WEBVTT 00:00:00.000 --> 00:00:03.439 I was sitting in traffic yesterday just staring 00:00:03.439 --> 00:00:05.480 at the car in front of me, and I started thinking 00:00:05.480 --> 00:00:10.000 about exhaustion, like real exhaustion. And it's 00:00:10.000 --> 00:00:12.820 rarely the big stuff, is it? It's not the marathon 00:00:12.820 --> 00:00:15.300 you run or the big project you finally ship. 00:00:15.550 --> 00:00:19.170 It's the friction. It's the hundreds of tiny, 00:00:19.170 --> 00:00:22.030 invisible decisions you have to make before you 00:00:22.030 --> 00:00:24.190 can actually do anything meaningful. It's the 00:00:24.190 --> 00:00:26.890 death by a thousand cuts. It's looking for the 00:00:26.890 --> 00:00:28.929 document, then realizing you need the password 00:00:28.929 --> 00:00:31.609 for the document, then formatting the email to 00:00:31.609 --> 00:00:33.770 send the document. Exactly. And the source material 00:00:33.770 --> 00:00:36.310 we're looking at today, frictionless efficiency. 00:00:38.230 --> 00:00:40.509 proposes something that honestly, it just stopped 00:00:40.509 --> 00:00:42.990 me in my tracks. It argues that we are living 00:00:42.990 --> 00:00:45.310 through this massive technological shift with 00:00:45.310 --> 00:00:48.590 AI, yet most of us are using these tools completely 00:00:48.590 --> 00:00:50.670 wrong. We're using the Ferrari to go to the mailbox. 00:00:50.909 --> 00:00:52.789 Yeah, that's the analogy. We're treating them 00:00:52.789 --> 00:00:55.310 like search engines. Right. If you go to ChatGPT 00:00:55.310 --> 00:00:57.950 and ask, what is the capital of France? You're 00:00:57.950 --> 00:01:00.590 using a supercomputer as an encyclopedia. You're 00:01:00.590 --> 00:01:03.390 using it for novelty. And the shift we're exploring 00:01:03.390 --> 00:01:07.049 today is moving from that novelty to utility. 00:01:07.469 --> 00:01:10.450 How do we stop using AI as an oracle and start 00:01:10.450 --> 00:01:12.950 using it as a workflow engine? A workflow engine. 00:01:13.129 --> 00:01:15.590 That's the key. That is a very distinct concept. 00:01:16.069 --> 00:01:18.269 So welcome to the deep dive. We are going to 00:01:18.269 --> 00:01:21.150 unpack some really practical AI workflow hacks, 00:01:21.450 --> 00:01:22.969 but we're going to filter them down to the ones 00:01:22.969 --> 00:01:24.930 that actually matter. We're looking at professional 00:01:24.930 --> 00:01:27.340 friction. domestic friction, which is a huge 00:01:27.340 --> 00:01:29.700 one for me, and the logistics of travel. And 00:01:29.700 --> 00:01:32.319 just to set expectations, the mission here isn't 00:01:32.319 --> 00:01:34.980 to add more apps to your phone. It's not about 00:01:34.980 --> 00:01:37.659 becoming a power user just for the sake of it. 00:01:37.939 --> 00:01:40.299 No, it's about subtraction. It's about using 00:01:40.299 --> 00:01:42.799 this tech to strip away all that administrative 00:01:42.799 --> 00:01:44.840 drag so you can actually get your brain back. 00:01:45.040 --> 00:01:46.819 OK, let's start with the professional side, because 00:01:46.819 --> 00:01:49.060 I think this is where that search engine mindset 00:01:49.060 --> 00:01:52.659 is the most entrenched. For sure. We use it to 00:01:52.659 --> 00:01:54.980 write an email, maybe summarize a report. But 00:01:54.980 --> 00:01:57.260 the source points out a much deeper bottleneck. 00:01:57.540 --> 00:02:00.680 Standard operating procedures. Documentation. 00:02:01.739 --> 00:02:04.180 The unsexy stuff that actually runs the world. 00:02:04.420 --> 00:02:06.420 Nobody wakes up excited to write a training manual. 00:02:06.939 --> 00:02:09.400 No. And that's a classic scaling problem, right? 00:02:09.699 --> 00:02:12.120 You know how to do a complex task. Maybe it's 00:02:12.120 --> 00:02:14.800 pulling a specific data report. But explaining 00:02:14.800 --> 00:02:16.500 it to someone else takes three times as long 00:02:16.500 --> 00:02:18.949 as just doing it yourself. So you never explain 00:02:18.949 --> 00:02:20.949 it. You just keep doing it, and you become the 00:02:20.949 --> 00:02:24.610 bottleneck. Exactly. But the hack here bridges 00:02:24.610 --> 00:02:28.250 that gap between doing and documenting using 00:02:28.250 --> 00:02:32.789 multimodal AI. Okay, how? The strategy is to 00:02:32.789 --> 00:02:35.210 open a voice recorder while you are actually 00:02:35.210 --> 00:02:37.729 doing the task. You just narrate your actions 00:02:37.729 --> 00:02:40.669 in real time. So literally talking to yourself 00:02:40.669 --> 00:02:42.469 while you work. Okay, now I'm clicking the settings 00:02:42.469 --> 00:02:44.789 gear, now I'm scrolling down. It sounds a little 00:02:44.789 --> 00:02:47.250 crazy in the moment, sure. It's stream of consciousness, 00:02:47.449 --> 00:02:49.930 it's gonna be messy. You might, you know, stumble 00:02:49.930 --> 00:02:52.490 or correct yourself. Right. But then, and this 00:02:52.490 --> 00:02:55.280 is the workflow part. You upload that raw audio 00:02:55.280 --> 00:02:58.439 file to the AI with a very specific prompt. You 00:02:58.439 --> 00:03:00.199 don't just say, summarize this. What do you say? 00:03:00.319 --> 00:03:02.560 You say, convert this transcript into a step 00:03:02.560 --> 00:03:05.439 -by -step standard operating procedure. Use bold 00:03:05.439 --> 00:03:08.159 headers for actions. Assume the reader is a beginner. 00:03:08.520 --> 00:03:10.740 I see. So you're offloading the structuring part 00:03:10.740 --> 00:03:13.120 to the model. You provide the raw knowledge. 00:03:13.539 --> 00:03:15.819 It provides the architecture. Precisely. You 00:03:15.819 --> 00:03:18.080 convert the physical action of doing the work 00:03:18.080 --> 00:03:20.900 into documentation, like instantly. Right. The 00:03:20.900 --> 00:03:22.939 friction of formatting and typing is just... 00:03:22.889 --> 00:03:25.930 gone. That seems like a really high ROI move. 00:03:26.110 --> 00:03:28.469 But what about friction with other people? Meetings, 00:03:28.469 --> 00:03:30.569 for instance. I think we're all guilty of the 00:03:30.569 --> 00:03:32.810 five -minute panic trap. Oh, yeah. You're about 00:03:32.810 --> 00:03:34.729 to jump on a call. You realize you have no context, 00:03:34.770 --> 00:03:37.490 so you frantically Google the person. And you 00:03:37.490 --> 00:03:39.789 end up with their job title and maybe their college. 00:03:40.110 --> 00:03:42.650 It's surface level. It's just noise. Exactly. 00:03:42.990 --> 00:03:45.629 So the workflow hack is to use AI to synthesize 00:03:45.629 --> 00:03:48.250 context, not just find facts. You take their 00:03:48.250 --> 00:03:50.849 LinkedIn URL, paste it into the model, and ask 00:03:50.849 --> 00:03:53.439 for Pattern recognition. Pattern recognition. 00:03:53.560 --> 00:03:55.680 OK, what does that look like in a prompt? You 00:03:55.680 --> 00:03:58.240 ask something like, based on this profile, what 00:03:58.240 --> 00:04:00.800 are three non -obvious conversation starters? 00:04:01.719 --> 00:04:04.879 What themes appear in their work history? Maybe 00:04:04.879 --> 00:04:07.819 they move from finance to nonprofit work. That's 00:04:07.819 --> 00:04:10.719 a story. The AI finds that story. So you start 00:04:10.719 --> 00:04:13.180 the meeting by saying, I noticed you made a fascinating 00:04:13.180 --> 00:04:16.939 pivot in 2018 instead of, so what do you do? 00:04:17.279 --> 00:04:20.180 Yes. It immediately changes the temperature of 00:04:20.180 --> 00:04:22.339 the room. It signals you've done your homework. 00:04:23.139 --> 00:04:25.560 Even if it only took you 30 seconds, it lets 00:04:25.560 --> 00:04:27.939 you skip the small talk. Speaking of high -stakes 00:04:27.939 --> 00:04:31.680 communication, what about job interviews or asking 00:04:31.680 --> 00:04:35.639 for a raise? The source mentions using simulation. 00:04:36.339 --> 00:04:38.399 It sounds a little like Ankin class, but without 00:04:38.399 --> 00:04:40.800 the embarrassment. Oh, right. The friction here 00:04:40.800 --> 00:04:43.660 is emotional. It's fear of the unknown. Yeah. 00:04:43.879 --> 00:04:45.980 We put off difficult conversations because we 00:04:45.980 --> 00:04:47.779 can't predict how the other person's going to 00:04:47.779 --> 00:04:50.339 react. So how do you simulate that? You use voice 00:04:50.339 --> 00:04:54.420 mode on a model like Gemini or Chad's GPT, and 00:04:54.420 --> 00:04:57.209 you set the stage. You tell the AI. Act as a 00:04:57.209 --> 00:04:59.529 skeptical CFO. I'm going to pitch you on a budget 00:04:59.529 --> 00:05:01.310 increase. I want you to interrupt me. I want 00:05:01.310 --> 00:05:04.370 you to be difficult. That sounds genuinely stressful. 00:05:04.509 --> 00:05:06.329 It should be. It's a flight simulator. You want 00:05:06.329 --> 00:05:08.189 to crash in the simulator so you don't crash 00:05:08.189 --> 00:05:10.930 the actual plane. If the AI asks you a tough 00:05:10.930 --> 00:05:13.149 question and you freeze, that's useful data. 00:05:13.449 --> 00:05:16.529 You can pause, regroup, try again. By the time 00:05:16.529 --> 00:05:18.769 you're in the real meeting, your brain has already 00:05:18.769 --> 00:05:21.959 lived it. the anxiety drops. It's using AI to 00:05:21.959 --> 00:05:24.379 build muscle memory, not just text. Exactly. 00:05:24.740 --> 00:05:27.420 And for research, the source highlights Notebook 00:05:27.420 --> 00:05:30.079 LM because of its grounding feature. So how does 00:05:30.079 --> 00:05:33.379 that work for, say, a job interview? Well, you 00:05:33.379 --> 00:05:35.560 upload the company's annual report, their last 00:05:35.560 --> 00:05:38.259 10 blog posts, and the job description into one 00:05:38.259 --> 00:05:41.500 notebook. Then you ask. Generate the 10 most 00:05:41.500 --> 00:05:44.319 likely interview questions based specifically 00:05:44.319 --> 00:05:47.139 on these texts. So it's not giving you generic 00:05:47.139 --> 00:05:49.180 questions like, what is your biggest weakness? 00:05:49.420 --> 00:05:52.199 No, not at all. It'll ask you, how would you 00:05:52.199 --> 00:05:54.980 apply your experience to our new Q3 strategy 00:05:54.980 --> 00:05:57.519 for the Asian market? It uses their own internal 00:05:57.519 --> 00:05:59.720 language. You walk in sounding like an insider. 00:05:59.860 --> 00:06:02.519 OK, let's pause on that. If AI can handle the 00:06:02.519 --> 00:06:04.920 preparation, the documentation, the research, 00:06:05.000 --> 00:06:07.319 and even simulate the conversation. I see where 00:06:07.319 --> 00:06:09.079 you're going with this. If AI handles all the 00:06:09.079 --> 00:06:11.360 prep, does that mean our only real job left is 00:06:11.360 --> 00:06:14.339 the actual decision making? Yes. It strips away 00:06:14.339 --> 00:06:16.899 the admin work so you can focus purely on judgment 00:06:16.899 --> 00:06:19.860 and execution. Okay, let's pivot. Because work 00:06:19.860 --> 00:06:22.600 friction is one thing, but home friction is. 00:06:23.279 --> 00:06:25.480 It's a different kind of exhausting. It's the 00:06:25.480 --> 00:06:27.639 invisible labor. The stuff that keeps the house 00:06:27.639 --> 00:06:29.860 running but nobody ever applauds you for. Right. 00:06:29.939 --> 00:06:32.279 And the visual capabilities of these tools seem 00:06:32.279 --> 00:06:34.879 to be the unlock here. The appliance repair example 00:06:34.879 --> 00:06:37.360 really struck me. You mean because you usually 00:06:37.360 --> 00:06:39.300 just ignore the weird noise until the appliance 00:06:39.300 --> 00:06:41.360 dies? Guilty. It's the most expensive way to 00:06:41.360 --> 00:06:44.980 deal with it. Well... The hack is multimodal 00:06:44.980 --> 00:06:47.800 analysis. You don't hunt for the manual you threw 00:06:47.800 --> 00:06:50.060 away. You just take a picture of the control 00:06:50.060 --> 00:06:52.660 panel, that error code or blinking light, and 00:06:52.660 --> 00:06:55.759 you ask the AI. Identify this model, explain 00:06:55.759 --> 00:06:58.240 this error code, and give me the step -by -step 00:06:58.240 --> 00:07:00.759 reset instructions. It's like having the technician's 00:07:00.759 --> 00:07:03.000 cheat sheet. It connects the visual data to the 00:07:03.000 --> 00:07:05.399 technical manual. It works for stains, too. Red 00:07:05.399 --> 00:07:08.019 wine on a carpet? Don't guess. Take a photo, 00:07:08.420 --> 00:07:10.699 identify the fabric, ask for the right solution. 00:07:10.860 --> 00:07:13.120 But the one that really hit home for me, and 00:07:13.120 --> 00:07:15.000 I'm going to be honest here, this is a daily 00:07:15.000 --> 00:07:17.699 struggle, is the what's for dinner problem. Oh, 00:07:17.699 --> 00:07:20.180 yeah. It's six pure zero p .m. I'm staring at 00:07:20.180 --> 00:07:23.319 a full fridge, but my brain just it stops. I 00:07:23.319 --> 00:07:25.939 can't compute a meal. That is decision fatigue 00:07:25.939 --> 00:07:28.720 in its purest form. Your executive function is 00:07:28.720 --> 00:07:31.339 just done for the day. So we order takeout. Right. 00:07:31.480 --> 00:07:34.519 The fix is to outsource the creativity. You open 00:07:34.519 --> 00:07:36.660 the fridge, you snap a photo of everything on 00:07:36.660 --> 00:07:38.899 the shelves and you upload it. The prompt is 00:07:38.899 --> 00:07:41.350 key. I have these ingredients. I have 20 minutes. 00:07:41.649 --> 00:07:45.329 Give me three simple, healthy meal options. Why 00:07:45.329 --> 00:07:48.029 three? Why not just ask for the best one? Because 00:07:48.029 --> 00:07:51.529 one option feels like an order, but 10 is overwhelming. 00:07:52.050 --> 00:07:54.069 Three gives you a sense of agency without the 00:07:54.069 --> 00:07:57.790 fatigue. Stir -fry, omelet, or pasta. The friction 00:07:57.790 --> 00:08:00.529 of infinite possibility is gone. You just pick 00:08:00.529 --> 00:08:03.660 B. I love that. It creates a menu based on your 00:08:03.660 --> 00:08:06.100 actual reality. What about the step before that 00:08:06.100 --> 00:08:08.540 though? The grocery shopping. The master grocery 00:08:08.540 --> 00:08:11.620 list hack. This solves what I call tab overload. 00:08:11.920 --> 00:08:13.459 You know when you have five different recipe 00:08:13.459 --> 00:08:15.040 tabs open in your browser? And you're scrolling 00:08:15.040 --> 00:08:17.319 back and forth trying to write it all down. And 00:08:17.319 --> 00:08:18.899 you end up with three bags of carrots because 00:08:18.899 --> 00:08:21.259 you didn't check the overlap. So the workflow 00:08:21.259 --> 00:08:25.079 is copy the URLs of all the recipes, paste them 00:08:25.079 --> 00:08:29.079 into the AI. Okay. Then you prompt, create a 00:08:29.079 --> 00:08:32.190 single master grocery list. deduplicate all the 00:08:32.190 --> 00:08:34.289 ingredients, so combine the flour amounts, and 00:08:34.289 --> 00:08:36.509 here's the important part. Organize the list 00:08:36.509 --> 00:08:39.070 by the aisle of the grocery store. Organize by 00:08:39.070 --> 00:08:41.309 aisle? Oh, that's the real friction remover right 00:08:41.309 --> 00:08:43.129 there. It turns a cognitive task planning and 00:08:43.129 --> 00:08:45.669 sorting into a pure execution task. You walk 00:08:45.669 --> 00:08:48.169 in, follow the list top to bottom, and you walk 00:08:48.169 --> 00:08:51.250 out. No backtracking. There's a theme here. It 00:08:51.250 --> 00:08:53.129 seems like we're always trying to separate the 00:08:53.129 --> 00:08:55.850 planning from the doing. That is the core insight. 00:08:56.230 --> 00:08:58.590 Friction happens when we try to plan and execute 00:08:58.590 --> 00:09:01.220 at the same time. AI is a great planner. It sets 00:09:01.220 --> 00:09:02.740 up the pins so you can just knock them down. 00:09:02.940 --> 00:09:06.539 But I have to ask, if we use AI to fix our appliances 00:09:06.539 --> 00:09:10.279 and plan our meals, are we de -skilling ourselves? 00:09:10.480 --> 00:09:12.379 Are we just becoming helpless without the bot? 00:09:12.779 --> 00:09:15.159 We aren't losing skills. We are reclaiming the 00:09:15.159 --> 00:09:17.919 mental energy usually wasted on logistics like 00:09:17.919 --> 00:09:20.740 sorting grocery lists. We are back. We've covered 00:09:20.740 --> 00:09:22.960 work and home. Now let's talk about moving through 00:09:22.960 --> 00:09:26.330 the world. Logistics. Travel. Yeah. Travel is 00:09:26.330 --> 00:09:28.730 supposed to be fun, but the planning part is 00:09:28.730 --> 00:09:31.690 often just high -stress logistics. The hack that 00:09:31.690 --> 00:09:34.309 caught my eye was the en route stop. This solves 00:09:34.309 --> 00:09:36.769 a problem that Google Maps is actually not very 00:09:36.769 --> 00:09:39.590 good at. Right. Google Maps is great at go from 00:09:39.590 --> 00:09:43.009 A to B. It is terrible at nuanced queries. Yeah. 00:09:43.169 --> 00:09:45.289 Let's say you're driving from Seattle to Portland. 00:09:46.009 --> 00:09:48.750 You want coffee, but you need easy parking. and 00:09:48.750 --> 00:09:50.649 you don't want to drive 10 minutes off the highway. 00:09:51.009 --> 00:09:53.169 If you just search coffee and maps, it shows 00:09:53.169 --> 00:09:55.610 you everything, including the tiny stand downtown 00:09:55.610 --> 00:09:58.129 you can't even get to. Exactly. So you use an 00:09:58.129 --> 00:10:01.490 LLM with web access. You say, I am driving south 00:10:01.490 --> 00:10:04.549 on I -5 from Seattle. Find me a highly rated 00:10:04.549 --> 00:10:06.590 coffee shop that is less than three minutes from 00:10:06.590 --> 00:10:09.230 an exit and has a large parking lot. It filters 00:10:09.230 --> 00:10:11.440 for the negative constraints. Not far from the 00:10:11.440 --> 00:10:13.600 highway, not small parking. And it gives you 00:10:13.600 --> 00:10:15.860 one or two perfect options. You've just saved 00:10:15.860 --> 00:10:19.200 yourself so much frustration. That idea of constraints 00:10:19.200 --> 00:10:22.700 applies to itinerary planning, too. I'm so guilty 00:10:22.700 --> 00:10:25.679 of the impossible itinerary, trying to see five 00:10:25.679 --> 00:10:28.419 museums in one morning. We all are. We're optimists 00:10:28.419 --> 00:10:32.519 when we plan. The AI is a realist, so the hack 00:10:32.519 --> 00:10:35.429 is constraint -based planning. How does that 00:10:35.429 --> 00:10:38.289 work? You act as the client and the AI is the 00:10:38.289 --> 00:10:40.409 travel agent. You tell it, I'm going to Paris 00:10:40.409 --> 00:10:43.029 for three days. I love art, but I hate crowds. 00:10:43.389 --> 00:10:46.029 I have a budget of X. Create an itinerary that 00:10:46.029 --> 00:10:48.669 allows for two hours of downtime every afternoon. 00:10:48.840 --> 00:10:51.220 You force it to prioritize downtime. You force 00:10:51.220 --> 00:10:53.500 reality onto the plan. It'll tell you, look, 00:10:53.580 --> 00:10:55.360 you can't do the Louvre and Versailles in the 00:10:55.360 --> 00:10:57.820 same day if you want downtime. It saves you from 00:10:57.820 --> 00:11:00.159 your own ambition. And what about packing? I 00:11:00.159 --> 00:11:02.860 am a chronic overpacker. Context aware lists. 00:11:03.159 --> 00:11:05.179 Don't just ask for a packing list for Chicago. 00:11:05.460 --> 00:11:07.740 Ask, I'm going to Chicago in November for a tech 00:11:07.740 --> 00:11:09.720 conference. I need professional clothes, but 00:11:09.720 --> 00:11:11.980 I'm carry on only. And I plan to run outside 00:11:11.980 --> 00:11:14.120 in the mornings. So it balances the suit, the 00:11:14.120 --> 00:11:16.419 winter coat, and the running shoes. It acts as 00:11:16.419 --> 00:11:18.799 a logic check. He remembers the adapter, the 00:11:18.799 --> 00:11:20.840 power bank, all the things you forget when you're 00:11:20.840 --> 00:11:23.580 rushing. And there's one more travel hack that 00:11:23.580 --> 00:11:25.820 circles back to that simulation idea we talked 00:11:25.820 --> 00:11:28.700 about. Language. This is huge for confidence. 00:11:28.919 --> 00:11:30.980 If you're going to Italy, you might know chow. 00:11:31.580 --> 00:11:35.299 But ordering a full meal is terrifying. You just 00:11:35.299 --> 00:11:38.580 don't want to look stupid. So role play it. Act 00:11:38.580 --> 00:11:41.779 as a waiter in a Roman trattoria. I want to ask 00:11:41.779 --> 00:11:43.879 for a table for two. Correct my pronunciation. 00:11:44.080 --> 00:11:46.340 You get to stumble in private. And when you get 00:11:46.340 --> 00:11:49.179 there, You've already said the words. The neural 00:11:49.179 --> 00:11:51.240 pathway is already there. It's incredible how 00:11:51.240 --> 00:11:53.519 these tools can lower the barrier to entry for 00:11:53.519 --> 00:11:56.279 real -life experiences, but I'm curious. Does 00:11:56.279 --> 00:11:59.539 the perfect efficiency of an AI itinerary kill 00:11:59.539 --> 00:12:02.360 the spontaneity of travel? No, because by handling 00:12:02.360 --> 00:12:06.179 the logistics hotel's routes, you actually free 00:12:06.179 --> 00:12:09.000 up mind space to be present and spontaneous in 00:12:09.000 --> 00:12:11.059 the moment. That makes a lot of sense. You aren't 00:12:11.059 --> 00:12:12.639 staring at your phone so you can actually look 00:12:12.639 --> 00:12:15.220 at the scenery. Okay, this last segment is where 00:12:15.220 --> 00:12:17.860 things get really interesting. We're moving from 00:12:17.860 --> 00:12:21.159 being consumers of these tools to being creators. 00:12:21.460 --> 00:12:23.200 This sounds a bit intimidating. Building tools 00:12:23.200 --> 00:12:25.159 sounds like I need to learn how to code. And 00:12:25.159 --> 00:12:27.899 that's the misconception. The source highlights 00:12:27.899 --> 00:12:31.240 this shift where just plain English becomes the 00:12:31.240 --> 00:12:33.419 coding language. You mean with these mini -apps. 00:12:33.600 --> 00:12:36.860 Exactly. What even is a mini -app in this context? 00:12:36.980 --> 00:12:39.580 Think of those tiny annoying problems that no 00:12:39.580 --> 00:12:41.679 software company is ever going to solve for you. 00:12:41.879 --> 00:12:44.399 Like maybe you and your roommates split rent 00:12:44.399 --> 00:12:46.820 based on the square footage of your bedrooms. 00:12:47.080 --> 00:12:49.700 That's a very specific math problem. Right. There's 00:12:49.700 --> 00:12:52.320 no app for that. But you can go to a tool like 00:12:52.320 --> 00:12:55.279 Claude or ChatGPT and just say, build me a simple 00:12:55.279 --> 00:12:58.139 calculator where I input the total rent and the 00:12:58.139 --> 00:13:00.700 square footage of three rooms, and it calculates 00:13:00.700 --> 00:13:02.960 the split. And it just builds it. It writes the 00:13:02.960 --> 00:13:06.980 code and renders a working clickable tool right 00:13:06.980 --> 00:13:09.240 there in the chat window. Whoa. You can use it 00:13:09.240 --> 00:13:11.259 on the spot. You've basically just wished a piece 00:13:11.259 --> 00:13:13.799 of software into existence. That is wild. You 00:13:13.799 --> 00:13:15.679 become a software engineer for 30 seconds to 00:13:15.679 --> 00:13:18.120 solve a one -time problem. It's disposable software. 00:13:18.320 --> 00:13:20.220 You stop waiting for an app to exist and you 00:13:20.220 --> 00:13:22.299 just build it yourself. Another great one from 00:13:22.299 --> 00:13:26.240 the source is the screenshot decoder. The desktop 00:13:26.240 --> 00:13:28.919 graveyard. I have hundreds of screenshots. I'll 00:13:28.919 --> 00:13:30.500 need this setting later. And you never look at 00:13:30.500 --> 00:13:32.039 them because you forget what they even mean. 00:13:32.960 --> 00:13:36.409 So the hack. You batch upload those screenshots 00:13:36.409 --> 00:13:40.009 to the AI, and you prompt, analyze these screenshots, 00:13:40.809 --> 00:13:42.909 explain what technical process is happening here, 00:13:43.309 --> 00:13:46.049 and convert it all into a text -based checklist. 00:13:46.269 --> 00:13:48.909 It decodes the visual memory into a text workflow. 00:13:49.190 --> 00:13:52.009 It turns a pile of random images into a user 00:13:52.009 --> 00:13:54.870 manual. It's incredibly powerful for learning 00:13:54.870 --> 00:13:57.919 new software. So looking at all of this, From 00:13:57.919 --> 00:14:01.399 voice noting SOPs to building our own rent calculators, 00:14:01.799 --> 00:14:03.419 it feels like we're not just consuming these 00:14:03.419 --> 00:14:05.399 tools anymore. We're engaging as architects. 00:14:05.659 --> 00:14:07.340 We're building the scaffolding to make our own 00:14:07.340 --> 00:14:10.159 lives easier. So this shifts the user from asking 00:14:10.159 --> 00:14:13.080 for help to building a solution. What does that 00:14:13.080 --> 00:14:16.240 do to our relationship with software? It democratizes 00:14:16.240 --> 00:14:18.480 software. We stop waiting for an app to exist 00:14:18.480 --> 00:14:21.460 and just build a mini tool to solve our specific 00:14:21.460 --> 00:14:23.590 friction point. Okay, let's try to unpack this 00:14:23.590 --> 00:14:26.049 whole thing. We've covered a lot of ground, from 00:14:26.049 --> 00:14:28.730 voice -noting SOPs to photographing our pantries. 00:14:29.009 --> 00:14:31.570 It is a lot. And that brings us to what I think 00:14:31.570 --> 00:14:33.789 is the most critical insight from the source 00:14:33.789 --> 00:14:36.929 material, the rule of one. Yes, because the temptation 00:14:36.929 --> 00:14:39.590 right now is to turn this off and try to do all 00:14:39.590 --> 00:14:42.090 21 hacks at once. Which would create massive 00:14:42.090 --> 00:14:44.070 friction. The irony would be terrible. The source 00:14:44.070 --> 00:14:46.190 is very clear about this. The instruction is 00:14:46.190 --> 00:14:49.370 to pick one hack, just one, and apply it this 00:14:49.370 --> 00:14:51.720 week. It's about building the muscle. Right. 00:14:52.080 --> 00:14:54.919 AI isn't for one -off answers. It's for repeated 00:14:54.919 --> 00:14:58.340 thinking. It only becomes a real time saver when 00:14:58.340 --> 00:15:00.740 it replaces a loop you do every single day. So 00:15:00.740 --> 00:15:02.539 if you struggle with dinner, do the pantry hack. 00:15:02.899 --> 00:15:05.360 If you hate meeting prep, do the LinkedIn hack. 00:15:05.519 --> 00:15:07.919 Just pick one. The whole shift is moving from 00:15:07.919 --> 00:15:12.539 AI as a novelty, a party trick, to AI as a cognitive 00:15:12.539 --> 00:15:15.580 assistant, a partner, something that takes the 00:15:15.580 --> 00:15:18.019 load off so you can be human. I love that. So 00:15:18.019 --> 00:15:20.519 here is the challenge to you, the listener. Identify 00:15:20.519 --> 00:15:22.820 that one source of friction in your week. Is 00:15:22.820 --> 00:15:25.379 it the grocery list? The email drafting? The 00:15:25.379 --> 00:15:27.980 travel planning? Pick that one thing. Apply the 00:15:27.980 --> 00:15:30.200 hack. See if you get that time back. Because 00:15:30.200 --> 00:15:32.559 if you aren't saving hours yet, it's probably 00:15:32.559 --> 00:15:35.779 not because the AI isn't smart enough. It's because 00:15:35.779 --> 00:15:38.580 you're using it as a search engine, not a workflow 00:15:38.580 --> 00:15:40.519 partner. Couldn't have said it better myself. 00:15:41.139 --> 00:15:42.659 Thanks for diving in with us. We'll see you next 00:15:42.659 --> 00:15:43.480 time. Take care.