WEBVTT 00:00:00.000 --> 00:00:03.919 We wait for perfect AI to replace humans. We 00:00:03.919 --> 00:00:06.379 stay completely paralyzed by that grand illusion. 00:00:06.679 --> 00:00:09.839 Beat. Meanwhile, agile teams are quietly taking 00:00:09.839 --> 00:00:13.039 over everything. They just let imperfect AI do 00:00:13.039 --> 00:00:16.300 the chores. Two sec sirens. It's the ultimate 00:00:16.300 --> 00:00:19.359 irony of modern business today. Welcome to the 00:00:19.359 --> 00:00:21.019 Deep Dive. I'm really glad you're here with us. 00:00:21.300 --> 00:00:24.460 Today, we're exploring a deeply fascinating blueprint. 00:00:24.820 --> 00:00:27.440 It's Maxan's successful AI project framework. 00:00:27.780 --> 00:00:31.440 This is the definitive March 2026 survival guide. 00:00:31.620 --> 00:00:33.460 Yeah, we're going to completely dismantle the 00:00:33.460 --> 00:00:36.939 hiring versus AI debate and explore the mathematical 00:00:36.939 --> 00:00:39.960 beauty of... you know, good enough. We'll also 00:00:39.960 --> 00:00:42.140 look at live automated business pipelines and 00:00:42.140 --> 00:00:44.380 we'll reveal the fatal mistakes that Sync projects. 00:00:44.679 --> 00:00:46.320 It's a fundamentally different way to approach 00:00:46.320 --> 00:00:49.119 work. You really have to rewire your entire brain. 00:00:49.320 --> 00:00:51.299 So let's start with a crucial mindset shift. 00:00:51.439 --> 00:00:53.859 To understand how to use AI effectively today, 00:00:53.979 --> 00:00:55.659 we first have to stop thinking about human job 00:00:55.659 --> 00:00:57.539 titles. Right. Most companies fail right at the 00:00:57.539 --> 00:00:59.479 starting line. They sit down and ask a terrible 00:00:59.479 --> 00:01:02.619 question. They ask, can AI replace my senior 00:01:02.619 --> 00:01:04.340 copywriter? Which is totally the wrong way to 00:01:04.340 --> 00:01:06.590 look at it. Oh, totally the wrong question. A 00:01:06.590 --> 00:01:08.489 job title is just a fictional bundle. It's a 00:01:08.489 --> 00:01:11.810 bundle of dozens of tiny tasks. Trying to automate 00:01:11.810 --> 00:01:14.849 a whole role leads to panic. It causes complete 00:01:14.849 --> 00:01:17.909 operational paralysis in an office. You can't 00:01:17.909 --> 00:01:20.349 just replace a broad human title. You can only 00:01:20.349 --> 00:01:25.079 replace specific, isolated daily tasks. Max Anne's 00:01:25.079 --> 00:01:28.439 2026 framework introduces the task -based framework. 00:01:28.739 --> 00:01:31.299 It deliberately breaks jobs down into single 00:01:31.299 --> 00:01:33.840 actions. Yeah, then you ask one highly specific 00:01:33.840 --> 00:01:37.060 focus question. Is AI good enough for this exact 00:01:37.060 --> 00:01:39.959 step today? This leads to a very practical, actionable 00:01:39.959 --> 00:01:42.659 step. It's called the task audit. I absolutely 00:01:42.659 --> 00:01:44.939 love the task audit exercise. It's incredibly 00:01:44.939 --> 00:01:46.939 revealing for any struggling business owner. 00:01:47.260 --> 00:01:49.060 You just pick one single department to start. 00:01:49.180 --> 00:01:51.040 Let's say you choose your internal content team. 00:01:51.099 --> 00:01:53.000 You list every single human action they take. 00:01:53.099 --> 00:01:55.719 Then you literally score each individual microstep. 00:01:55.920 --> 00:01:58.560 Yes, no, or maybe for its AI readiness. Let's 00:01:58.560 --> 00:02:01.200 look at a real content team's workflow. Humans 00:02:01.200 --> 00:02:04.280 absolutely still need to record raw video. The 00:02:04.280 --> 00:02:06.200 human provides the original thought and charisma. 00:02:06.560 --> 00:02:10.120 Exactly. That biological spark is incredibly 00:02:10.120 --> 00:02:12.740 hard to fake. But after that recording stops, 00:02:13.020 --> 00:02:15.990 the process shifts. AI tools like Whisper or 00:02:15.990 --> 00:02:18.669 Gemini step in. They transcribe the entire raw 00:02:18.669 --> 00:02:21.270 audio file instantly. Then human video editors 00:02:21.270 --> 00:02:24.330 don't manually cut silence. Automated tools handle 00:02:24.330 --> 00:02:26.849 that tedious dead air instantly. Right. And a 00:02:26.849 --> 00:02:29.830 copywriter doesn't draft social captions. Modern 00:02:29.830 --> 00:02:33.169 models like GPT 5 .4 or Claude Sonnet do. They 00:02:33.169 --> 00:02:35.270 write dozens of platform -specific variations 00:02:35.270 --> 00:02:38.289 in seconds. Then tools like Zapier schedule the 00:02:38.289 --> 00:02:40.469 actual posts. The human basically becomes the 00:02:40.469 --> 00:02:42.370 director, not the crew. Yeah, it's like stacking 00:02:42.370 --> 00:02:44.840 Lego blocks of data. You break the big job into 00:02:44.840 --> 00:02:47.259 tiny plastic bricks. Then you rebuild the entire 00:02:47.259 --> 00:02:50.020 workflow brick by brick. You swap out the fragile 00:02:50.020 --> 00:02:52.879 human blocks for AI blocks. Let me ask a practical 00:02:52.879 --> 00:02:54.919 question here. What if a workflow is already 00:02:54.919 --> 00:02:57.259 a tangled mess? Won't breaking it down just give 00:02:57.259 --> 00:02:59.780 us messy Lego blocks? Oh, absolutely. Automating 00:02:59.780 --> 00:03:02.539 chaos just scales your errors much faster. You 00:03:02.539 --> 00:03:04.580 must manually clean the business process first. 00:03:04.759 --> 00:03:06.740 Figure out exactly how the data should logically 00:03:06.740 --> 00:03:09.560 move. Only then do you plug in the AI automation. 00:03:09.919 --> 00:03:12.259 Clean the messy process first or automate chaos? 00:03:12.759 --> 00:03:15.789 Precisely. And once you actually map these specific 00:03:15.789 --> 00:03:19.250 tasks out, you realize something quite profound 00:03:19.250 --> 00:03:22.389 about human labor. Many tasks simply don't require 00:03:22.389 --> 00:03:25.830 human perfection. Two -sec silence. That brings 00:03:25.830 --> 00:03:28.370 us to the new operational standard, the good 00:03:28.370 --> 00:03:30.830 -enough standard of the 2026 landscape. Right, 00:03:30.889 --> 00:03:33.770 because waiting for perfect AI is a massive risk. 00:03:34.110 --> 00:03:36.550 Agile teams don't wait for flawless androids. 00:03:36.610 --> 00:03:38.830 They look at a tool and ask a simple question. 00:03:39.009 --> 00:03:41.990 Is this good enough to use right now? Speed to 00:03:41.990 --> 00:03:44.789 market is the ultimate defining edge. The defining 00:03:44.789 --> 00:03:47.330 metric is no longer pure, unadulterated quality. 00:03:47.530 --> 00:03:50.430 It's now entirely about your quality per dollar. 00:03:50.629 --> 00:03:52.669 We really have to look at the actual math here. 00:03:52.870 --> 00:03:55.389 The baseline math is honestly staggering to consider. 00:03:55.610 --> 00:03:58.069 Let's look at a standard basic social media caption. 00:03:58.229 --> 00:04:01.830 An AI caption might only be 80 % polished, but 00:04:01.830 --> 00:04:04.129 it costs your business exactly one single cent. 00:04:04.250 --> 00:04:06.050 And it takes just three seconds to fully generate. 00:04:06.439 --> 00:04:09.080 Compare that to a human copywriter's daily effort. 00:04:09.259 --> 00:04:11.520 A human might take 20 full minutes for perfection. 00:04:11.879 --> 00:04:14.360 Yeah, and that 20 minutes cost the company $25. 00:04:14.580 --> 00:04:18.540 So 80 % quality at one single cent versus 100 00:04:18.540 --> 00:04:22.339 % quality at 25 steep dollars. It's a mathematical 00:04:22.339 --> 00:04:24.980 winner for most business cases. You simply can't 00:04:24.980 --> 00:04:28.019 ignore that level of extreme efficiency. Modern 00:04:28.019 --> 00:04:30.339 language models are basically indistinguishable 00:04:30.339 --> 00:04:33.829 from humans anyway. They handle 90 % of repetitive 00:04:33.829 --> 00:04:37.670 writing flawlessly. But I do have to push back 00:04:37.670 --> 00:04:41.389 on this. Isn't settling for 80 % polished a dangerous 00:04:41.389 --> 00:04:44.069 race? Doesn't that eventually destroy a premium 00:04:44.069 --> 00:04:46.350 brand's reputation? Well, not if you structure 00:04:46.350 --> 00:04:48.829 the workflow properly. The 80 % is just for the 00:04:48.829 --> 00:04:51.850 heavy lifting. The lean human layer adds the 00:04:51.850 --> 00:04:54.209 final strategic polish. You get the quality, 00:04:54.269 --> 00:04:56.970 but skip the grunt work. The AI writes the rough 00:04:56.970 --> 00:05:00.500 draft. The human makes it sing. AI handles heavy 00:05:00.500 --> 00:05:03.100 lifting while humans add strategic power. Exactly. 00:05:03.300 --> 00:05:05.779 It's a powerful collaboration, not a sad compromise. 00:05:06.160 --> 00:05:08.180 So we accept good enough for the heavy lifting. 00:05:08.379 --> 00:05:11.019 But what does this machinery look like in motion? 00:05:11.279 --> 00:05:13.519 Let's look at the actual live business pipelines. 00:05:13.879 --> 00:05:16.040 The modern content engine is a beautiful thing. 00:05:16.139 --> 00:05:19.339 You sit down and record one long video. That 00:05:19.339 --> 00:05:22.620 is your only required manual input step. From 00:05:22.620 --> 00:05:25.459 there, the automated machinery completely takes 00:05:25.459 --> 00:05:28.300 over. It's a totally seamless transition of digital 00:05:28.300 --> 00:05:31.379 labor. The AI automatically transcribes the entire 00:05:31.379 --> 00:05:34.379 video file first. Then it actively scans the 00:05:34.379 --> 00:05:36.879 text for high engagement hooks. Right. It finds 00:05:36.879 --> 00:05:39.399 those moments and cuts short clips. It removes 00:05:39.399 --> 00:05:41.839 filler words and formats the video aspect ratio. 00:05:42.100 --> 00:05:45.160 Then it writes highly specific captions for different 00:05:45.160 --> 00:05:48.300 platforms. LinkedIn gets a somewhat serious professional 00:05:48.300 --> 00:05:51.259 tone. Instagram gets a much more casual, punchy 00:05:51.259 --> 00:05:53.500 tone. Then it auto publishes everything on a 00:05:53.500 --> 00:05:55.920 predetermined schedule. What's fascinating is 00:05:55.920 --> 00:05:58.740 the system learns over time. It analyzes exactly 00:05:58.740 --> 00:06:01.279 which hooks get the most engagement. Yeah, it 00:06:01.279 --> 00:06:03.279 adapts future selections based on real world 00:06:03.279 --> 00:06:05.600 data. But this isn't just for creative marketing 00:06:05.600 --> 00:06:07.980 content. Back office operations use the exact 00:06:07.980 --> 00:06:10.500 same underlying logic. And those transformations 00:06:10.500 --> 00:06:14.000 are arguably even more lucrative. Voice AI handles 00:06:14.000 --> 00:06:16.699 lead intake completely effortlessly. It talks 00:06:16.699 --> 00:06:19.480 to prospects and instantly updates the CRM. It 00:06:19.480 --> 00:06:22.500 categorizes the caller's urgency without any 00:06:22.500 --> 00:06:25.759 human intervention. The B2B proposal drafting 00:06:25.759 --> 00:06:29.300 process is completely revolutionized too. AI 00:06:29.300 --> 00:06:32.019 pulls industry trends and budget data from the 00:06:32.019 --> 00:06:35.399 CRM. It drafts a customized proposal in just 00:06:35.399 --> 00:06:38.620 15 minutes. That exact same process used to take 00:06:38.620 --> 00:06:43.079 two solid hours. Exactly. AI chat bots also handle 00:06:43.079 --> 00:06:45.839 all tier one FAQ support tickets. Yeah. They 00:06:45.839 --> 00:06:47.839 answer routine questions about open invoices 00:06:47.839 --> 00:06:51.100 or timelines. Humans are left strictly for complex, 00:06:51.180 --> 00:06:53.199 nuanced decisions. They handle the edge cases 00:06:53.199 --> 00:06:55.800 and the actual relationship building. But with 00:06:55.800 --> 00:06:57.439 all these different tools talking to each other, 00:06:57.519 --> 00:06:59.399 how does the business owner not get completely 00:06:59.399 --> 00:07:01.740 overwhelmed? It sounds incredibly complicated 00:07:01.740 --> 00:07:03.879 to maintain all this tech. Well, they use integration 00:07:03.879 --> 00:07:06.779 platforms like Zapier as a simple bridge. You 00:07:06.779 --> 00:07:08.680 don't need to write any complex custom code. 00:07:08.819 --> 00:07:11.180 You just use basic logic to connect them together. 00:07:11.240 --> 00:07:13.060 These are essentially what we call agentic workflows. 00:07:13.439 --> 00:07:15.420 How exactly would you define agentic workflows 00:07:15.420 --> 00:07:18.500 for our listeners? AI systems autonomously linking 00:07:18.500 --> 00:07:21.240 actions together to achieve goals. They do the 00:07:21.240 --> 00:07:23.579 intelligent routing for you. Simple logic bridges 00:07:23.579 --> 00:07:26.319 linking autonomous AI systems together seamlessly. 00:07:26.779 --> 00:07:29.399 Spot on. It keeps the tech stack manageable and 00:07:29.399 --> 00:07:32.379 invisible. We've talked extensively about automating 00:07:32.379 --> 00:07:35.180 text and data pipelines, but the most disruptive 00:07:35.180 --> 00:07:37.899 pipeline taking over is entirely different. It 00:07:37.899 --> 00:07:39.800 doesn't use a traditional screen or keyboard. 00:07:40.240 --> 00:07:43.259 Let's discuss the rapidly disappearing user interface. 00:07:43.680 --> 00:07:46.399 This is the massive shift to conversational voice 00:07:46.399 --> 00:07:49.660 AI. It's completely rewiring the baseline customer 00:07:49.660 --> 00:07:53.259 experience. Tools like 11 Labs allow customers 00:07:53.259 --> 00:07:55.720 to call a standard number. They speak completely 00:07:55.720 --> 00:07:58.600 naturally to a highly responsive AI agent. There 00:07:58.600 --> 00:08:00.920 are no wait times. and no rigid numeric menus. 00:08:01.439 --> 00:08:04.839 The system actually understands nuanced, messy 00:08:04.839 --> 00:08:07.980 human context. It can book calendar calls or 00:08:07.980 --> 00:08:11.040 route to humans dynamically. Phone calls are 00:08:11.040 --> 00:08:13.100 still the highest converting touchpoint in business. 00:08:13.420 --> 00:08:16.259 Whoa, imagine scaling to a billion queries. Just 00:08:16.259 --> 00:08:18.759 perfectly pleasant, instant voice responses for 00:08:18.759 --> 00:08:20.879 everyone globally. It's honestly mind -blowing 00:08:20.879 --> 00:08:22.720 to think about that volume. The AI never gets 00:08:22.720 --> 00:08:24.800 tired or annoyed by repetitive questions. If 00:08:24.800 --> 00:08:27.079 your competitor's AI picks up the phone instantly 00:08:27.079 --> 00:08:29.939 while you send your worn, eager leads to a voicemail, 00:08:30.019 --> 00:08:32.080 you will lose that business every single time. 00:08:32.279 --> 00:08:35.659 Oh, 100%. The window to adopt this technology 00:08:35.659 --> 00:08:39.279 is right now. Building an AI voice agent is surprisingly 00:08:39.279 --> 00:08:42.059 easy today. You don't need a degree in computer 00:08:42.059 --> 00:08:44.919 science. The tools are highly accessible and 00:08:44.919 --> 00:08:47.600 often entirely free. It just requires mapping 00:08:47.600 --> 00:08:50.960 the desired conversation flow properly. But does 00:08:50.960 --> 00:08:53.500 an AI voice agent actually make people feel heard? 00:08:53.940 --> 00:08:56.220 Or do customers just get frustrated they aren't 00:08:56.220 --> 00:08:58.960 talking to humans? Speed and accuracy basically 00:08:58.960 --> 00:09:01.629 always win out in the end. Customers heavily 00:09:01.629 --> 00:09:04.450 prefer an instant, highly competent AI assistant. 00:09:04.629 --> 00:09:06.710 It's much better than waiting on hold forever. 00:09:06.950 --> 00:09:09.769 Customers prefer instant, competent AI over waiting 00:09:09.769 --> 00:09:12.850 for humans. Every single time. We value our limited 00:09:12.850 --> 00:09:15.429 time above absolutely everything else. Sponsor. 00:09:15.450 --> 00:09:17.850 It all sounds a bit like futuristic science fiction. 00:09:18.190 --> 00:09:20.350 Business owners usually assume science fiction 00:09:20.350 --> 00:09:23.149 is highly expensive. But the actual cost math 00:09:23.149 --> 00:09:26.210 tells a very different story. That is, if you 00:09:26.210 --> 00:09:28.490 manage to avoid the hidden traps. The cost math 00:09:28.490 --> 00:09:30.909 is probably the most compelling part. Setting 00:09:30.909 --> 00:09:34.190 up a basic workflow might cost $2 ,000. A highly 00:09:34.190 --> 00:09:37.370 complex agency build might hit $20 ,000. Your 00:09:37.370 --> 00:09:39.970 monthly API subscriptions run a few hundred bucks. 00:09:40.190 --> 00:09:42.429 But look at what that initial investment actually 00:09:42.429 --> 00:09:46.049 replaces. It replaces mundane roles costing $4 00:09:46.049 --> 00:09:48.769 ,000 to $15 ,000 monthly. The financial break 00:09:48.769 --> 00:09:51.750 -even point happens incredibly fast. Usually 00:09:51.750 --> 00:09:53.870 you see a return in just one to three months. 00:09:54.509 --> 00:09:57.269 After that break -even point, your business fundamentally 00:09:57.269 --> 00:10:01.169 changes. Every automated step becomes pure, unadulterated 00:10:01.169 --> 00:10:03.649 margin expansion. You radically increase output 00:10:03.649 --> 00:10:06.590 with significantly fewer daily resources. But 00:10:06.590 --> 00:10:08.070 we really have to look at the fatal mistakes. 00:10:08.370 --> 00:10:10.889 This is why AI projects actually fail in the 00:10:10.889 --> 00:10:13.309 early stages. The underlying technology usually 00:10:13.309 --> 00:10:16.029 works just fine. The human execution is where 00:10:16.029 --> 00:10:18.730 most companies stumble, beat. Over -engineering 00:10:18.730 --> 00:10:20.970 is an absolute massive trap. Right. People love 00:10:20.970 --> 00:10:23.230 to jump straight to complex custom software builds. 00:10:23.370 --> 00:10:25.320 They think, they need proprietary coded software 00:10:25.320 --> 00:10:28.080 to win. But simple, easily connected tools usually 00:10:28.080 --> 00:10:30.480 work much better. You start simple and upgrade 00:10:30.480 --> 00:10:33.019 your stack only when necessary. The second fatal 00:10:33.019 --> 00:10:35.840 mistake is skipping the human review layer. AI 00:10:35.840 --> 00:10:38.000 without a human check inevitably drifts over 00:10:38.000 --> 00:10:40.860 time. The machine output slowly becomes stale 00:10:40.860 --> 00:10:44.320 or robotic or off -brand to sex silence. I still 00:10:44.320 --> 00:10:46.419 wrestle with prompt drift myself. Oh, it's a 00:10:46.419 --> 00:10:48.299 very real struggle. You build a great prompt. 00:10:48.990 --> 00:10:51.269 And a few weeks later, the tone just suddenly 00:10:51.269 --> 00:10:54.509 shifts. It drifts away from your core brand voice. 00:10:54.730 --> 00:10:57.029 It happens to absolutely everyone who uses these 00:10:57.029 --> 00:11:00.570 tools. Language models are basically just giant 00:11:00.570 --> 00:11:03.590 statistical prediction engines. Without a rigid 00:11:03.590 --> 00:11:06.769 framework, they slowly regress to the mean. They 00:11:06.769 --> 00:11:09.389 start sounding like a generic corporate brochure. 00:11:09.490 --> 00:11:11.669 Which is why the human review layer is completely 00:11:11.669 --> 00:11:14.039 non -negotiable. A quick weekly review keeps 00:11:14.039 --> 00:11:16.399 the system's quality stable. You have to maintain 00:11:16.399 --> 00:11:19.100 that critical human feedback loop. But if the 00:11:19.100 --> 00:11:22.460 ROI is this obvious and incredibly fast, why 00:11:22.460 --> 00:11:24.399 is anyone still hesitating to implement this 00:11:24.399 --> 00:11:27.600 today? Well, it's almost entirely a deep psychological 00:11:27.600 --> 00:11:30.740 barrier for management. Leaders are deeply afraid 00:11:30.740 --> 00:11:32.399 of breaking their current functional systems. 00:11:32.940 --> 00:11:35.820 Their legacy systems are incredibly slow, but 00:11:35.820 --> 00:11:38.139 they technically function. They fear the messy 00:11:38.139 --> 00:11:40.279 transition period more than anything. Leaders 00:11:40.279 --> 00:11:42.340 fear breaking their slow but functional legacy 00:11:42.340 --> 00:11:45.159 systems. Yeah. Change is terrifying, even when 00:11:45.159 --> 00:11:48.019 it's massively profitable. So let's synthesize 00:11:48.019 --> 00:11:50.700 this entire deep dive for our listeners. The 00:11:50.700 --> 00:11:53.820 businesses winning in 2026 share a very specific 00:11:53.820 --> 00:11:56.539 philosophy. They aren't waiting around for flawless 00:11:56.539 --> 00:11:59.179 humanoid androids. They don't expect a glowing 00:11:59.179 --> 00:12:02.100 AI to sit at a desk. They're actively breaking 00:12:02.100 --> 00:12:05.179 work down to the atomic task level. They map 00:12:05.179 --> 00:12:07.700 the tiny tasks before touching any software tools. 00:12:07.980 --> 00:12:10.500 They deploy highly affordable, good enough AI 00:12:10.500 --> 00:12:13.299 for repetitive steps. And they fiercely maintain 00:12:13.299 --> 00:12:17.000 a lean, highly strategic human layer. Those expert 00:12:17.000 --> 00:12:19.620 humans sit at the top to guide everything. Speed 00:12:19.620 --> 00:12:21.860 to market decisively beats the grand illusion 00:12:21.860 --> 00:12:24.419 of perfection. Your execution matters infinitely 00:12:24.419 --> 00:12:27.019 more than your specific tool choice. You have 00:12:27.019 --> 00:12:28.980 to clean your messy internal processes first. 00:12:29.200 --> 00:12:31.399 Then you let the automation scale your newfound 00:12:31.399 --> 00:12:34.399 efficiency. It's a profound shift in how we fundamentally 00:12:34.399 --> 00:12:36.379 structure work. Thank you for taking this deep 00:12:36.379 --> 00:12:38.039 dive with us today. Before we go, I want to leave 00:12:38.039 --> 00:12:39.940 you with this thought. We started by talking 00:12:39.940 --> 00:12:43.200 about waiting for perfect AI. But consider the 00:12:43.200 --> 00:12:45.519 reality of a totally leveled playing field. If 00:12:45.519 --> 00:12:47.820 every single company on earth eventually has 00:12:47.820 --> 00:12:50.840 access, access to the exact same three second, 00:12:50.919 --> 00:12:53.759 one cent AI to generate them creative content 00:12:53.759 --> 00:12:56.610 and politely answer their ringing phones. beat 00:12:56.610 --> 00:12:59.269 what becomes the ultimate irreplaceable premium 00:12:59.269 --> 00:13:01.990 commodity in modern business out tarot music