WEBVTT 00:00:00.000 --> 00:00:03.580 Most people think AI only threatens certain jobs, 00:00:04.160 --> 00:00:07.660 jobs like tech or customer support. But the real 00:00:07.660 --> 00:00:10.380 dividing line isn't your job title. It is whether 00:00:10.380 --> 00:00:13.679 you choose to learn AI or ignore it until it's 00:00:13.679 --> 00:00:15.919 too late. Yeah, that is the crucial threshold 00:00:15.919 --> 00:00:18.019 right there. The people getting caught off guard 00:00:18.019 --> 00:00:20.320 are the ones waiting for permission to adapt. 00:00:20.780 --> 00:00:22.899 While the ground is... Well, it is shifting right 00:00:22.899 --> 00:00:25.679 beneath them. Welcome to today's deep dive. We 00:00:25.679 --> 00:00:28.379 are unpacking five essential practical skills 00:00:28.379 --> 00:00:31.239 today, skills to help you thrive in the AI economy 00:00:31.239 --> 00:00:33.920 without needing a new degree. We have a really 00:00:33.920 --> 00:00:36.600 packed roadmap for you. We are going to explore 00:00:36.600 --> 00:00:39.240 how to become the AI person on your team. We 00:00:39.240 --> 00:00:41.619 will look at why judgment always beats speed, 00:00:42.200 --> 00:00:44.920 plus mastering context over prompts, the power 00:00:44.920 --> 00:00:47.820 of fast iteration, and finally, how to bulletproof 00:00:47.820 --> 00:00:50.119 your career with a second income stream. OK, 00:00:50.119 --> 00:00:52.060 let's unpack this. Let's start with this idea 00:00:52.060 --> 00:00:54.859 of becoming the AI person. It sounds intimidating, 00:00:55.140 --> 00:00:56.960 like you need to be a software engineer, but 00:00:56.960 --> 00:00:59.380 the landscape is shifting locally. A report that 00:00:59.380 --> 00:01:01.619 used to take two hours now takes 30 minutes. 00:01:01.979 --> 00:01:04.040 The change happens right at your desk. And you 00:01:04.040 --> 00:01:06.950 really don't need a technical background. Becoming 00:01:06.950 --> 00:01:09.950 the AI person just means knowing slightly more 00:01:09.950 --> 00:01:12.230 than the person sitting next to you. It's about 00:01:12.230 --> 00:01:15.750 finding one repeating task and making it noticeably 00:01:15.750 --> 00:01:18.109 faster. Which brings us to the Exxon Enterprise 00:01:18.109 --> 00:01:20.469 case study. They built hardware and software 00:01:20.469 --> 00:01:23.129 for law enforcement. Yeah. And they didn't roll 00:01:23.129 --> 00:01:26.090 out some massive, intimidating AI initiative. 00:01:26.189 --> 00:01:28.829 Not at all. They built a tool called DraftOne 00:01:28.829 --> 00:01:31.750 using the Azure OpenAI service. The mechanics 00:01:31.750 --> 00:01:34.950 were actually... Quite focused, they took a massive 00:01:34.950 --> 00:01:37.129 bottleneck, police officers typing up incident 00:01:37.129 --> 00:01:40.250 reports. Right, which takes hours. Exactly. And 00:01:40.250 --> 00:01:43.769 they aimed the AI squarely at it. Draft 1 basically 00:01:43.769 --> 00:01:46.109 drafts the report using the audio transcript 00:01:46.109 --> 00:01:48.390 from the officer's body camera. And the result 00:01:48.390 --> 00:01:51.370 was an 82 % reduction in police officer report 00:01:51.370 --> 00:01:54.069 writing time. Huge. Which completely flips the 00:01:54.069 --> 00:01:56.670 ROI of an officer. You aren't paying them to 00:01:56.670 --> 00:01:59.209 type. You are paying them to patrol. Exactly. 00:01:59.329 --> 00:02:03.209 It shifts their cognitive load from manual data 00:02:03.209 --> 00:02:06.349 entry to simple editing and verification. But 00:02:06.349 --> 00:02:09.150 the critical lesson here is how it started. This 00:02:09.150 --> 00:02:11.870 wasn't a top -down mandate where executives forced 00:02:11.979 --> 00:02:14.139 everyone to learn prompt engineering. Someone 00:02:14.139 --> 00:02:17.240 tested a highly specific repeatable use case. 00:02:17.580 --> 00:02:20.860 They proved it worked on one specific task, and 00:02:20.860 --> 00:02:23.379 then the adoption just spread naturally. It makes 00:02:23.379 --> 00:02:25.479 me think about finding a new gear on a bicycle. 00:02:25.759 --> 00:02:28.060 Yeah. You don't need to understand the physics 00:02:28.060 --> 00:02:30.159 of the derailleur to realize you are suddenly 00:02:30.159 --> 00:02:32.960 pedaling easier and moving faster. You just show 00:02:32.960 --> 00:02:35.500 your coworkers how to shift into that gear. Right, 00:02:35.539 --> 00:02:37.240 and suddenly everyone wants to ride that way. 00:02:37.560 --> 00:02:41.180 It's the exact same dynamic. Pick one tool, stick 00:02:41.180 --> 00:02:43.400 with it. Don't scatter your attention across 00:02:43.400 --> 00:02:46.580 50 different applications. Apply it to one task 00:02:46.580 --> 00:02:49.340 you repeat every single day and track the delta 00:02:49.340 --> 00:02:52.259 in your time. Does being the AI person mean you 00:02:52.259 --> 00:02:54.259 suddenly have to fix everyone's tech issues? 00:02:54.460 --> 00:02:56.919 No. It just means finding one repeating task 00:02:56.919 --> 00:02:59.520 and making it noticeably faster. You become the 00:02:59.520 --> 00:03:01.639 proof of concept, not the IT department. That 00:03:01.639 --> 00:03:04.610 is a crucial distinction. But... You know, finding 00:03:04.610 --> 00:03:07.169 shortcuts is one thing. Trusting them blindly 00:03:07.169 --> 00:03:09.310 is where companies are getting into serious trouble. 00:03:09.870 --> 00:03:12.189 Which leads us to the second skill, building 00:03:12.189 --> 00:03:14.990 judgment before speed. Judgment is where the 00:03:14.990 --> 00:03:18.750 real premium lives now. Speed is completely commoditized. 00:03:19.090 --> 00:03:21.590 AI can generate a thousand words in three seconds. 00:03:22.289 --> 00:03:24.150 But judgment knowing if those thousand words 00:03:24.150 --> 00:03:27.120 are actually correct. is rare. I still wrestle 00:03:27.120 --> 00:03:29.500 with prompt drift myself. Sometimes I just want 00:03:29.500 --> 00:03:31.659 to accept the first draft because it looks so 00:03:31.659 --> 00:03:33.819 polished. It is a very human temptation. Oh, 00:03:33.939 --> 00:03:36.759 we all feel it. I do it too. But that polish 00:03:36.759 --> 00:03:39.860 is precisely the danger. The AI's output almost 00:03:39.860 --> 00:03:41.860 always looks perfectly fine on the first try. 00:03:42.099 --> 00:03:44.759 It utilizes impeccable grammar, excellent structure, 00:03:45.120 --> 00:03:47.379 and a highly authoritative tone. Which brings 00:03:47.379 --> 00:03:50.340 us to the Deloitte, Australia disaster. I looked 00:03:50.340 --> 00:03:52.259 at this case study. What fascinated me wasn't 00:03:52.259 --> 00:03:54.960 the technology failing. It was the human psychology 00:03:54.960 --> 00:03:57.960 of the failure. It is a textbook example of trusting 00:03:57.960 --> 00:04:02.680 the Polish. Deloitte prepared a 237 -page report 00:04:02.680 --> 00:04:04.500 for the Australian Department of Employment. 00:04:04.919 --> 00:04:09.639 And recently, they had to refund $440 ,000 to 00:04:09.639 --> 00:04:12.740 the government. Wow. Because the report was entirely 00:04:12.740 --> 00:04:15.659 compromised by hallucinations. Completely compromised. 00:04:15.879 --> 00:04:18.779 The AI generated a flawless looking document. 00:04:19.100 --> 00:04:21.399 But when a researcher at the University of Sydney 00:04:21.399 --> 00:04:25.259 actually audited the citations, it just unraveled. 00:04:25.720 --> 00:04:28.040 The report featured fabricated quotes from a 00:04:28.040 --> 00:04:30.639 federal court judgment. It referenced academic 00:04:30.639 --> 00:04:33.300 research papers that literally did not exist. 00:04:33.660 --> 00:04:36.120 The AI just invented them out of thin air to 00:04:36.120 --> 00:04:39.800 satisfy the prompt. Why do incredibly smart people 00:04:39.800 --> 00:04:42.519 at top Stop consulting firms. Miss these obvious 00:04:42.519 --> 00:04:44.959 AI hallucinations. Because the initial output 00:04:44.959 --> 00:04:47.139 usually looks perfectly formatted and highly 00:04:47.139 --> 00:04:49.519 confident. It doesn't look like a mistake. Exactly. 00:04:49.600 --> 00:04:52.019 It looks like a finished product. Which means 00:04:52.019 --> 00:04:55.360 the workflow has to physically change. You cannot 00:04:55.360 --> 00:04:57.959 just read the output. You have to actively verify 00:04:57.959 --> 00:05:00.680 it against a known standard. And more importantly, 00:05:00.819 --> 00:05:02.939 you can't just fix the errors manually and move 00:05:02.939 --> 00:05:05.600 on. That is a trap so many people fall into. 00:05:05.699 --> 00:05:08.439 They get a bad draft from the AI, they rewrite 00:05:08.439 --> 00:05:10.899 it themselves, and they try again the next day. 00:05:11.540 --> 00:05:14.000 The machine learns nothing. You have to teach 00:05:14.000 --> 00:05:17.000 it your standard. If it uses overly formal language, 00:05:17.199 --> 00:05:19.199 you don't just delete the words. You reply and 00:05:19.199 --> 00:05:22.000 say, make this shorter, remove the academic tone. 00:05:22.339 --> 00:05:24.399 Use conversational language. Tell it what you 00:05:24.399 --> 00:05:27.139 changed and why. You are essentially calibrating 00:05:27.139 --> 00:05:30.240 the machine's judgment to match your own. But 00:05:30.240 --> 00:05:33.300 even with good feedback loops, the best way to 00:05:33.300 --> 00:05:35.800 prevent the AI from making those confident mistakes 00:05:35.800 --> 00:05:39.339 in the first place is to put guardrails up. You 00:05:39.339 --> 00:05:41.620 do this before it even starts generating words. 00:05:42.600 --> 00:05:46.240 Which brings us to our third skill, context engineering. 00:05:46.519 --> 00:05:48.399 This is fundamentally different from what most 00:05:48.399 --> 00:05:50.100 people are doing right now. We should define 00:05:50.100 --> 00:05:52.560 that clearly. Context engineering simply means 00:05:52.560 --> 00:05:54.920 giving AI your background info before asking 00:05:54.920 --> 00:05:58.000 a question. Exactly. Prompt engineering is focusing 00:05:58.000 --> 00:06:00.620 on how you ask the question. Context engineering 00:06:00.620 --> 00:06:02.779 is controlling what the AI already knows before 00:06:02.779 --> 00:06:05.720 you even open your mouth. A generic prompt with 00:06:05.720 --> 00:06:08.899 no context will always produce a generic hallucination 00:06:08.899 --> 00:06:12.399 prone result. I think of it like briefing. a 00:06:12.399 --> 00:06:15.120 highly intelligent, but totally on easy act intern. 00:06:15.699 --> 00:06:17.860 You can't just walk up to this intern and say, 00:06:18.300 --> 00:06:20.139 write a strategy report. Right. You have to hand 00:06:20.139 --> 00:06:22.699 them the company playbook, the style guide. And 00:06:22.699 --> 00:06:25.019 three examples of last year's successful reports 00:06:25.019 --> 00:06:27.759 before they even touch a keyboard. That is the 00:06:27.759 --> 00:06:30.079 perfect analogy. And when you look at how enterprise 00:06:30.079 --> 00:06:32.579 companies are doing this, the results are staggering. 00:06:33.079 --> 00:06:35.980 Look at Wells Fargo. They ruled out an AI tool 00:06:35.980 --> 00:06:40.120 for 35 ,000 bankers across 4 ,000 branches. But 00:06:40.120 --> 00:06:42.420 they didn't just give their employees a blank 00:06:42.420 --> 00:06:45.430 chat window. Right. Giving a bank teller a blank 00:06:45.430 --> 00:06:48.389 chat GPT screen sounds like a compliance nightmare. 00:06:48.750 --> 00:06:51.110 Oh, it would be catastrophic. So instead, they 00:06:51.110 --> 00:06:54.029 built a Microsoft Teams app and loaded it with 00:06:54.029 --> 00:06:57.329 1700 internal procedures. They essentially built 00:06:57.329 --> 00:07:00.620 a walled garden of context. The AI can only pull 00:07:00.620 --> 00:07:03.560 answers from those specific approved internal 00:07:03.560 --> 00:07:05.779 documents. And the operational shift there is 00:07:05.779 --> 00:07:08.519 massive. The source notes that response times 00:07:08.519 --> 00:07:11.519 plummeted from 10 minutes down to just 30 seconds. 00:07:11.600 --> 00:07:14.079 Because the banker doesn't have to search a massive 00:07:14.079 --> 00:07:16.079 intranet. They don't have to open three different 00:07:16.079 --> 00:07:18.920 PDFs and synthesize an answer while the customer 00:07:18.920 --> 00:07:23.279 is waiting. The AI does the synthesis, but it 00:07:23.279 --> 00:07:25.939 is constrained by the context. Constraining the 00:07:25.939 --> 00:07:28.569 AI is the secret. It removes the guesswork. You 00:07:28.569 --> 00:07:30.649 don't need a massive corporate system to do this 00:07:30.649 --> 00:07:33.569 yourself. You just need to build a context package 00:07:33.569 --> 00:07:36.129 for your own work. What does a personal context 00:07:36.129 --> 00:07:38.389 package look like in practice? It's basically 00:07:38.389 --> 00:07:41.149 a simple text document where you store your rules, 00:07:41.569 --> 00:07:44.569 your brand voice, information about your target 00:07:44.569 --> 00:07:48.029 audience, examples of your best past work, past 00:07:48.029 --> 00:07:50.550 mistakes or cliches you want the AI to avoid. 00:07:51.050 --> 00:07:53.490 You feed that document into the AI first and 00:07:53.490 --> 00:07:56.079 then you ask it to do the work. So is context 00:07:56.079 --> 00:07:57.939 engineering really just writing an incredibly 00:07:57.939 --> 00:08:00.600 long, detailed prompt? It is more like giving 00:08:00.600 --> 00:08:03.060 the AI a custom instruction manual before assigning 00:08:03.060 --> 00:08:05.300 work. Okay, we are going to pause for a quick 00:08:05.300 --> 00:08:07.379 word from our sponsor. We will be right back. 00:08:07.560 --> 00:08:09.819 And we are back! So once you've loaded up that 00:08:09.819 --> 00:08:11.939 context and you understand the value of judgment, 00:08:12.439 --> 00:08:14.579 the temptation is to spend six months planning 00:08:14.579 --> 00:08:17.639 the perfect, flawless deployment. But the sources 00:08:17.639 --> 00:08:20.879 point in the exact opposite direction. You have 00:08:20.879 --> 00:08:25.660 to move to skill four. Mastering fast, iteration. 00:08:26.199 --> 00:08:28.560 Speed of testing will always beat speed of planning 00:08:28.560 --> 00:08:31.220 when it comes to AI. You cannot predict how these 00:08:31.220 --> 00:08:33.379 models will behave in the wild until you actually 00:08:33.379 --> 00:08:35.179 release them. It's like teaching someone to ride 00:08:35.179 --> 00:08:38.220 a bike. You do not sit them down in a classroom 00:08:38.220 --> 00:08:40.700 and explain the physics of gyroscopic balance. 00:08:40.899 --> 00:08:43.399 Yeah, right. You don't draw diagrams, you put 00:08:43.399 --> 00:08:45.159 them on the seat, you let them pedal, and you 00:08:45.159 --> 00:08:47.539 catch them when they wobble. You fix the small 00:08:47.539 --> 00:08:50.419 mistakes in real time. And that requires a major 00:08:50.419 --> 00:08:52.940 shift in corporate culture. You have to be willing 00:08:52.940 --> 00:08:56.080 to ship a rough version. Let's look at Klarna. 00:08:56.419 --> 00:08:58.820 They are the gold standard for fast iteration 00:08:58.820 --> 00:09:01.960 right now. In February 2024, they launched an 00:09:01.960 --> 00:09:04.519 open AI powered assistant to handle customer 00:09:04.519 --> 00:09:06.820 service. The scale of this rollout is what caught 00:09:06.820 --> 00:09:09.340 my attention. In a single month, it handled 2 00:09:09.340 --> 00:09:12.419 .3 million conversations. That was two thirds 00:09:12.419 --> 00:09:14.519 of all their customer service interactions globally. 00:09:14.759 --> 00:09:17.179 The efficiency gains are hard to comprehend. 00:09:17.779 --> 00:09:20.580 Their average resolution time dropped from 11 00:09:20.580 --> 00:09:23.139 minutes down to under two minutes. Oh, imagine 00:09:23.139 --> 00:09:25.720 scaling to a billion queries seamlessly like 00:09:25.720 --> 00:09:28.019 that. It completely changes the economics of 00:09:28.019 --> 00:09:30.639 the company. And repeat inquiries on the exact 00:09:30.639 --> 00:09:34.259 same issue fell by 25%, meaning the AI wasn't 00:09:34.259 --> 00:09:37.240 just faster. It was actually solving the root 00:09:37.240 --> 00:09:40.000 problems better than the manual process. But 00:09:40.000 --> 00:09:42.240 to your earlier point about chipping a rough 00:09:42.240 --> 00:09:44.940 version, They didn't just build this perfect 00:09:44.940 --> 00:09:47.440 system in a lab and then turn it on. How did 00:09:47.440 --> 00:09:49.600 they actually manage that deployment without 00:09:49.600 --> 00:09:52.000 alienating their customers? They built an aggressive 00:09:52.000 --> 00:09:54.879 audit loop. They didn't wait for angry customer 00:09:54.879 --> 00:09:58.580 reviews. They had human teams sampling hundreds 00:09:58.580 --> 00:10:00.960 of chats every single week. They were constantly 00:10:00.960 --> 00:10:03.299 hunting for friction points. When they found 00:10:03.299 --> 00:10:06.440 a content gap, say, the AI didn't know how to 00:10:06.440 --> 00:10:09.620 handle a specific type of return, they updated 00:10:09.620 --> 00:10:12.179 the context immediately. They were fixing the 00:10:12.179 --> 00:10:14.639 wobbles while the bike was moving. Doesn't shipping 00:10:14.639 --> 00:10:17.500 a rough version risk frustrating your actual 00:10:17.500 --> 00:10:20.019 customers? Not if you monitor it closely, catch 00:10:20.019 --> 00:10:22.860 errors early, and fix them immediately. The friction 00:10:22.860 --> 00:10:25.960 of a minor AI error is usually lower than the 00:10:25.960 --> 00:10:28.700 friction of waiting on hold for 45 minutes. That 00:10:28.700 --> 00:10:31.620 brings us to our final skill. This is where all 00:10:31.620 --> 00:10:34.299 the previous pieces lock together. When you combine 00:10:34.299 --> 00:10:37.340 fast iteration, deep context engineering, and 00:10:37.340 --> 00:10:39.840 human judgment, you suddenly possess the leverage 00:10:39.840 --> 00:10:42.470 of an entire team. You could do the work of five 00:10:42.470 --> 00:10:45.350 people. You become a highly leveraged individual. 00:10:45.590 --> 00:10:48.690 Which raises a structural question. Why restrict 00:10:48.690 --> 00:10:51.970 that immense power to just one fragile day job? 00:10:52.509 --> 00:10:55.110 Relying entirely on one paycheck is a vulnerable 00:10:55.110 --> 00:10:58.370 position. If that company downsizes, your leverage 00:10:58.370 --> 00:11:01.029 vanishes. That is skill number five. Building 00:11:01.029 --> 00:11:03.549 a second income stream using the domain expertise 00:11:03.549 --> 00:11:05.830 you already have. And we have a phenomenal case 00:11:05.830 --> 00:11:08.120 study for this with Rick Chorney. Rick's story 00:11:08.120 --> 00:11:10.220 is a perfect illustration of how to apply AI 00:11:10.220 --> 00:11:12.899 to physical, traditional businesses. He wasn't 00:11:12.899 --> 00:11:15.039 building a software startup. He was running a 00:11:15.039 --> 00:11:17.320 commercial cleaning company called Echo Janitorial 00:11:17.320 --> 00:11:19.639 Services in Vancouver. And he was doing it the 00:11:19.639 --> 00:11:22.200 hard way. He spent his first year working seven 00:11:22.200 --> 00:11:24.960 days a week with absolutely no days off. He was 00:11:24.960 --> 00:11:27.100 out in the field managing cleaning crews by 7 00:11:27.100 --> 00:11:29.539 a .m., getting home at 8 t .m., and then opening 00:11:29.539 --> 00:11:32.039 his laptop to do administrative work until 1 00:11:32.039 --> 00:11:35.340 a .m. Calculating quotes, replying to customer 00:11:35.340 --> 00:11:38.440 emails, scheduling, the invisible work that actually 00:11:38.440 --> 00:11:41.399 runs the business. He was earning roughly $14 00:11:41.399 --> 00:11:44.159 an hour when you calculated the sheer volume 00:11:44.159 --> 00:11:47.100 of his time. Burnout was inevitable at that pace, 00:11:47.299 --> 00:11:49.639 but instead of just grinding harder, he spent 00:11:49.639 --> 00:11:52.360 about four hours researching how AI could carry 00:11:52.360 --> 00:11:54.679 that administrative load. He wanted to buy his 00:11:54.679 --> 00:11:57.179 time back. Let's talk about the actual mechanics 00:11:57.179 --> 00:11:59.720 of what Rick built. The source material highlights 00:11:59.720 --> 00:12:03.600 three specific tools. Claude, and A .N. and Obsidian. 00:12:05.460 --> 00:12:07.500 But listing tools doesn't explain the workflow. 00:12:07.940 --> 00:12:10.080 How did he actually connect these things? This 00:12:10.080 --> 00:12:12.759 is where it gets brilliant. He essentially built 00:12:12.759 --> 00:12:14.980 an automated nervous system for his business. 00:12:15.539 --> 00:12:17.600 Imagine a potential customer fills out a contact 00:12:17.600 --> 00:12:20.000 form on his website at 10 p .m. asking for a 00:12:20.000 --> 00:12:22.379 quote. In the old days, Rick reads that at 1 00:12:22.379 --> 00:12:25.080 a .m. and manually types a reply. Total bottleneck. 00:12:25.120 --> 00:12:29.379 Right. So now he uses N8n. N8n is basically a 00:12:29.379 --> 00:12:31.840 digital bridge. It connects different applications 00:12:31.840 --> 00:12:34.399 together based on triggers. It sees that a new 00:12:34.399 --> 00:12:36.539 form was submitted. It acts as a dispatcher. 00:12:36.759 --> 00:12:39.659 Exactly. N8n grabs the customer's request and 00:12:39.659 --> 00:12:42.039 hands it over to Claude, which is the AI language 00:12:42.039 --> 00:12:44.820 model. But Claude needs to know Rick's pricing 00:12:44.820 --> 00:12:47.820 structure to write a quote. That is where Obsidian 00:12:47.820 --> 00:12:50.700 comes into play. Obsidian is a specialized note 00:12:50.700 --> 00:12:53.639 -taking app. It stores files as plain text on 00:12:53.639 --> 00:12:56.840 your hard drive. Yes. And Rick uses it as his 00:12:56.840 --> 00:12:59.039 context library. It holds his pricing tiers, 00:12:59.220 --> 00:13:02.259 his scheduling rules, his brand voice. So Claude 00:13:02.259 --> 00:13:04.700 reads the customer's request, pulls the correct 00:13:04.700 --> 00:13:08.000 context from Obsidian, and drafts a highly personalized, 00:13:08.299 --> 00:13:11.480 accurate quote. Then ends the draft back to N8N, 00:13:11.500 --> 00:13:13.899 which automatically sends the email. So a customer 00:13:13.899 --> 00:13:16.600 asks for a quote at 10 p .m. and within two minutes, 00:13:16.779 --> 00:13:18.820 they have a fully customized proposal sitting 00:13:18.820 --> 00:13:20.799 in their inbox. Yeah. And Rick was sleeping. 00:13:20.970 --> 00:13:23.289 He completely removed himself from the administrative 00:13:23.289 --> 00:13:25.889 bottleneck. He set up an AI receptionist to handle 00:13:25.889 --> 00:13:28.409 phone inquiries. He automated his intake forms. 00:13:28.690 --> 00:13:30.789 And the entire setup process for that automated 00:13:30.789 --> 00:13:33.129 workflow took him about half a day. Half a day 00:13:33.129 --> 00:13:36.629 of building to reclaim hours of time every single 00:13:36.629 --> 00:13:39.490 day. And the financial result is undeniable. 00:13:40.099 --> 00:13:42.360 Fortune magazine actually reviewed his business 00:13:42.360 --> 00:13:44.960 records. EcoGenitorial is projected to clear 00:13:44.960 --> 00:13:48.659 $1 .3 million in sales this year. Because he 00:13:48.659 --> 00:13:51.799 finally had the time to focus on growth, rather 00:13:51.799 --> 00:13:54.980 than drowning in paperwork. He used tech to scale 00:13:54.980 --> 00:13:57.340 a physical cleaning company. He didn't abandon 00:13:57.340 --> 00:13:59.500 his industry to become a prompt engineer. Do 00:13:59.500 --> 00:14:02.220 you have to invent a totally new tech -focused 00:14:02.220 --> 00:14:05.059 business to pull this off? No. You just apply 00:14:05.059 --> 00:14:07.259 AI to automate the administrative parts of your 00:14:07.259 --> 00:14:09.039 current skills. Stay close to the domain you 00:14:09.039 --> 00:14:11.480 already understand. Exactly. We do need to offer 00:14:11.480 --> 00:14:13.679 a practical warning here. If you are building 00:14:13.679 --> 00:14:16.460 a second income stream, check your current employment 00:14:16.460 --> 00:14:19.460 contract. Look out for non -compete clauses or 00:14:19.460 --> 00:14:21.860 policies about using company equipment. And most 00:14:21.860 --> 00:14:24.899 importantly, do not quit your main job prematurely. 00:14:25.320 --> 00:14:27.340 You wait until that side income is completely 00:14:27.340 --> 00:14:29.840 stable and proven. It's about creating a safety 00:14:29.840 --> 00:14:33.000 net, not jumping without a parachute. Let's recap 00:14:33.000 --> 00:14:35.519 the big ideas we've explored today. These are 00:14:35.519 --> 00:14:37.960 the levers you can pull right now. It really 00:14:37.960 --> 00:14:41.340 synthesizes down to five executable steps. First, 00:14:41.620 --> 00:14:44.139 step up and be the AI person by finding just 00:14:44.139 --> 00:14:47.419 one repeatable shortcut for your team. Second, 00:14:47.799 --> 00:14:50.399 build a strict feedback loop to enforce human 00:14:50.399 --> 00:14:54.419 judgment over AI speed. Third, engineer your 00:14:54.419 --> 00:14:56.759 context before you engineer your prompts. Give 00:14:56.759 --> 00:15:00.299 the machine the playbook. Fourth, iterate fast. 00:15:00.639 --> 00:15:03.440 Ship a rough version and fix the wobbles in real 00:15:03.440 --> 00:15:06.379 time. And finally, take all that newly created 00:15:06.379 --> 00:15:08.679 leverage and build a secondary income stream 00:15:08.679 --> 00:15:11.179 that isn't tied to a single employer's payroll. 00:15:11.379 --> 00:15:13.059 You don't need to change your career today. Just 00:15:13.059 --> 00:15:15.720 pick one of those skills. Build one context package. 00:15:16.039 --> 00:15:19.139 Automate one repeating email. That single action 00:15:19.139 --> 00:15:20.960 puts you leagues ahead of the people waiting 00:15:20.960 --> 00:15:23.179 to see what happens. You have the roadmap. It's 00:15:23.179 --> 00:15:25.389 time to start pedaling the bike. Sure. But before 00:15:25.389 --> 00:15:26.889 we wrap up, I want to leave you with a thought. 00:15:27.230 --> 00:15:28.970 A slightly deeper question about where all this 00:15:28.970 --> 00:15:30.750 technological leverage is actually heading. I 00:15:30.750 --> 00:15:33.330 love these. If these AI models are making everyone 00:15:33.330 --> 00:15:36.080 10 times faster. faster at producing reports, 00:15:36.659 --> 00:15:39.220 faster at writing code, faster at analyzing massive 00:15:39.220 --> 00:15:41.980 data sets. What happens to the fundamental value 00:15:41.980 --> 00:15:44.720 of speed itself? In a world where being fast 00:15:44.720 --> 00:15:47.879 is essentially free, maybe deep, slow, critical 00:15:47.879 --> 00:15:50.679 thinking becomes the ultimate premium skill. 00:15:50.799 --> 00:15:52.679 Two sec silence. That changes the whole game. 00:15:53.019 --> 00:15:54.679 Thank you for joining us on this deep dive. Take 00:15:54.679 --> 00:15:56.080 care of yourself. We will see you next time.