you? It's the private data your customers 00:06:22.720 --> 00:06:25.420 generate. It's the unique industry data you collect. 00:06:25.660 --> 00:06:28.480 These create learning loops. Like with ChatGPT, 00:06:28.639 --> 00:06:30.680 it gets better for you because it remembers your 00:06:30.680 --> 00:06:33.860 conversations. And no competitor can replicate 00:06:33.860 --> 00:06:37.300 that specific personal feedback loop. That's 00:06:37.300 --> 00:06:40.060 the moat. So how does a company that's not, you 00:06:40.060 --> 00:06:43.399 know, open AI build one of these? You have to 00:06:43.399 --> 00:06:45.240 be intentional about it. There's a three -step 00:06:45.240 --> 00:06:49.149 framework. Step one is... You have to clean your 00:06:49.149 --> 00:06:51.490 data. The unglamorous part. The most critical 00:06:51.490 --> 00:06:53.449 part. Garbage in, garbage out. You have to fix 00:06:53.449 --> 00:06:56.069 inaccuracies, validate everything. Okay, so your 00:06:56.069 --> 00:07:00.370 data is clean. Step two. AI analysis. Use AI 00:07:00.370 --> 00:07:03.350 to find the patterns that humans would just miss. 00:07:03.769 --> 00:07:06.949 It can see correlations across years of data 00:07:06.949 --> 00:07:10.029 that are invisible to us. So it discovers a proprietary 00:07:10.029 --> 00:07:12.860 insight for you. It does. And then... Step three 00:07:12.860 --> 00:07:14.699 is you close the loop with AI suggested next 00:07:14.699 --> 00:07:17.579 steps. AI monitors data, sees a trend, recommends 00:07:17.579 --> 00:07:20.120 an action, human approves it, the AI executes. 00:07:20.139 --> 00:07:22.120 And the whole system gets smarter from the outcome. 00:07:22.339 --> 00:07:24.740 That's the loop. The brutal reality is the company 00:07:24.740 --> 00:07:26.540 with the best data and the fastest learning loop 00:07:26.540 --> 00:07:29.800 is going to win. So. If AI can handle features 00:07:29.800 --> 00:07:32.699 and data analysis, what does that mean for all 00:07:32.699 --> 00:07:35.899 the repetitive internal stuff, the back office 00:07:35.899 --> 00:07:39.439 functions? It means they are also ripe for near 00:07:39.439 --> 00:07:42.839 total automation. We're talking up to 98%. And 00:07:42.839 --> 00:07:45.579 that takes us to shift four, the autonomous back 00:07:45.579 --> 00:07:49.120 office. For decades, we've staffed things like 00:07:49.120 --> 00:07:52.160 finance, HR, and legal with full -time employees. 00:07:52.279 --> 00:07:54.019 Right, assuming you needed human judgment for 00:07:54.019 --> 00:07:55.759 every little thing. But if you actually look 00:07:55.759 --> 00:07:57.920 at the work. Most of it isn't judgment. It's 00:07:57.920 --> 00:08:00.899 just applying a policy to a request. It's pattern 00:08:00.899 --> 00:08:03.160 -based execution. Which is exactly what AI is 00:08:03.160 --> 00:08:06.560 built for. The new reality is 98 % automation 00:08:06.560 --> 00:08:09.439 with just 2 % human oversight for the exceptions. 00:08:09.740 --> 00:08:11.819 Let's go back to that finance example. The old 00:08:11.819 --> 00:08:13.379 way is someone gets an invoice, they manually 00:08:13.379 --> 00:08:15.920 type it in, manually check it. So much friction. 00:08:16.279 --> 00:08:19.139 Now the invoice arrives, an AI extracts the data, 00:08:19.240 --> 00:08:21.240 checks it against your expense policy and your 00:08:21.240 --> 00:08:23.639 CRM. Instantly. And if it matches the rules? 00:08:23.819 --> 00:08:26.319 It processes and reconciles it. Done. The work 00:08:26.319 --> 00:08:28.899 of days now takes seconds. So the key to implementing 00:08:28.899 --> 00:08:31.360 this is just having really clear documented policies. 00:08:31.800 --> 00:08:35.039 The AI needs rules. It needs access to the data. 00:08:35.279 --> 00:08:37.759 And it needs a clear point where it hands off 00:08:37.759 --> 00:08:40.320 an exception to a human. The core principle seems 00:08:40.320 --> 00:08:43.919 to be exceptions deserve people, patterns deserve 00:08:43.919 --> 00:08:47.120 code. That's it. Automate the 98 % so your people 00:08:47.120 --> 00:08:49.809 can focus their brilliance on... The 2 % that 00:08:49.809 --> 00:08:51.830 really matters. Okay, so let's put it all together. 00:08:52.090 --> 00:08:55.009 The product is easy to build. The data is your 00:08:55.009 --> 00:08:58.470 defense. The back office is automated. What is 00:08:58.470 --> 00:09:01.230 the one thing left that determines if you win 00:09:01.230 --> 00:09:04.570 or lose? Distribution. Yeah. Your ability to 00:09:04.570 --> 00:09:06.809 actually reach the customers who need what you've 00:09:06.809 --> 00:09:09.309 built. And that's the final and maybe the most 00:09:09.309 --> 00:09:12.330 important change. Shift five. From a development 00:09:12.330 --> 00:09:15.250 advantage to a distribution advantage. For 30 00:09:15.250 --> 00:09:16.950 years, the company with the smartest engineers 00:09:16.950 --> 00:09:19.919 won. That was the game. But that advantage is 00:09:19.919 --> 00:09:23.940 basically gone. It's been destroyed. I mean, 00:09:23.980 --> 00:09:26.000 code quality still matters for stability, of 00:09:26.000 --> 00:09:28.080 course, but it's not a competitive edge anymore. 00:09:28.279 --> 00:09:31.259 When a kid can build your core feature set, code 00:09:31.259 --> 00:09:33.820 has been competitively neutralized. So the entire 00:09:33.820 --> 00:09:36.419 game shifts to distribution, reaching people, 00:09:36.659 --> 00:09:39.340 building trust, converting attention. And that 00:09:39.340 --> 00:09:41.519 distribution really comes in three main forms. 00:09:41.700 --> 00:09:43.620 You've got your own audiences, like an email 00:09:43.620 --> 00:09:46.039 list, that's the most valuable. Then there's 00:09:46.039 --> 00:09:47.759 paid distribution ads, which are getting more 00:09:47.759 --> 00:09:49.960 and more expensive. And finally, partnerships. 00:09:50.440 --> 00:09:52.340 Right. Working with people who already have the 00:09:52.340 --> 00:09:54.500 audience you need to reach. So there's a playbook 00:09:54.500 --> 00:09:58.399 for this new world. There is. Step one. You have 00:09:58.399 --> 00:10:00.659 to build distribution into your business model 00:10:00.659 --> 00:10:03.440 from day one. You have to ask, how will we reach 00:10:03.440 --> 00:10:06.600 customers before you write a line of code? Because 00:10:06.600 --> 00:10:08.919 great distribution will beat a great product. 00:10:09.059 --> 00:10:12.710 Every time. Every time. Step two. Attach your 00:10:12.710 --> 00:10:15.730 brand to one clear problem. People buy clarity, 00:10:15.870 --> 00:10:18.669 not a list of features. Be super clear about 00:10:18.669 --> 00:10:20.789 the outcome you deliver. And step three. Pre 00:10:20.789 --> 00:10:23.409 -sell before you build. Go get commitments or 00:10:23.409 --> 00:10:25.570 at least real interest before you spend months 00:10:25.570 --> 00:10:27.649 building something nobody actually wants. It 00:10:27.649 --> 00:10:30.669 de -risks everything. Completely. And if you 00:10:30.669 --> 00:10:32.889 don't have an audience, the shortcut is to partner 00:10:32.889 --> 00:10:35.809 with creators in your space and share the revenue. 00:10:36.129 --> 00:10:38.649 It's that simple. The meta lesson here is that 00:10:38.649 --> 00:10:41.710 when anyone can build, the only edge left. is 00:10:41.710 --> 00:10:44.029 who can reach that's the whole game now what 00:10:44.029 --> 00:10:47.590 an incredible breakdown so to recap the big idea 00:10:47.590 --> 00:10:51.610 here ai is fundamentally devaluing the work of 00:10:51.610 --> 00:10:54.970 pure execution it is it's destroying the value 00:10:54.970 --> 00:10:58.389 of building simple features success now demands 00:10:58.389 --> 00:11:00.970 a total transformation into a strategic director 00:11:00.970 --> 00:11:03.289 right you have to build data modes you have to 00:11:03.289 --> 00:11:05.350 automate all the pattern -based work and you 00:11:05.350 --> 00:11:08.570 have to master distribution it's a uh a complete 00:11:08.570 --> 00:11:11.250 rewrite of business strategy And this isn't something 00:11:11.250 --> 00:11:13.110 that's coming in a few years. It is happening 00:11:13.110 --> 00:11:15.470 right now. The choice for you, the listener, 00:11:15.629 --> 00:11:18.370 is pretty stark. Either lead this transformation 00:11:18.370 --> 00:11:20.769 in your organization or you're going to be disrupted 00:11:20.769 --> 00:11:22.529 by someone who does. The time to act is getting 00:11:22.529 --> 00:11:25.090 very short. So here's the challenge for you this 00:11:25.090 --> 00:11:27.600 week. Audit your business through the lens of 00:11:27.600 --> 00:11:30.059 these five shifts. And then, and this is the 00:11:30.059 --> 00:11:32.700 actionable part, pick just one high -impact area 00:11:32.700 --> 00:11:34.879 to start with. Maybe it's defining your first 00:11:34.879 --> 00:11:36.960 data loop, or maybe it's just doing that time 00:11:36.960 --> 00:11:39.700 audit to justify hiring an AI ops role. Whatever 00:11:39.700 --> 00:11:42.820 it is, start this week. Because every week you 00:11:42.820 --> 00:11:45.059 wait, the gap gets wider and harder to close. 00:11:45.840 --> 00:11:47.120 Thank you for this deep dive.