WEBVTT 00:00:00.000 --> 00:00:02.100 I was looking at the notes for today's deep dive 00:00:02.100 --> 00:00:06.459 this morning, and there was this one phrase in 00:00:06.459 --> 00:00:08.919 the research that just stopped me cold. Oh, the 00:00:08.919 --> 00:00:11.359 empty company concept. Yeah, the empty company. 00:00:11.460 --> 00:00:13.800 It's not about the neural networks or the processing 00:00:13.800 --> 00:00:16.559 speed. Right. It's about us. It's a really haunting 00:00:16.559 --> 00:00:19.019 image, isn't it? It really is. It's this idea 00:00:19.019 --> 00:00:23.899 that you can look at a spreadsheet in 2026 and 00:00:24.059 --> 00:00:26.839 Everything looks perfect. The costs are down. 00:00:26.940 --> 00:00:29.039 The efficiency is through the roof. Exactly. 00:00:29.199 --> 00:00:31.539 Because you've fired the humans and plugged in 00:00:31.539 --> 00:00:34.179 the AI agents, but then you walk into the building. 00:00:34.539 --> 00:00:36.579 Or you log into the Slack channel. Right. And 00:00:36.579 --> 00:00:39.119 there is nobody home. The soul has just been 00:00:39.119 --> 00:00:40.640 automated right out of the building. And the 00:00:40.640 --> 00:00:43.920 customers feel it. That is the kicker. Yeah. 00:00:44.060 --> 00:00:46.140 The data we're looking at today suggests this 00:00:46.140 --> 00:00:50.240 massive paradox. Business owners think they are 00:00:50.240 --> 00:00:52.920 optimizing for profit. But they're actually optimizing 00:00:52.920 --> 00:00:55.759 for irrelevance. Exactly. They are building these 00:00:55.759 --> 00:00:58.460 incredibly efficient engines that nobody wants 00:00:58.460 --> 00:01:01.539 to ride in. Optimizing for irrelevance. That 00:01:01.539 --> 00:01:05.319 is a heavy concept. It is. So that is our mission 00:01:05.319 --> 00:01:07.540 today. For everyone listening, we aren't talking 00:01:07.540 --> 00:01:09.879 about apocalypse scenarios. No, no hype cycles 00:01:09.879 --> 00:01:12.640 today. Right. We are doing a deep dive into the 00:01:12.640 --> 00:01:17.040 12 AI traps of 2026. Basically a guide on how 00:01:17.040 --> 00:01:19.799 to stop tripping over our own shoelaces. Yeah, 00:01:19.900 --> 00:01:21.719 while trying to sprint into the future. Exactly. 00:01:22.060 --> 00:01:26.420 We are going to dissect why automating the wrong 00:01:26.420 --> 00:01:30.000 things actually breaks a business. Why AI is 00:01:30.000 --> 00:01:32.859 terrible at strategic decisions, despite what 00:01:32.859 --> 00:01:34.879 the tech trolls say. Yeah. And then we're going 00:01:34.879 --> 00:01:37.219 to get very tactical with the toolbox. Because 00:01:37.219 --> 00:01:40.140 if you're using chat GPT for everything in 2026, 00:01:40.420 --> 00:01:42.900 you are doing it wrong. Totally wrong. Let's 00:01:42.900 --> 00:01:45.439 start with that empty company concept. OK. The 00:01:45.439 --> 00:01:47.219 source material leads with what they call the 00:01:47.219 --> 00:01:50.260 big cost saving myth. Right. Trap one. And I 00:01:50.260 --> 00:01:52.219 have to be honest here. As someone who runs a 00:01:52.219 --> 00:01:54.599 business, I feel this temptation. Oh, sure. There 00:01:54.599 --> 00:01:57.099 is a vulnerability in leadership where you see 00:01:57.099 --> 00:01:59.579 the line item for salaries. And then you see 00:01:59.579 --> 00:02:02.439 the monthly subscription for an AI agent. Right. 00:02:02.680 --> 00:02:05.739 And the math is just so seductive. The math is 00:02:05.739 --> 00:02:09.919 seductive, but it is incomplete. The source argues 00:02:09.919 --> 00:02:13.020 that on paper, the AI looks like the perfect 00:02:13.020 --> 00:02:16.719 employee. It doesn't sleep. It doesn't ask for 00:02:16.719 --> 00:02:19.000 a raise. It doesn't take lunch breaks. So the 00:02:19.000 --> 00:02:21.199 instinct is just to replace the humans. Right. 00:02:21.659 --> 00:02:24.120 But the source says this kills the energy of 00:02:24.120 --> 00:02:27.199 the company. Now, that sounds a bit vague to 00:02:27.199 --> 00:02:29.479 me. What does that actually mean in a profit 00:02:29.479 --> 00:02:32.159 and loss statement? Think about it this way. 00:02:33.199 --> 00:02:35.599 Business is fundamentally about the transfer 00:02:35.599 --> 00:02:38.639 of emotion. Okay. You buy from people you trust. 00:02:39.280 --> 00:02:42.259 You work harder for leaders you respect. Humans 00:02:42.259 --> 00:02:44.520 read the room. And a computer doesn't. Right. 00:02:44.620 --> 00:02:46.340 A computer doesn't know if a client is hesitant 00:02:46.340 --> 00:02:49.479 or just busy. A computer creates output, but 00:02:49.479 --> 00:02:51.819 a human creates connection. So if you strip that 00:02:51.819 --> 00:02:54.520 away to save cash... You lose the mechanism that 00:02:54.520 --> 00:02:56.520 actually generates the revenue. You're saving 00:02:56.520 --> 00:02:58.840 money on the expenses, but capping your upside 00:02:58.840 --> 00:03:01.060 on the income. Precisely. You are efficient, 00:03:01.180 --> 00:03:04.000 but you're dead. Wow. And this bleeds right into 00:03:04.000 --> 00:03:06.479 trap three, which is customer service. This is 00:03:06.479 --> 00:03:08.199 where I see the most damage happening right now. 00:03:08.300 --> 00:03:10.599 Oh, I completely agree. Companies see a chat 00:03:10.599 --> 00:03:12.689 bot and think... Great. I never have to deal 00:03:12.689 --> 00:03:14.870 with a complaint again. I've been on the receiving 00:03:14.870 --> 00:03:17.949 end of that. It is infuriating. It is. You just 00:03:17.949 --> 00:03:20.289 want to scream representative into your phone. 00:03:20.449 --> 00:03:22.750 It's worse than infuriating. It's just bad strategy. 00:03:23.090 --> 00:03:25.909 The source calls a complaint a gift. A gift. 00:03:26.169 --> 00:03:28.449 That is a very generous way to frame it. When 00:03:28.449 --> 00:03:31.370 a customer complains, they are telling you exactly 00:03:31.370 --> 00:03:34.090 how to fix your product. They're giving you free 00:03:34.090 --> 00:03:37.449 consulting. Right. If a bot deflects that with 00:03:37.449 --> 00:03:40.939 a generic apology, you lose the data. But more 00:03:40.939 --> 00:03:43.719 importantly, you miss the emotional opportunity. 00:03:44.039 --> 00:03:46.840 The torn shirt analogy from the notes. Yes. The 00:03:46.840 --> 00:03:49.139 torn shirt analogy nails this. Walk me through 00:03:49.139 --> 00:03:53.800 that. So imagine you buy a shirt online, it arrives, 00:03:54.259 --> 00:03:56.280 and it's torn. You're annoyed. Very annoyed. 00:03:56.479 --> 00:03:58.219 You go to the chat support. If a bot handled 00:03:58.219 --> 00:04:00.500 that, what does it do? It scans your text for 00:04:00.500 --> 00:04:03.699 the word damage. Exactly. And it pastes a link 00:04:03.699 --> 00:04:06.099 to the return policy. It follows the rules. It 00:04:06.099 --> 00:04:08.400 follows the rules, but it fails the human. It 00:04:08.400 --> 00:04:11.740 is cold. It feels bureaucratic. Now imagine a 00:04:11.740 --> 00:04:14.099 human handles it. They see the picture and say, 00:04:14.280 --> 00:04:16.800 oh no, I am so sorry that happened. They validate 00:04:16.800 --> 00:04:18.899 the frustration. Right. They say, that looks 00:04:18.899 --> 00:04:21.459 awful. Tell you what, I'm sending a replacement 00:04:21.459 --> 00:04:23.500 overnight and I'm throwing in a discount code. 00:04:23.620 --> 00:04:26.360 And suddenly I am not angry. No. I am actually 00:04:26.360 --> 00:04:28.420 a bigger fan than I was before the shirt tour. 00:04:28.699 --> 00:04:31.620 Exactly. You converted a failure into loyalty. 00:04:32.639 --> 00:04:35.540 A bot cannot do that because a bot doesn't feel 00:04:35.540 --> 00:04:37.740 empathy. That makes sense. The rule of thumb 00:04:37.740 --> 00:04:40.459 the source gives here is super useful. Use bots 00:04:40.459 --> 00:04:44.319 for facts. Like opening hours or tracking numbers. 00:04:44.360 --> 00:04:47.060 Right. And use humans for feelings. Facts versus 00:04:47.060 --> 00:04:49.360 feelings. That seems like the dividing line for 00:04:49.360 --> 00:04:52.399 job security in 2026. Absolutely. But there is 00:04:52.399 --> 00:04:54.639 a flip side to this. We talk about leaders wanting 00:04:54.639 --> 00:04:58.420 to replace people. But the source also mentions 00:04:58.420 --> 00:05:02.339 the fear coming from the employees. Trap 10 touches 00:05:02.339 --> 00:05:05.720 on this IT silo mentality. Oh, yeah. The secret 00:05:05.720 --> 00:05:08.379 weapon problem. Right. Often, companies treat 00:05:08.379 --> 00:05:11.579 AI as this high -tech thing that only the engineering 00:05:11.579 --> 00:05:14.060 team or the IT guys are allowed to touch. Which 00:05:14.060 --> 00:05:16.500 makes everyone else feel like they are outdated. 00:05:16.879 --> 00:05:19.379 It creates a class system. It does. If you work 00:05:19.379 --> 00:05:23.000 in sales or HR, and you see the engineers playing 00:05:23.000 --> 00:05:25.620 with these powerful tools. We start to think, 00:05:25.660 --> 00:05:27.819 wait, is that thing going to take my job? And 00:05:27.819 --> 00:05:30.839 that fear. robs the culture. People start hiding 00:05:30.839 --> 00:05:33.160 information. They stop collaborating. So how 00:05:33.160 --> 00:05:35.459 do you break that silo? You can't just force 00:05:35.459 --> 00:05:38.399 people to use technology they're afraid of. The 00:05:38.399 --> 00:05:41.040 source suggests something surprisingly low tech. 00:05:41.720 --> 00:05:44.959 The Friday pizza party. Please tell me it is 00:05:44.959 --> 00:05:47.300 more than just free food. It's about the environment. 00:05:47.439 --> 00:05:49.879 You don't do a seminar. You don't do a mandatory 00:05:49.879 --> 00:05:52.560 training module. You get the team together, order 00:05:52.560 --> 00:05:55.250 pizza, and say, hey, Let's look at this one simple 00:05:55.250 --> 00:05:57.889 trick together. Maybe it's taking a rude email 00:05:57.889 --> 00:06:00.350 and asking the AI to rewrite it to be polite. 00:06:00.709 --> 00:06:02.730 Everyone tries it. They laugh about the results. 00:06:02.829 --> 00:06:05.290 So they gamify it. Exactly. You take the scary 00:06:05.290 --> 00:06:07.310 monster and turn it into a toy. You take the 00:06:07.310 --> 00:06:09.709 sci -fi out of it and make it a utility. Yes. 00:06:10.129 --> 00:06:12.689 So if the goal isn't replacement, but safety, 00:06:12.970 --> 00:06:15.009 what is the core psychological shift a leader 00:06:15.009 --> 00:06:17.930 needs to make? Explicitly tell your team the 00:06:17.930 --> 00:06:20.449 tool helps them. it does not replace them. I 00:06:20.449 --> 00:06:22.649 like that. Let's move to the actual mechanics 00:06:22.649 --> 00:06:25.189 of work. Okay. Because trap two is automating 00:06:25.189 --> 00:06:27.689 everything. And I think this is where the efficiency 00:06:27.689 --> 00:06:30.129 mindset really backfires. This is definitely 00:06:30.129 --> 00:06:32.889 for the speed junkies. The source drops a hard 00:06:32.889 --> 00:06:35.730 truth here. Yeah. Speeding up a bad process just 00:06:35.730 --> 00:06:39.269 means you fail faster. Wow. Speeding up a bad 00:06:39.269 --> 00:06:42.529 process means you fail faster. That is profound. 00:06:42.649 --> 00:06:45.310 It is. We often equate velocity with progress. 00:06:45.410 --> 00:06:48.769 If I am moving fast, I must be winning. But... 00:06:48.360 --> 00:06:50.939 If you are running in the wrong direction, velocity 00:06:50.939 --> 00:06:53.519 is your enemy. Right. The example they give is 00:06:53.519 --> 00:06:56.620 email outreach. Let's say you write a boring, 00:06:56.839 --> 00:06:59.759 ineffective sales email. OK. In the old days, 00:06:59.779 --> 00:07:02.759 you would send 10 of them, get no replies, and 00:07:02.759 --> 00:07:04.639 realize you need to rewrite it. The feedback 00:07:04.639 --> 00:07:07.600 loop was slow, but it was safe. Exactly. Now 00:07:07.600 --> 00:07:11.300 you use an AI agent to scrape 1 ,000 leads, and 00:07:11.300 --> 00:07:13.560 you fire off that boring email to all of them 00:07:13.560 --> 00:07:15.740 in three minutes. You haven't scaled your success. 00:07:15.920 --> 00:07:18.529 No. You have just annoyed 1 ,000 people instantly. 00:07:18.910 --> 00:07:21.329 You have burned your market. So the advice is 00:07:21.329 --> 00:07:23.829 actually counterintuitive. It's do things that 00:07:23.829 --> 00:07:27.290 don't scale first. Manual first. Always. Yeah. 00:07:27.569 --> 00:07:30.470 Write 10 personal emails. See if anyone replies. 00:07:30.709 --> 00:07:33.230 And once you have the words that work, then you 00:07:33.230 --> 00:07:35.790 automate. Exactly. This reminds me of the bakery 00:07:35.790 --> 00:07:38.050 story from Trap 5. Oh, that's a great one. It 00:07:38.050 --> 00:07:39.910 really resonated with me because it highlights 00:07:39.910 --> 00:07:42.470 the difference between building tools and building 00:07:42.470 --> 00:07:44.610 a business. It's the archetype of the engineer 00:07:44.610 --> 00:07:47.259 founder. Right. So this guy wants to start a 00:07:47.259 --> 00:07:50.019 bakery. But instead of buying flour, he spends 00:07:50.019 --> 00:07:54.579 three months coding a complex AI system. A system 00:07:54.579 --> 00:07:57.560 to predict bread sales based on local weather 00:07:57.560 --> 00:08:00.060 patterns. And foot traffic data. Right, right. 00:08:00.420 --> 00:08:02.579 He is building a prediction engine for a product 00:08:02.579 --> 00:08:05.600 that doesn't exist. Zero bread. He hasn't baked 00:08:05.600 --> 00:08:08.220 a single loaf. And what happens? When he finally 00:08:08.220 --> 00:08:10.699 opens... It turns out people in that neighborhood 00:08:10.699 --> 00:08:12.600 didn't even want bread. They wanted chocolate 00:08:12.600 --> 00:08:15.220 chip cookies. Exactly. His system was totally 00:08:15.220 --> 00:08:17.839 useless. Start with flour and water. Fix the 00:08:17.839 --> 00:08:20.720 bread. Sell it on the street. Use technology 00:08:20.720 --> 00:08:23.920 to grow, not to learn the basics. We see this 00:08:23.920 --> 00:08:26.980 with what the source calls shiny object syndrome. 00:08:27.220 --> 00:08:30.339 Oh, we are all guilty of this. We see a cool 00:08:30.339 --> 00:08:33.480 video of a new tool and we invent a problem just 00:08:33.480 --> 00:08:36.169 so we can use it. Right. But the source suggests 00:08:36.169 --> 00:08:39.090 using a mental model called the theory of constraints. 00:08:39.470 --> 00:08:43.129 The bottleneck theory. Yes. Imagine your work 00:08:43.129 --> 00:08:46.309 is water flowing through a pipe. OK. There's 00:08:46.309 --> 00:08:48.970 always one spot where the pipe is narrowest. 00:08:49.570 --> 00:08:51.950 That is the block. If you are great at shooting 00:08:51.950 --> 00:08:54.610 video, but it takes you four hours to write a 00:08:54.610 --> 00:08:57.950 title. The title is the block. Right. So buying 00:08:57.950 --> 00:09:00.070 a faster camera doesn't help. It does nothing. 00:09:00.190 --> 00:09:02.169 It just piles up more video footage that you 00:09:02.169 --> 00:09:04.710 can't title. Exactly. You need a tool specifically 00:09:04.710 --> 00:09:07.450 for titles. So it seems like the trap is putting 00:09:07.450 --> 00:09:09.590 the tool before the problem. How do we discipline 00:09:09.590 --> 00:09:12.169 ourselves? Identify the bottleneck first, the 00:09:12.169 --> 00:09:14.350 one thing taking your time, then solve only that. 00:09:14.629 --> 00:09:16.809 We are going to take a quick break for our sponsors, 00:09:16.909 --> 00:09:19.049 but when we come back, we are going to get into 00:09:19.049 --> 00:09:21.690 the limits of machine intelligence. Sounds good. 00:09:21.929 --> 00:09:27.870 And we are back. Segment 3, traps 4, 7, and 8. 00:09:28.169 --> 00:09:30.889 This is the philosophical part. It is. It really 00:09:30.889 --> 00:09:33.629 challenges the way we view intelligence itself. 00:09:33.710 --> 00:09:35.309 You have to be honest about what the machine 00:09:35.309 --> 00:09:38.230 actually is. Right. Now it is amazing that these 00:09:38.230 --> 00:09:40.409 models can read a thousand pages in a minute. 00:09:40.850 --> 00:09:44.450 Whoa. Imagine scaling to a billion queries. It's 00:09:44.450 --> 00:09:46.350 incredible. I definitely don't want to downplay 00:09:46.350 --> 00:09:49.509 that. Yeah. But the source defines AI simply 00:09:49.509 --> 00:09:52.860 as a guessing machine. A guessing machine? It 00:09:52.860 --> 00:09:55.080 is just predicting the next word. It's basically 00:09:55.080 --> 00:09:57.600 autocorrect on steroids. On massive steroids, 00:09:57.820 --> 00:10:01.399 yes. But because its goal is to predict the most 00:10:01.399 --> 00:10:04.620 likely next word, it wants to please you. And 00:10:04.620 --> 00:10:08.159 this leads to trap four, hallucinations. Right. 00:10:08.259 --> 00:10:10.720 We hear this term a lot. I think people misunderstand 00:10:10.720 --> 00:10:13.100 why it happens, though. It happens because the 00:10:13.100 --> 00:10:16.580 model prioritizes completion over accuracy. Let's 00:10:16.580 --> 00:10:19.200 define it simply. Hallucination is when the computer 00:10:19.200 --> 00:10:21.970 dreams up fake facts to please you. Right. Like 00:10:21.970 --> 00:10:24.210 if you ask it, how many cars did we sell last 00:10:24.210 --> 00:10:26.110 month? And it doesn't have access to your database? 00:10:26.289 --> 00:10:29.009 It might just say 500. Because 500 sounds like 00:10:29.009 --> 00:10:31.269 a plausible number for a car dealership. It sounds 00:10:31.269 --> 00:10:34.269 like a good sentence. It's not lying in the human 00:10:34.269 --> 00:10:37.529 sense. It's filling in the blank. Which is terrifying 00:10:37.529 --> 00:10:40.029 if you are reporting to a bank. You have to verify 00:10:40.029 --> 00:10:42.149 everything. This leads to the bigger issue of 00:10:42.149 --> 00:10:46.110 decision making. Trap seven. I see leaders who 00:10:46.110 --> 00:10:49.090 want the AI to make the big strategic calls. 00:10:49.230 --> 00:10:51.980 Oh, yeah. Should we expand to Europe? Right. 00:10:52.240 --> 00:10:55.200 Should I hire this VP? And that is a fatal error. 00:10:55.799 --> 00:10:58.840 Because remember, the AI is trained on the past. 00:10:59.120 --> 00:11:01.740 All of its data is historical. When you ask it 00:11:01.740 --> 00:11:04.360 for a decision, it gives you the average of what 00:11:04.360 --> 00:11:06.639 has worked before. It creates a regression to 00:11:06.639 --> 00:11:08.899 the meme. Exactly. Regression to the meme. Let's 00:11:08.899 --> 00:11:11.539 unpack that. So it pushes you toward mediocrity. 00:11:11.580 --> 00:11:13.919 It pushes you toward the average. Think about 00:11:13.919 --> 00:11:16.429 it. If you ask it to write a pop song, it will 00:11:16.429 --> 00:11:18.210 write a song that sounds like every top song. 00:11:18.330 --> 00:11:21.509 Because it's trained on all of them. Right. But 00:11:21.509 --> 00:11:24.870 big business success, the outliers, usually come 00:11:24.870 --> 00:11:28.429 from weird, brave, counterintuitive ideas. And 00:11:28.429 --> 00:11:31.110 data cannot predict those. If you listen to the 00:11:31.110 --> 00:11:33.940 AI... You will build a business that looks exactly 00:11:33.940 --> 00:11:35.840 like everyone else's. You will build a perfectly 00:11:35.840 --> 00:11:38.240 average business. The source uses a beautiful 00:11:38.240 --> 00:11:41.080 analogy here about creativity. Yeah. A human 00:11:41.080 --> 00:11:45.080 can watch a bird fly and get an idea for an airplane 00:11:45.080 --> 00:11:47.759 part. We connect unrelated things. But an AI 00:11:47.759 --> 00:11:50.259 just looks at a database of old airplane parts. 00:11:50.679 --> 00:11:53.419 Exactly. It mixes old ideas. It doesn't create 00:11:53.419 --> 00:11:56.799 new futures. And that links to trap sex, chasing 00:11:56.799 --> 00:12:00.220 the news. The hype cycle. With how fast this 00:12:00.220 --> 00:12:03.360 stuff moves, there is a lot of anxiety. Oh, a 00:12:03.360 --> 00:12:05.580 new model dropped today. I need to change my 00:12:05.580 --> 00:12:08.059 whole strategy. It is exhausting. The source 00:12:08.059 --> 00:12:10.019 says to just ignore it. Focus on what doesn't 00:12:10.019 --> 00:12:12.600 change. Jeff Bezos famously said this. Customers 00:12:12.600 --> 00:12:15.159 will always want cheap prices and fast shipping. 00:12:15.500 --> 00:12:18.039 No new AI model changes that. Build your business 00:12:18.039 --> 00:12:20.429 around those constants. not the daily software 00:12:20.429 --> 00:12:23.210 updates. If the machine effectively regresses 00:12:23.210 --> 00:12:26.730 to the mean, what is the human's specific role 00:12:26.730 --> 00:12:29.830 in the final decision? Humans must use gut feeling 00:12:29.830 --> 00:12:32.250 and personal risk tolerance to make the final 00:12:32.250 --> 00:12:34.509 call. Let's get practical. Segment four is the 00:12:34.509 --> 00:12:38.250 tactical toolkit. Yes. Trap 12, the one tool 00:12:38.250 --> 00:12:41.090 mistake. We want one app to rule them all. The 00:12:41.090 --> 00:12:45.049 all in one. We like simplicity. But in 2026, 00:12:45.450 --> 00:12:47.470 specialization is the reality. You wouldn't use 00:12:47.470 --> 00:12:50.210 a hammer to cut wood. But people use ChatGPT 00:12:50.210 --> 00:12:52.549 for everything. They really do. So let's break 00:12:52.549 --> 00:12:55.250 down the 2026 Toolbox as the source describes 00:12:55.250 --> 00:12:58.309 it. We have three main players, Claude, Gemini, 00:12:58.450 --> 00:13:00.450 and ChatGPT. Right. Let's start with Claude. 00:13:00.549 --> 00:13:02.750 OK. Think of Claude as your thoughtful writer 00:13:02.750 --> 00:13:05.769 or your analyst. The source describes his voice 00:13:05.769 --> 00:13:09.649 as natural, softer. It has a massive context 00:13:09.649 --> 00:13:11.809 window too. Right. Meaning you can feed it huge 00:13:11.809 --> 00:13:14.529 documents. So if I need to summarize a 50 page 00:13:14.529 --> 00:13:17.169 contract. Or write a sensitive email to a client. 00:13:17.360 --> 00:13:20.080 You use Claude. Next is Gemini? Gemini is your 00:13:20.080 --> 00:13:22.379 connection to the live internet. It's Google. 00:13:22.759 --> 00:13:25.679 So this is for facts. Checking the weather, looking 00:13:25.679 --> 00:13:28.299 up stock prices, verifying news events. Because 00:13:28.299 --> 00:13:31.080 if you ask ChatGPT for today's news, it might 00:13:31.080 --> 00:13:33.720 hallucinate. Exactly. Gemini is grounded in the 00:13:33.720 --> 00:13:36.879 live web. And then the classic, ChatGPT. What 00:13:36.879 --> 00:13:42.179 is its lane now? Logic, reasoning, coding, structure. 00:13:42.200 --> 00:13:44.639 OK. If you need to break down a complex problem 00:13:44.639 --> 00:13:48.200 into steps or write a Python script, ChatGPT 00:13:48.200 --> 00:13:50.860 is the sharpest tool in the shed. So it's a woodworker's 00:13:50.860 --> 00:13:53.220 bench. I like that. Claude is the sander, getting 00:13:53.220 --> 00:13:56.539 things smooth. Gemini is the tape measure, getting 00:13:56.539 --> 00:14:00.360 facts straight. And ChatGPT is the saw, cutting 00:14:00.360 --> 00:14:03.220 through the logic. Perfect analogy. If you try 00:14:03.220 --> 00:14:06.080 to use the logic bot to write a feeling email, 00:14:06.159 --> 00:14:08.799 it sounds robotic. And if you use the creative 00:14:08.799 --> 00:14:11.340 bot to check facts, it lies to you. You have 00:14:11.340 --> 00:14:13.539 to pick the right tool. Trap9 warns us about 00:14:13.539 --> 00:14:16.200 how we talk to these tools too. The more data 00:14:16.200 --> 00:14:19.129 myth. We tend to think if we paste a whole book 00:14:19.129 --> 00:14:21.830 into the chat, the answer will be smarter. Right. 00:14:21.830 --> 00:14:25.149 Here is everything I know. Fix my life. It's 00:14:25.149 --> 00:14:27.169 like asking for directions and getting 50 maps 00:14:27.169 --> 00:14:29.129 at once. You just get confused. And the computer 00:14:29.129 --> 00:14:32.389 gets confused too. The advice is a clean desk 00:14:32.389 --> 00:14:35.929 policy. Key prompt short. One page of notes equals 00:14:35.929 --> 00:14:39.279 one summary. If you want a clear output, you 00:14:39.279 --> 00:14:41.700 must provide a clear input. So specialization 00:14:41.700 --> 00:14:44.700 is key. What is the immediate action step regarding 00:14:44.700 --> 00:14:46.600 these different tools? Make a folder linking 00:14:46.600 --> 00:14:48.620 all three and pick the right one for the job. 00:14:48.940 --> 00:14:51.620 As we wrap up this deep dive, let's synthesize 00:14:51.620 --> 00:14:54.919 this. We have covered 12 traps today. From the 00:14:54.919 --> 00:14:57.960 empty company to the one tool mistake. But the 00:14:57.960 --> 00:15:00.559 overarching theme here seems to be about confidence. 00:15:01.159 --> 00:15:03.440 Confidence in our own humanity. That is the big 00:15:03.440 --> 00:15:06.480 idea. The narrative has been man versus machine. 00:15:06.720 --> 00:15:08.940 But the research shows it's actually man plus 00:15:08.940 --> 00:15:12.340 machine. The real winners in 2026 aren't the 00:15:12.340 --> 00:15:14.279 ones who automated everything into oblivion. 00:15:14.399 --> 00:15:17.200 They are the ones who mixed human skills like 00:15:17.200 --> 00:15:21.299 empathy and bravery with machine speed. Technology 00:15:21.299 --> 00:15:24.259 should be used to grow, not to learn. You have 00:15:24.259 --> 00:15:26.779 to learn the business manually first. So for 00:15:26.779 --> 00:15:29.059 everyone listening, what is the move for tomorrow 00:15:29.059 --> 00:15:31.000 morning? We don't want to overwhelm them. Pick 00:15:31.000 --> 00:15:34.720 one thing. Don't try to fix all 12 traps at once. 00:15:34.940 --> 00:15:37.620 Maybe tomorrow you stop using the bot for customer 00:15:37.620 --> 00:15:40.220 complaints. Or maybe you just organize your toolbar 00:15:40.220 --> 00:15:42.200 with the three different AIs. Just start with 00:15:42.200 --> 00:15:44.480 one thing. I love that. The source ends with 00:15:44.480 --> 00:15:46.220 a provocative thought that I want to leave you 00:15:46.220 --> 00:15:48.460 with. Yeah. It mentions that a machine can never 00:15:48.460 --> 00:15:51.779 understand the soul of an idea. It leads me to 00:15:51.779 --> 00:15:55.039 ask you, the listener, which parts of your work 00:15:55.039 --> 00:15:57.600 require a soul? And are you accidentally trying 00:15:57.600 --> 00:16:00.179 to automate them? That is the question. Something 00:16:00.179 --> 00:16:03.259 to think about. Thank you for joining us on this 00:16:03.259 --> 00:16:04.620 deep dive. Take care, everyone.