WEBVTT 00:00:00.000 --> 00:00:02.960 OK, let's talk about that frustration. When you've 00:00:02.960 --> 00:00:04.839 spent maybe 40 minutes, could be an hour, just 00:00:04.839 --> 00:00:07.099 trying to get the right words in a prompt, you're 00:00:07.099 --> 00:00:10.400 tweaking, you're fixing, and the output, it's 00:00:10.400 --> 00:00:12.619 still not quite there. And you just get that 00:00:12.619 --> 00:00:14.980 sinking feeling, right? That maybe it would just 00:00:14.980 --> 00:00:17.339 been faster to write the darn thing yourself. 00:00:17.399 --> 00:00:19.879 Oh, absolutely. That feeling is so common. And 00:00:19.879 --> 00:00:23.160 it really comes from treating these AIs like 00:00:23.160 --> 00:00:25.280 some kind of magic black box. We're just guessing, 00:00:25.500 --> 00:00:27.320 throwing prompts at the wall. Yeah, trial and 00:00:27.320 --> 00:00:30.359 error. Exactly and we're wasting so much potential 00:00:30.359 --> 00:00:34.320 doing that. So today we're gonna dive into four 00:00:34.320 --> 00:00:38.539 Really practical reliable ways to work with AI 00:00:38.539 --> 00:00:41.259 techniques that kind of turn that guesswork into 00:00:41.259 --> 00:00:44.329 Well predictable results. Yeah getting smarter 00:00:44.329 --> 00:00:46.869 about how we talk to the AI. Right. So the mission 00:00:46.869 --> 00:00:48.530 today is really clear, then. We want to help 00:00:48.530 --> 00:00:51.009 you move past that frustrating loop of back and 00:00:51.009 --> 00:00:53.090 forth, getting almost what you want. Yeah, we're 00:00:53.090 --> 00:00:55.250 aiming to give you a toolkit, a set of structured 00:00:55.250 --> 00:00:58.009 methods. OK. And these steps should help optimize 00:00:58.009 --> 00:01:00.729 time, but also, crucially, improve the quality 00:01:00.729 --> 00:01:03.170 of what you're getting back from the AI. Precisely. 00:01:03.670 --> 00:01:06.549 So the roadmap for this deep dive is super practical. 00:01:07.150 --> 00:01:09.469 First up, how to capture that perfect prompt. 00:01:09.719 --> 00:01:12.739 the one you need every single time. Second, we'll 00:01:12.739 --> 00:01:14.659 look at how you can take one really good piece 00:01:14.659 --> 00:01:18.519 of content and multiply it, turn it into maybe 00:01:18.519 --> 00:01:21.159 10 different things without losing that core 00:01:21.159 --> 00:01:23.840 quality. That sounds incredibly useful. Then 00:01:23.840 --> 00:01:26.920 third. Third, a neat trick for finding the weaknesses, 00:01:27.340 --> 00:01:29.719 the holes in your ideas before someone else does, 00:01:30.000 --> 00:01:33.599 like your boss or a client. Preemptive critique. 00:01:33.859 --> 00:01:38.250 I like that. Exactly. Finally, number four. How 00:01:38.250 --> 00:01:41.090 to structure those big complex projects so you 00:01:41.090 --> 00:01:43.489 don't just get back a giant messy wall of text. 00:01:43.609 --> 00:01:45.310 Which happens all the time. Okay, fantastic. 00:01:45.409 --> 00:01:47.189 Let's unpack this toolkit then. Where do we start? 00:01:47.310 --> 00:01:49.129 We kick things off with what we're calling the 00:01:49.129 --> 00:01:51.510 work backwards trick. This one is really the 00:01:51.510 --> 00:01:53.349 antidote to all that prompt guesswork. Okay, 00:01:53.469 --> 00:01:55.849 work backwards. I like the analogy you used earlier, 00:01:56.189 --> 00:01:58.530 finding a light switch in a dark room. Instead 00:01:58.530 --> 00:02:01.170 of just flailing around hoping to hit it, you 00:02:01.170 --> 00:02:03.310 somehow turn the light on first and then you 00:02:03.310 --> 00:02:06.010 ask the AI, hey, draw me a map straight to that 00:02:06.010 --> 00:02:08.379 switch for next time. Precisely. You nailed it. 00:02:08.439 --> 00:02:10.479 You don't stress about crafting the perfect prompt 00:02:10.479 --> 00:02:12.300 right at the beginning. You actually generate 00:02:12.300 --> 00:02:14.659 it at the end based on the perfect output. Ah, 00:02:14.740 --> 00:02:16.620 okay. So walk us through that. You mentioned 00:02:16.620 --> 00:02:19.199 five steps using maybe an email example. Yeah, 00:02:19.199 --> 00:02:21.960 let's use a common task. Writing a team email 00:02:21.960 --> 00:02:25.259 about a meeting. Step one. Start super simple, 00:02:25.379 --> 00:02:28.180 just the basics. Write an email to my team about 00:02:28.180 --> 00:02:31.180 the Monday 9 a .m. meeting on the Blue Sky Project. 00:02:31.340 --> 00:02:33.340 Okay, and you get back that first draft. Probably 00:02:33.340 --> 00:02:36.139 pretty robotic, right? Usually, yeah. So step 00:02:36.139 --> 00:02:40.349 two. you immediately give feedback. You look 00:02:40.349 --> 00:02:42.090 at it and say, okay, this is way too formal. 00:02:42.590 --> 00:02:44.330 Rewrite it, make the tone more friendly, maybe 00:02:44.330 --> 00:02:46.409 a bit excited. Oh, and add a quick joke about 00:02:46.409 --> 00:02:48.610 Monday mornings. All right, so you're iterating, 00:02:48.909 --> 00:02:51.810 then step three is more refining. Exactly. Keep 00:02:51.810 --> 00:02:53.810 tweaking. Maybe you say, actually make it much 00:02:53.810 --> 00:02:55.530 shorter. It needs to be readable in like under 00:02:55.530 --> 00:02:58.110 30 seconds. And you know what? Ditch the joke. 00:02:58.349 --> 00:03:01.819 It didn't land. But do add a clear call action, 00:03:02.000 --> 00:03:04.539 ask them to prep one idea, and boom, now the 00:03:04.539 --> 00:03:06.719 email is perfect. Just how you want it. OK, so 00:03:06.719 --> 00:03:08.900 we have the perfect email after a few rounds 00:03:08.900 --> 00:03:12.060 of feedback. What's next? This is where the magic 00:03:12.060 --> 00:03:15.280 happens. This is the magic step. Step four, instead 00:03:15.280 --> 00:03:17.419 of just copying that final perfect email and 00:03:17.419 --> 00:03:21.039 being done, you immediately give the AI a new 00:03:21.039 --> 00:03:23.580 instruction. You say, OK, now look back at our 00:03:23.580 --> 00:03:25.900 entire conversation, analyze everything we just 00:03:25.900 --> 00:03:29.050 did, and write me one single prompt. that would 00:03:29.050 --> 00:03:31.150 have produced this final email directly right 00:03:31.150 --> 00:03:33.409 from the start. Whoa. OK. So you're asking the 00:03:33.409 --> 00:03:36.590 AI to reverse engineer its own process to give 00:03:36.590 --> 00:03:39.030 you the ideal input. A -sicely. And it does. 00:03:39.090 --> 00:03:41.030 It gives you that golden prompt, which leads 00:03:41.030 --> 00:03:44.430 to step five. Save that prompt. Test it in a 00:03:44.430 --> 00:03:46.490 fresh chat to make sure it works. Then stick 00:03:46.490 --> 00:03:48.949 it in your prompt toolbox, Notion, Google Doc, 00:03:49.110 --> 00:03:51.969 whatever. Organized. OK. I see the value in having 00:03:51.969 --> 00:03:55.050 that golden prompt saved. But playing devil's 00:03:55.050 --> 00:03:57.330 advocate here, Doesn't this whole process mean 00:03:57.330 --> 00:03:59.909 you're kind of writing the email twice? Isn't 00:03:59.909 --> 00:04:02.189 that more work up front? It feels like it, but 00:04:02.189 --> 00:04:03.990 it's really investment. Think of a time saved 00:04:03.990 --> 00:04:07.229 on all future similar emails. But more importantly, 00:04:08.469 --> 00:04:10.129 it genuinely teaches you how to prompt better. 00:04:10.169 --> 00:04:13.389 How so? Because you see exactly which words made 00:04:13.389 --> 00:04:16.509 the difference. You see that asking for a friendly 00:04:16.509 --> 00:04:19.350 and excited tone worked better than just friendly. 00:04:19.910 --> 00:04:22.889 Or specifying under 30 seconds read time was 00:04:22.889 --> 00:04:27.370 key. You learn the AI's nuances. Ah, okay. So 00:04:27.370 --> 00:04:29.589 that upfront effort isn't just about this one 00:04:29.589 --> 00:04:32.290 email. It's about learning the specific descriptive 00:04:32.290 --> 00:04:34.389 language that gets results across the board. 00:04:34.649 --> 00:04:37.089 Exactly. Seeing those patterns highlights the 00:04:37.089 --> 00:04:39.509 words that truly move the needle. You become 00:04:39.509 --> 00:04:41.810 a better prompter. Got it. That makes sense. 00:04:42.250 --> 00:04:44.709 So once we've nailed getting that perfect prompt, 00:04:45.329 --> 00:04:48.000 what's the next challenge? Well, that knowledge 00:04:48.000 --> 00:04:50.360 is crucial because the next big hurdle is something 00:04:50.360 --> 00:04:52.579 called prompt drift, especially when you want 00:04:52.579 --> 00:04:55.579 to repurpose content. Ah, yes, prompt drift, 00:04:55.699 --> 00:04:58.259 where the AI kind of loses the plot over a long 00:04:58.259 --> 00:05:00.860 conversation, right? The context fades and the 00:05:00.860 --> 00:05:03.480 outputs get worse. Precisely. It loses focus. 00:05:03.800 --> 00:05:05.500 And I'll admit, I still wrestle with prompt drift 00:05:05.500 --> 00:05:07.500 myself sometimes, especially if I rush things. 00:05:07.579 --> 00:05:10.579 It happens. It does. And that leads us needily 00:05:10.579 --> 00:05:13.579 into our second trick, the content multiplier. 00:05:14.240 --> 00:05:16.680 This one tackles that common issue, where you 00:05:16.680 --> 00:05:19.139 have this amazing piece of content, say, big 00:05:19.139 --> 00:05:22.180 report or a great blog post, and it just sits 00:05:22.180 --> 00:05:24.579 there. Yeah, collecting digital dust. Right. 00:05:24.600 --> 00:05:26.720 So we use the AI like a super fast assistant 00:05:26.720 --> 00:05:29.500 editor. We take that one high quality piece and 00:05:29.500 --> 00:05:31.500 multiply it out into loads of different formats 00:05:31.500 --> 00:05:33.579 suitable for different platforms. OK, I love 00:05:33.579 --> 00:05:37.829 this idea. Taking a... Maybe 1500 word article 00:05:37.829 --> 00:05:40.589 and spinning it into what LinkedIn posts tweets 00:05:40.589 --> 00:05:43.029 and email summary exactly for LinkedIn posts 00:05:43.029 --> 00:05:46.310 three tweets and Internal memo you name it, but 00:05:46.310 --> 00:05:48.889 the execution needs structure to avoid that drift. 00:05:48.970 --> 00:05:52.129 Okay. How do we do it reliably three steps first? 00:05:52.350 --> 00:05:54.490 You paste the entire source material into the 00:05:54.490 --> 00:05:56.129 chat, the whole blog post, the whole report. 00:05:56.149 --> 00:05:58.230 All of it. Second, and this is key, you set the 00:05:58.230 --> 00:06:00.189 context. You give the AI a specific command. 00:06:00.730 --> 00:06:03.370 Reply only with the single word read once you 00:06:03.370 --> 00:06:05.569 have fully processed all of this text. Don't 00:06:05.569 --> 00:06:07.730 ask for anything else yet. Ah, so you're making 00:06:07.730 --> 00:06:09.730 sure it's actually ingested everything before 00:06:09.730 --> 00:06:11.889 you give it the real task, like a confirmation 00:06:11.889 --> 00:06:14.930 step. Exactly. It guarantees the model has the 00:06:14.930 --> 00:06:18.189 full context loaded. Only after it replies read, 00:06:18.569 --> 00:06:21.050 Do you move to step three? Which is asking for 00:06:21.050 --> 00:06:23.350 the different formats. Right. Now you say, OK, 00:06:23.649 --> 00:06:25.670 based on the test you just read, write me three 00:06:25.670 --> 00:06:28.529 tweets. Focus on different tips. Use a casual 00:06:28.529 --> 00:06:32.430 but expert tone. Or draft a short email newsletter 00:06:32.430 --> 00:06:35.230 summary. Needs a catchy intro and a strong call 00:06:35.230 --> 00:06:37.949 to action based on the main findings. Makes sense. 00:06:38.170 --> 00:06:40.310 But I guess the quality of the output still depends 00:06:40.310 --> 00:06:42.550 heavily on the input, right? The little garbage 00:06:42.550 --> 00:06:45.360 in, garbage out rule. Absolutely critical. This 00:06:45.360 --> 00:06:47.339 trick only works if you're multiplying your best 00:06:47.339 --> 00:06:49.920 stuff. High -quality source material is non -negotiable. 00:06:50.100 --> 00:06:52.519 You can't magically make a weak idea strong just 00:06:52.519 --> 00:06:54.680 by multiplying it. Good reminder. And you mentioned 00:06:54.680 --> 00:06:58.360 a pro tip about the audience. Yes. Always. Always 00:06:58.360 --> 00:07:01.600 tell the AI who the new format is for. This fights 00:07:01.600 --> 00:07:04.120 prompt drift and improves relevance dramatically. 00:07:04.879 --> 00:07:07.160 Just asking for a summary is bad. Too vague. 00:07:07.459 --> 00:07:10.060 Way too vague. Asking, write a three bullet point 00:07:10.060 --> 00:07:12.339 summary for my busy boss who only cares about 00:07:12.339 --> 00:07:15.319 the bottom line business impact. It's good prompting. 00:07:15.620 --> 00:07:18.500 Specific audience, specific needs. OK, so that 00:07:18.500 --> 00:07:21.240 context setting with the read confirmation, does 00:07:21.240 --> 00:07:24.000 that really make a measurable difference? Or 00:07:24.000 --> 00:07:27.139 is it just good practice? It makes a huge difference. 00:07:27.139 --> 00:07:31.459 It ensures the AI has fully processed your specific 00:07:31.459 --> 00:07:33.480 source material before it starts generating. 00:07:33.819 --> 00:07:36.180 So the outputs are grounded in your facts, not 00:07:36.180 --> 00:07:38.379 just its general knowledge base. It prevents 00:07:38.379 --> 00:07:41.339 hallucination and keeps things relevant. Gotcha. 00:07:41.379 --> 00:07:43.279 So it guarantees the AI is working off the right 00:07:43.279 --> 00:07:45.459 playbook. Okay, that brings us to trick number 00:07:45.459 --> 00:07:49.560 three. The other side trick. Now this one, this 00:07:49.560 --> 00:07:51.579 might be the most valuable, I think, for high 00:07:51.579 --> 00:07:54.000 stakes communication. Things like proposals, 00:07:54.240 --> 00:07:57.139 asking for a raise, pitching investors. Okay, 00:07:57.259 --> 00:08:00.439 the other side. Sounds intriguing, like seeing 00:08:00.439 --> 00:08:02.519 things from another perspective. Exactly. It's 00:08:02.519 --> 00:08:05.319 like having a really smart, maybe slightly cynical 00:08:05.319 --> 00:08:08.220 friend. play devil's advocate for you. You create 00:08:08.220 --> 00:08:10.420 your piece of content, your pitch, whatever it 00:08:10.420 --> 00:08:12.579 is. And then you immediately flip the script. 00:08:12.600 --> 00:08:14.860 You make the AI act as your toughest critic. 00:08:15.399 --> 00:08:17.279 It stress tests your argument before you face 00:08:17.279 --> 00:08:19.420 the real deal. OK, let's use that raise request 00:08:19.420 --> 00:08:23.459 example. So I draft my email, I outline my achievements, 00:08:23.920 --> 00:08:26.500 why I deserve more money, all the good stuff. 00:08:26.680 --> 00:08:29.639 Perfect. You've got your draft. Then prompt number 00:08:29.639 --> 00:08:32.500 two, right away. What does that look like? That's 00:08:32.500 --> 00:08:34.399 what the magic is. You say something like, OK, 00:08:34.480 --> 00:08:39.120 AI. Now, you are Jane, my manager. She's super 00:08:39.120 --> 00:08:41.279 busy. She has a really tight budget this quarter, 00:08:41.539 --> 00:08:44.340 and she's generally risk -averse. Read my email. 00:08:44.639 --> 00:08:47.220 What's your immediate gut reaction? Which specific 00:08:47.220 --> 00:08:49.679 sentence annoys you the most? And give me the 00:08:49.679 --> 00:08:52.679 top three reasons you'd probably say no right 00:08:52.679 --> 00:08:56.299 now. Wow. Okay. That's direct. But I can see 00:08:56.299 --> 00:08:58.320 the power. It forces you to see your own blind 00:08:58.320 --> 00:09:00.779 spots, doesn't it? Totally. When we write something 00:09:00.779 --> 00:09:03.159 persuasive, we're naturally defending our position. 00:09:03.200 --> 00:09:05.509 We think it's great. Of course. But the audience, 00:09:05.669 --> 00:09:08.470 the boss, the client, the investor, they're often 00:09:08.470 --> 00:09:11.210 actively looking for reasons to say no or poke 00:09:11.210 --> 00:09:14.110 holes in it. This trick helps you find and fix 00:09:14.110 --> 00:09:16.710 those weaknesses first. You preempt the objections. 00:09:16.970 --> 00:09:19.009 That's incredibly useful. Are there any specific 00:09:19.009 --> 00:09:20.929 tips for making this critique really effective? 00:09:21.289 --> 00:09:23.750 Yes. Three key things. First, as we just touched 00:09:23.750 --> 00:09:27.049 on, be extremely specific about the persona you 00:09:27.049 --> 00:09:29.529 want the AI to adopt. Don't just say, be a critic. 00:09:29.889 --> 00:09:33.169 Right. Be Jane, the busy, budget -conscious manager. 00:09:33.250 --> 00:09:35.889 Or, you're a 50 -year -old chief financial officer. 00:09:36.210 --> 00:09:38.450 You hate unnecessary risk. Your main concern 00:09:38.450 --> 00:09:41.909 is Q4 cost savings. The more detailed the persona, 00:09:42.090 --> 00:09:44.649 the better the critique. OK, specificity is number 00:09:44.649 --> 00:09:48.120 one. What's two? Ask for a ranking. Don't just 00:09:48.120 --> 00:09:50.200 ask for weaknesses, ask for the top weaknesses. 00:09:50.980 --> 00:09:53.500 List the top three weakest points in my argument 00:09:53.500 --> 00:09:56.179 in order from most serious down to least serious. 00:09:56.720 --> 00:09:59.240 Ah, that helps prioritize revisions. Smart. It 00:09:59.240 --> 00:10:01.440 focuses your energy where it matters most. And 00:10:01.440 --> 00:10:04.399 the third pro tip, always close the loop. Close 00:10:04.399 --> 00:10:07.139 the loop? How? Don't just stop at identifying 00:10:07.139 --> 00:10:09.299 the flaws. Immediately follow up and ask the 00:10:09.299 --> 00:10:11.840 AI for help fixing them. OK, based on the top 00:10:11.840 --> 00:10:13.899 three weaknesses you just pointed out, help me 00:10:13.899 --> 00:10:15.820 rewrite the specific sentences that are most 00:10:15.820 --> 00:10:18.399 problematic. Turn that critique directly into 00:10:18.399 --> 00:10:20.399 actionable improvement. That makes perfect sense. 00:10:20.600 --> 00:10:23.279 Identify rank, then fix with the AI's help. But 00:10:23.279 --> 00:10:26.460 quick question. If we make that persona too harsh, 00:10:26.720 --> 00:10:29.840 like that super skeptical CFO, is there a risk 00:10:29.840 --> 00:10:32.360 we over -correct? Make the pitch too defensive 00:10:32.360 --> 00:10:34.740 or negative. That's a fair point. You don't want 00:10:34.740 --> 00:10:37.279 to water down your core message completely. The 00:10:37.279 --> 00:10:40.080 goal is balanced preparation. By focusing on 00:10:40.080 --> 00:10:42.519 the ranked weaknesses, you address the most critical 00:10:42.519 --> 00:10:45.179 issues identified by that tough persona without 00:10:45.179 --> 00:10:47.740 necessarily overhauling everything. It's about 00:10:47.740 --> 00:10:50.669 strengthening, not undermining. Right. balanced 00:10:50.669 --> 00:10:52.490 prep, focusing on the biggest ranked weaknesses. 00:10:52.710 --> 00:10:55.350 Got it. OK, that leads us to the final trick 00:10:55.350 --> 00:10:57.870 in the toolkit. Indeed, the blueprint trick. 00:10:58.250 --> 00:11:00.529 This one directly tackles that frustrating problem 00:11:00.529 --> 00:11:02.649 we mentioned earlier, asking for something complex 00:11:02.649 --> 00:11:05.090 and just getting back this huge, undifferentiated, 00:11:05.250 --> 00:11:08.210 often generic wall of text. The wall of text, 00:11:08.409 --> 00:11:11.470 yes. How does the blueprint help? It forces the 00:11:11.470 --> 00:11:14.450 AI to think about structure first before generating 00:11:14.450 --> 00:11:18.090 the content. It's exactly like a finalizing the 00:11:18.090 --> 00:11:20.090 detailed architectural blueprint for a house 00:11:20.090 --> 00:11:22.570 before you pour a single drop of concrete. Okay, 00:11:22.570 --> 00:11:25.490 I see. If you spot a flaw in the blueprint, you 00:11:25.490 --> 00:11:27.549 fix it instantly, drag a wall, change a room 00:11:27.549 --> 00:11:31.009 size. It's easy on paper or on the screen. But 00:11:31.009 --> 00:11:34.970 if you find a major structural flaw after the 00:11:34.970 --> 00:11:37.610 AI has already generated 10 pages of detailed 00:11:37.610 --> 00:11:40.409 interconnected text based on a bad structure... 00:11:40.409 --> 00:11:43.289 You have to tear it all down. Hours wasted. Exactly. 00:11:43.649 --> 00:11:46.440 Huge waste of time and effort. So the trick is 00:11:46.440 --> 00:11:49.960 to insert a specific blueprint request into your 00:11:49.960 --> 00:11:51.980 initial prompt. How would that look? Say, for 00:11:51.980 --> 00:11:54.299 that travel blog content plan example. Instead 00:11:54.299 --> 00:11:56.179 of just saying, write me a content plan for my 00:11:56.179 --> 00:11:58.720 travel blog, you'd say something like, I need 00:11:58.720 --> 00:12:01.620 a detailed content plan for my travel blog. But 00:12:01.620 --> 00:12:04.600 first, outline the standard sections typically 00:12:04.600 --> 00:12:06.960 found in a professional content plan. For each 00:12:06.960 --> 00:12:08.799 section, just give me a one sentence description. 00:12:09.179 --> 00:12:12.059 Do not write the full content plan yet. Uh, so 00:12:12.059 --> 00:12:13.960 you're explicitly telling it not to write the 00:12:13.960 --> 00:12:16.279 content, just the structure first. Precisely. 00:12:16.480 --> 00:12:19.519 That's step one. Step two. You review that structural 00:12:19.519 --> 00:12:21.639 outline it provides, it might list things like 00:12:21.639 --> 00:12:25.080 audience analysis, content pillars, keyword strategy, 00:12:25.700 --> 00:12:27.639 editorial calendar, promotion plan. Right, the 00:12:27.639 --> 00:12:31.570 standard components. Then, step three. you correct 00:12:31.570 --> 00:12:34.009 the course before the heavy lifting starts. You 00:12:34.009 --> 00:12:36.029 look at the blueprint and say, okay, wait, I 00:12:36.029 --> 00:12:37.669 already know my audience really well, so remove 00:12:37.669 --> 00:12:40.190 audience analysis, and I handle promotion separately, 00:12:40.269 --> 00:12:43.169 so remove promotion plan two. I only want you 00:12:43.169 --> 00:12:45.470 to focus on content pillars, keyword strategy, 00:12:45.830 --> 00:12:48.610 and the editorial calendar. So you approve the 00:12:48.610 --> 00:12:50.730 specific sections you want included. You approve 00:12:50.730 --> 00:12:53.730 the final blueprint, and then step four, you 00:12:53.730 --> 00:12:56.350 tell the AI, okay, now generate the full content 00:12:56.350 --> 00:12:58.470 plan, but only based on these specific sections 00:12:58.470 --> 00:13:02.129 we just agreed on. The result is targeted, structured, 00:13:02.350 --> 00:13:05.509 and exactly what you need. No waste. Wow. Okay, 00:13:05.710 --> 00:13:07.909 that's incredibly efficient. You can just imagine 00:13:07.909 --> 00:13:10.470 scaling that kind of structured planning across, 00:13:10.470 --> 00:13:14.070 I don't know, dozens of big complex business 00:13:14.070 --> 00:13:16.289 reports or technical documents. That feels like 00:13:16.289 --> 00:13:19.350 real leverage. It really is. True leverage. So, 00:13:19.509 --> 00:13:22.929 just to crystallize it. What's the main danger, 00:13:23.009 --> 00:13:25.409 the biggest risk, if you skip doing this blueprint 00:13:25.409 --> 00:13:27.649 step when you're tackling a really big multi 00:13:27.649 --> 00:13:30.330 -part project with an AI? The biggest danger 00:13:30.330 --> 00:13:33.830 is wasted time and effort on a massive scale. 00:13:33.960 --> 00:13:37.799 You risk the AI generating pages and pages, maybe 00:13:37.799 --> 00:13:42.000 10, 20 pages of detailed content based on a flawed 00:13:42.000 --> 00:13:44.639 or irrelevant structure. And then you're forced 00:13:44.639 --> 00:13:46.840 to scrap most of it and start over rebuilding 00:13:46.840 --> 00:13:49.299 from the ground up. Right, you waste hours generating 00:13:49.299 --> 00:13:51.940 irrelevant stuff that has to be torn down. Okay, 00:13:52.279 --> 00:13:55.370 blueprint first. Got it. Okay, so let's recap 00:13:55.370 --> 00:13:57.429 the toolkit we have assembled here. It's pretty 00:13:57.429 --> 00:13:59.110 powerful when you put it all together. Yeah, 00:13:59.190 --> 00:14:01.330 definitely. We've got the work backwards trick 00:14:01.330 --> 00:14:03.429 that's for capturing those perfect prompts and 00:14:03.429 --> 00:14:07.009 learning the AIs language. Then the content multiplier 00:14:07.009 --> 00:14:09.450 for taking your best stuff and scaling it efficiently 00:14:09.450 --> 00:14:11.629 across different formats. Then the other side 00:14:11.629 --> 00:14:14.230 trick for stress testing your crucial ideas and 00:14:14.230 --> 00:14:16.529 finding weaknesses before anyone else does. Then 00:14:16.529 --> 00:14:19.250 finally the blueprint trick, ensuring structured 00:14:19.250 --> 00:14:22.409 logical output for complex projects, avoiding 00:14:22.409 --> 00:14:25.289 that dreaded wall of text. It really feels like 00:14:25.289 --> 00:14:27.649 these build on each other. They absolutely do. 00:14:27.730 --> 00:14:30.929 They stack together like Lego blocks of data. 00:14:31.070 --> 00:14:34.409 almost. Think about creating a full social media 00:14:34.409 --> 00:14:36.750 campaign plan, for instance. You could use all 00:14:36.750 --> 00:14:38.809 four. OK, how would that work? Walk me through 00:14:38.809 --> 00:14:41.330 it. All right. First, you'd use the blueprint 00:14:41.330 --> 00:14:44.289 trick. You'd force the AI to outline the essential 00:14:44.289 --> 00:14:47.149 sections of a good campaign plan goals, target 00:14:47.149 --> 00:14:49.549 audience, key messages, content assets, schedule, 00:14:49.669 --> 00:14:52.029 metrics, before it writes a single word of the 00:14:52.029 --> 00:14:54.570 actual plan. OK, structure first. Makes sense. 00:14:54.830 --> 00:14:57.370 Then, once you have a draft plan based on that 00:14:57.370 --> 00:15:00.049 approved blueprint, you immediately use the other 00:15:00.049 --> 00:15:03.230 side trick. You tell the AI, OK, now be a cynical 00:15:03.230 --> 00:15:05.509 follower of our brand. You hate feeling sold 00:15:05.509 --> 00:15:07.809 to. Read this campaign plan. What feels off? 00:15:07.889 --> 00:15:10.029 What sounds too salesy? What part of the schedule 00:15:10.029 --> 00:15:12.509 makes no sense? Ah, getting that critical audience 00:15:12.509 --> 00:15:15.649 perspective early. Nice. Exactly. Once you've 00:15:15.649 --> 00:15:17.649 refined the plan based on that critique, then 00:15:17.649 --> 00:15:19.830 you bring in the content multiplier. You take 00:15:19.830 --> 00:15:22.389 the main campaign announcement post, maybe, and 00:15:22.389 --> 00:15:25.490 tell the AI, OK, turn this into five short Instagram 00:15:25.490 --> 00:15:28.330 story scripts, a longer email for our subscriber 00:15:28.330 --> 00:15:30.809 list, and bullet points for a short explainer 00:15:30.809 --> 00:15:33.289 video. Multiplying the core message efficiently. 00:15:33.570 --> 00:15:36.730 Got it. And the last piece. Finally, because 00:15:36.730 --> 00:15:40.309 that whole process, blueprint, critique, multiply, 00:15:40.610 --> 00:15:43.070 worked so well and gave you great results, you 00:15:43.070 --> 00:15:45.470 use the work backwards trick. You ask the AI. 00:15:46.009 --> 00:15:48.429 Analyze this whole workflow. Give me the optimized 00:15:48.429 --> 00:15:50.750 set of props that achieve this entire campaign 00:15:50.750 --> 00:15:53.690 plan and asset generation. And you save that 00:15:53.690 --> 00:15:56.350 sequence for the next campaign launch. Wow. OK. 00:15:56.350 --> 00:15:58.210 Seeing it stacked like that really shows the 00:15:58.210 --> 00:16:00.970 power. Blueprint, then other side, then multiplier, 00:16:01.090 --> 00:16:03.049 then work backwards to save the whole workflow. 00:16:03.429 --> 00:16:05.509 That's a system. It really is a system. And as 00:16:05.509 --> 00:16:07.210 you start using these, just remember the two 00:16:07.210 --> 00:16:09.309 big pitfalls we talked about. Yeah. Critical 00:16:09.309 --> 00:16:11.519 mistakes to avoid. Right. Number one was being 00:16:11.519 --> 00:16:13.419 too vague when you use the other side critique, 00:16:13.460 --> 00:16:15.500 wasn't it? Exactly. Don't just say be a critic. 00:16:15.820 --> 00:16:18.899 Be super specific about the persona, the role, 00:16:18.960 --> 00:16:21.480 the concerns, the biases you want the AI to embody. 00:16:21.820 --> 00:16:24.779 Specificity and critique. And number two. Forgetting 00:16:24.779 --> 00:16:27.500 to close the loop. Don't just find the weaknesses 00:16:27.500 --> 00:16:30.259 with the other side trick. Always take that next 00:16:30.259 --> 00:16:33.500 step. Ask the AI to actively help you rewrite 00:16:33.500 --> 00:16:36.720 and fix the specific sentences or sections it 00:16:36.720 --> 00:16:39.360 identified as flawed. Turn insight into action. 00:16:39.440 --> 00:16:43.519 Identify, then immediately fix. OK, so for listeners 00:16:43.519 --> 00:16:45.379 wanting to try this out, what's the best way 00:16:45.379 --> 00:16:47.740 to start? It feels like a lot to implement all 00:16:47.740 --> 00:16:50.019 at once. Yeah, definitely don't try to master 00:16:50.019 --> 00:16:52.519 all four overnight. That's overwhelming. The 00:16:52.519 --> 00:16:55.240 best approach is to pick the one trick that addresses 00:16:55.240 --> 00:16:57.919 your biggest pain point right now. OK, tackle 00:16:57.919 --> 00:17:00.620 the biggest frustration first. Exactly. If you 00:17:00.620 --> 00:17:02.840 find yourself constantly wasting time tweaking 00:17:02.840 --> 00:17:06.220 initial prompts, start with work backwards. Spend 00:17:06.220 --> 00:17:08.799 an hour really nailing that process for one type 00:17:08.799 --> 00:17:11.599 of task you do often. Makes sense. Or, if you 00:17:11.599 --> 00:17:13.519 have a really important presentation or proposal 00:17:13.519 --> 00:17:16.579 coming up, try the other side trick today. Stress 00:17:16.579 --> 00:17:18.660 test your arguments before you finalize them. 00:17:18.940 --> 00:17:21.259 Just pick one method, apply it to a real task 00:17:21.259 --> 00:17:23.500 this week, and just observe the difference it 00:17:23.500 --> 00:17:26.140 makes. That focused approach feels much more 00:17:26.140 --> 00:17:28.240 manageable. Start with one, feel the benefit, 00:17:28.339 --> 00:17:31.500 then maybe add another later. Precisely. And... 00:17:31.480 --> 00:17:34.940 You know, ultimately, the real deep value here 00:17:34.940 --> 00:17:37.640 isn't just about getting slightly better output 00:17:37.640 --> 00:17:39.759 from the AI, although that's great. What's the 00:17:39.759 --> 00:17:43.200 deeper value then? By consciously using these 00:17:43.200 --> 00:17:46.079 methods, forcing the AI to reflect on its process, 00:17:46.519 --> 00:17:49.180 to critique from another perspective, to outline 00:17:49.180 --> 00:17:52.220 before creating this game, you're actually training 00:17:52.220 --> 00:17:54.559 yourself. You're building your own critical thinking 00:17:54.559 --> 00:17:57.619 muscles, your own strategic discipline. Demanding 00:17:57.619 --> 00:17:59.819 structure from the AI forces you to become a 00:17:59.819 --> 00:18:02.369 more structured thinker. That's a really interesting 00:18:02.369 --> 00:18:05.049 final thought. We become better thinkers by demanding 00:18:05.049 --> 00:18:07.109 better thinking processes from the tools we use. 00:18:07.329 --> 00:18:08.910 I like that. It's a powerful feedback loop.