WEBVTT 00:00:00.000 --> 00:00:02.580 If you've ever actually tried to read a 10K, 00:00:02.700 --> 00:00:05.940 you know, that huge 100 -page annual report from 00:00:05.940 --> 00:00:08.000 a big company, you know the feeling. Your brain 00:00:08.000 --> 00:00:10.119 just gets completely bogged down. All that complexity, 00:00:10.160 --> 00:00:14.259 it just turns into this dense, overwhelming fog 00:00:14.259 --> 00:00:16.920 of detail. Yeah, it's instant information overload, 00:00:17.059 --> 00:00:19.179 isn't it? But, okay, now imagine you have this 00:00:19.179 --> 00:00:22.260 tireless, completely emotion -free research assistant, 00:00:22.699 --> 00:00:25.739 one who can read, like... thousands of pages 00:00:25.739 --> 00:00:28.579 and seconds and instantly find the exact financial 00:00:28.579 --> 00:00:30.899 ratios or the specific risk factors you're looking 00:00:30.899 --> 00:00:33.020 for. That's really the power of AI we're going 00:00:33.020 --> 00:00:35.280 to try and unlock today. Welcome to the deep 00:00:35.280 --> 00:00:37.479 dive. We're doing a specialized deep dive today 00:00:37.479 --> 00:00:39.619 looking at source material focus on using some 00:00:39.619 --> 00:00:42.600 pretty advanced AI tools, things like specialized 00:00:42.600 --> 00:00:45.840 versions of chat GPT and notebook LM to basically 00:00:45.840 --> 00:00:48.179 revolutionize how you can do smarter investing. 00:00:48.759 --> 00:00:51.380 Exactly. And our mission here is to get past 00:00:51.380 --> 00:00:54.119 the simple summaries you see everywhere and give 00:00:54.119 --> 00:00:56.700 you a real actionable roadmap. We're breaking 00:00:56.700 --> 00:00:59.539 down five core steps today. Things like sophisticated 00:00:59.539 --> 00:01:03.200 charting analysis, coding your own custom indicators, 00:01:03.280 --> 00:01:06.000 even if you can't code, automating that dense 00:01:06.000 --> 00:01:08.640 data extraction from PDFs, cross -referencing 00:01:08.640 --> 00:01:11.219 legal documents, which is super important, and 00:01:11.219 --> 00:01:13.239 finally, building an auto -updating tracking 00:01:13.239 --> 00:01:15.980 dashboard. The real key takeaway here, I think, 00:01:15.980 --> 00:01:18.079 is just a radical boost in your research. speed 00:01:18.079 --> 00:01:20.719 and depth. Okay, so let's maybe unpack the foundational 00:01:20.719 --> 00:01:22.920 mindset first, because this is important, right? 00:01:23.000 --> 00:01:25.599 A lot of newer investors, they kind of mistakenly 00:01:25.599 --> 00:01:27.599 believe they can treat AI like some kind of oracle. 00:01:28.000 --> 00:01:30.260 They just ask, should I buy Tesla next week? 00:01:30.430 --> 00:01:32.549 or something. Right. And that expectation is, 00:01:32.549 --> 00:01:34.810 well, it's fundamentally flawed. AI is really 00:01:34.810 --> 00:01:36.769 at its most powerful when you combine it with 00:01:36.769 --> 00:01:39.489 human knowledge and, critically, market judgment. 00:01:40.150 --> 00:01:42.590 You really need to view this tech as a, hmm, 00:01:42.870 --> 00:01:46.549 let's say a dedicated, smart, and just relentlessly 00:01:46.549 --> 00:01:49.030 hardworking research assistant, not some kind 00:01:49.030 --> 00:01:51.010 of guru. So it's not the decision maker itself. 00:01:51.069 --> 00:01:53.719 It's more like the processor. Exactly. And AI 00:01:53.719 --> 00:01:56.140 solves, I think, three huge problems for any 00:01:56.140 --> 00:01:59.379 serious investor. First is just information overload. 00:01:59.459 --> 00:02:02.180 It can rapidly summarize the essential bits from 00:02:02.180 --> 00:02:05.719 massive reports and news feeds. Second, the time 00:02:05.719 --> 00:02:07.920 constraint issue. It handles those repetitive 00:02:07.920 --> 00:02:10.439 jobs like gathering data points completely automatically, 00:02:10.639 --> 00:02:13.539 saves so much time. And third, maybe the most 00:02:13.539 --> 00:02:16.620 interesting, emotional decisions. It strictly 00:02:16.620 --> 00:02:19.340 looks at data logically. That helps you avoid 00:02:19.340 --> 00:02:22.400 making choices driven purely by like, short -term 00:02:22.400 --> 00:02:25.139 fear or greed. Mm -hmm. That emotional distance, 00:02:25.340 --> 00:02:27.159 that's crucial in markets, isn't it? So okay, 00:02:27.240 --> 00:02:29.379 if AI is handling all the heavy lifting of the 00:02:29.379 --> 00:02:31.539 research side, what's left for the investor? 00:02:31.620 --> 00:02:34.080 What's their remaining most critical job? Applying 00:02:34.080 --> 00:02:36.099 their own unique knowledge and ultimately their 00:02:36.099 --> 00:02:38.319 final judgment to all that analyzed data. That's 00:02:38.319 --> 00:02:40.199 the human element. Okay, let's jump straight 00:02:40.199 --> 00:02:42.120 into the technical side then. Starting with step 00:02:42.120 --> 00:02:46.430 one, using AI for... quantitative technical analysis. 00:02:46.610 --> 00:02:48.750 And this really needs specificity. You can't 00:02:48.750 --> 00:02:50.870 just ask, what's this chart doing? We need to 00:02:50.870 --> 00:02:53.490 prompt the AI like a domain expert would. Yeah, 00:02:53.509 --> 00:02:54.930 we looked at a really good example involving 00:02:54.930 --> 00:02:58.569 a chart for Tesla, ticker TSLA. You upload an 00:02:58.569 --> 00:03:01.569 image, one that shows common indicators like 00:03:01.569 --> 00:03:04.439 RSI. which measures the speed and change of price 00:03:04.439 --> 00:03:06.780 moves, and Mazze for momentum. And then you give 00:03:06.780 --> 00:03:09.800 it this highly structured prompt. That's the 00:03:09.800 --> 00:03:12.520 secret weapon, really. We basically forced the 00:03:12.520 --> 00:03:14.819 AI to produce four distinct components in its 00:03:14.819 --> 00:03:18.180 analysis. One, trend and structure. So identify 00:03:18.180 --> 00:03:20.620 the main trend and find key support or resistance 00:03:20.620 --> 00:03:24.620 zones. Two, volatility analysis. Are the Bollinger 00:03:24.620 --> 00:03:26.979 Bands the ones that track price deviations? Are 00:03:26.979 --> 00:03:29.120 they contracting or expanding? Tells you about 00:03:29.120 --> 00:03:31.240 volatility. Right. And three is the momentum 00:03:31.240 --> 00:03:34.349 diagnosis. specifically checking the RSI for 00:03:34.349 --> 00:03:37.009 like overbought or oversold signals and looking 00:03:37.009 --> 00:03:40.169 for those MedZ crossovers. And the fourth component, 00:03:40.330 --> 00:03:43.030 this is maybe the most vital, synthesis. This 00:03:43.030 --> 00:03:45.629 forces the AI to actually put all three factors 00:03:45.629 --> 00:03:47.770 together for a probabilistic short -term outlook. 00:03:48.129 --> 00:03:49.969 This is where we integrate that golden rule of 00:03:49.969 --> 00:03:51.889 prompting we talked about. Always break the problem 00:03:51.889 --> 00:03:54.629 down. I noticed that. That four -part structure, 00:03:55.009 --> 00:03:57.669 it makes sure the AI can't just... waffle or 00:03:57.669 --> 00:03:59.770 give you some vague prediction right has a check 00:03:59.770 --> 00:04:03.099 multiple specific boxes and the output might 00:04:03.099 --> 00:04:05.419 highlight something really specific, like a bullish 00:04:05.419 --> 00:04:08.120 divergence. Precisely. And a bullish divergence, 00:04:08.120 --> 00:04:11.300 that's a highly specific technical signal. It's 00:04:11.300 --> 00:04:13.599 where the stock price makes a lower low, but 00:04:13.599 --> 00:04:16.240 the RSI indicator, it doesn't follow along. It 00:04:16.240 --> 00:04:18.980 makes a higher low instead. And that signals 00:04:18.980 --> 00:04:22.079 that selling pressure might be weakening, even 00:04:22.079 --> 00:04:24.160 if the price itself is still dropping. It's a 00:04:24.160 --> 00:04:26.259 human insight, really, but delivered by machine 00:04:26.259 --> 00:04:29.120 analysis. So how does forcing the AI to combine 00:04:29.120 --> 00:04:31.480 these specific technical indicators, how does 00:04:31.480 --> 00:04:34.189 that elevate the analysis beyond just a general 00:04:34.189 --> 00:04:37.149 summary. It compels a highly structured you could 00:04:37.149 --> 00:04:39.810 say expert level breakdown of those confluence 00:04:39.810 --> 00:04:42.550 factors. It forces rigor. Okay now here's where 00:04:42.550 --> 00:04:44.350 for me anyway gets really interesting especially 00:04:44.350 --> 00:04:47.120 for the quantitatively curious folks. turning 00:04:47.120 --> 00:04:51.540 a complex trading idea into actual working PineScript 00:04:51.540 --> 00:04:54.319 code for platforms like TradingView. That's step 00:04:54.319 --> 00:04:56.100 two. And it sounds like it's accessible even 00:04:56.100 --> 00:04:58.860 if you, like me, can't really code. Oh, it's 00:04:58.860 --> 00:05:01.220 a total game changer for scaling up your analysis. 00:05:01.339 --> 00:05:04.439 It really is. Let's detail that bullish pullback 00:05:04.439 --> 00:05:07.079 setup strategy from the source. It's actually 00:05:07.079 --> 00:05:09.439 quite complex. It requires four conditions to 00:05:09.439 --> 00:05:12.240 be met simultaneously. First, the uptrend condition. 00:05:12.750 --> 00:05:15.009 The price has to be trading above both the 50 00:05:15.009 --> 00:05:17.230 -day and the 200 -day simple moving averages, 00:05:17.430 --> 00:05:20.670 the SMAs. Clear uptrend. OK, makes sense. Second, 00:05:20.769 --> 00:05:23.329 the pullback condition. Yeah. The price needs 00:05:23.329 --> 00:05:26.189 to fall back, sort of pause, near the 20 -day 00:05:26.189 --> 00:05:28.350 exponential moving average, the EMAs, the market 00:05:28.350 --> 00:05:30.550 takes a breather. Exactly. And then come the 00:05:30.550 --> 00:05:33.230 crucial confirmation layers. Third, the healthy 00:05:33.230 --> 00:05:36.060 momentum condition. During that pullback, the 00:05:36.060 --> 00:05:39.360 RSI indicator must hold above the 40 level. This 00:05:39.360 --> 00:05:41.240 suggests sellers haven't really taken control 00:05:41.240 --> 00:05:44.339 despite the dip. Right. And fourth, volume exhaustion. 00:05:44.860 --> 00:05:47.079 The volume on that specific signal candle, the 00:05:47.079 --> 00:05:49.259 one that might trigger the entry, it must be 00:05:49.259 --> 00:05:51.240 lower than the average volume of the previous 00:05:51.240 --> 00:05:54.319 10 candles. So you feed those four detailed rules 00:05:54.319 --> 00:05:57.339 to the AI, and it just generates executable code 00:05:57.339 --> 00:06:00.060 that scans the market for you. Yeah. It spits 00:06:00.060 --> 00:06:02.300 out PineScript you can plug right into TradingView. 00:06:02.860 --> 00:06:05.220 The feeling, honestly, of turning this abstract 00:06:05.220 --> 00:06:08.100 multi -part strategy into code that just runs. 00:06:08.339 --> 00:06:11.939 It's powerful. Beat. Whoa. I mean, imagine scaling 00:06:11.939 --> 00:06:15.160 this kind of system across thousands of stocks 00:06:15.160 --> 00:06:18.160 instantly, running these complex screens constantly. 00:06:18.240 --> 00:06:21.480 That sounds incredible, yeah. But, okay, if I'm 00:06:21.480 --> 00:06:23.600 not a coder myself, how much risk am I taking? 00:06:23.790 --> 00:06:25.930 I mean, couldn't the AI -generated pine script 00:06:25.930 --> 00:06:28.629 have some subtle bug I wouldn't catch? Isn't 00:06:28.629 --> 00:06:30.649 this just trading one kind of complexity for 00:06:30.649 --> 00:06:33.189 another? That's a fair point. It is a trade -off. 00:06:33.449 --> 00:06:35.389 To be honest, I still wrestle with prompt drift 00:06:35.389 --> 00:06:37.509 myself sometimes where the AI doesn't quite get 00:06:37.509 --> 00:06:39.810 it right. And this is where we learn the value 00:06:39.810 --> 00:06:41.629 of using different models for their strengths. 00:06:42.149 --> 00:06:43.829 You might use a model like Claude, which seems 00:06:43.829 --> 00:06:46.149 to be better at writing and fixing complex code, 00:06:46.149 --> 00:06:48.649 and then, this is non -negotiable, you must back 00:06:48.649 --> 00:06:51.170 -test it rigorously and visually verify the results 00:06:51.170 --> 00:06:54.069 on charts. You can't just trust it blindly. Okay, 00:06:54.069 --> 00:06:56.449 that makes sense. Trust, but verify. Heavily. 00:06:56.790 --> 00:06:59.269 Alright, moving to step three. Automating the 00:06:59.269 --> 00:07:02.449 fundamental data extraction. Using AI to analyze 00:07:02.449 --> 00:07:05.610 those official PDF annual reports. The 10Ks. 00:07:05.949 --> 00:07:08.709 This used to take analysts... Days, right? Manual 00:07:08.709 --> 00:07:11.290 data entry for really rigorous work. Oh, easily. 00:07:11.509 --> 00:07:14.649 Days. Our source material used a detailed Microsoft 00:07:14.649 --> 00:07:17.790 MSFT example here. And the key was that the prompt 00:07:17.790 --> 00:07:20.930 demanded several things way beyond just raw data. 00:07:21.110 --> 00:07:23.069 It wasn't just asking what's the revenue. That's 00:07:23.069 --> 00:07:25.310 too simple. Right. It required extracting four 00:07:25.310 --> 00:07:28.769 specific core metrics. Total revenue, gross profit, 00:07:29.170 --> 00:07:31.689 operating income, and R &D spending for three 00:07:31.689 --> 00:07:34.449 consecutive years. Yes. And then, crucially, 00:07:34.709 --> 00:07:37.310 it had to calculate three specific ratios based 00:07:37.310 --> 00:07:39.310 on that data, like gross margin and operating 00:07:39.310 --> 00:07:42.089 margin, all within the same prompt request. But 00:07:42.089 --> 00:07:44.949 here's the really critical part. The prompt demanded 00:07:44.949 --> 00:07:47.329 the AI provide the exact page number from the 00:07:47.329 --> 00:07:49.769 source document for every single number it extracted. 00:07:50.230 --> 00:07:52.050 Ah, that verification step again. Absolutely 00:07:52.050 --> 00:07:54.810 key. That page number requirement, it turns the 00:07:54.810 --> 00:07:57.110 AI from just a summarizer into a precise data 00:07:57.110 --> 00:07:59.910 locator, doesn't it? Exactly. So what's the biggest 00:07:59.910 --> 00:08:02.069 vulnerability then when you're trusting an AI 00:08:02.069 --> 00:08:05.209 to generate code or pull financial ratios directly 00:08:05.209 --> 00:08:08.230 from a PDF like that the risk of hallucination 00:08:08.230 --> 00:08:11.230 basically the AI making things up or just outright 00:08:11.230 --> 00:08:14.170 errors which means those manual checks using 00:08:14.170 --> 00:08:16.410 those page numbers are absolutely necessary if 00:08:16.410 --> 00:08:21.170 double -check okay moving to step four deploying 00:08:21.170 --> 00:08:23.449 the AI almost like an investigative journalist 00:08:23.449 --> 00:08:26.509 using tools like notebook LM can you define notebook 00:08:26.509 --> 00:08:29.889 LM for us quickly why use that specifically over 00:08:29.889 --> 00:08:32.269 maybe a general chat model when you're handling 00:08:32.269 --> 00:08:35.009 documents. Sure. Notebook LM is specifically 00:08:35.009 --> 00:08:37.470 designed by Google to ground its responses only 00:08:37.470 --> 00:08:39.750 in the documents you provide it. It's built for 00:08:39.750 --> 00:08:42.470 source citation and research tasks. So that inherently 00:08:42.470 --> 00:08:44.269 gives it a much lower hallucination rate when 00:08:44.269 --> 00:08:46.590 you're dealing with uploaded PDFs or transcripts 00:08:46.590 --> 00:08:48.509 compared to more general models. It tends to 00:08:48.509 --> 00:08:51.429 stick to the facts presented. Got it. So for 00:08:51.429 --> 00:08:54.110 this step, it involves some serious cross -referencing. 00:08:54.330 --> 00:08:57.590 The task in the example was comparing Nvidia's, 00:08:57.590 --> 00:09:00.850 NVDA's earnings call transcript, which is often 00:09:00.850 --> 00:09:03.490 full of, you know, spoken optimism and forward 00:09:03.490 --> 00:09:05.909 -looking statements. Comparing that against the 00:09:05.909 --> 00:09:08.590 static official 10k report, the legal document 00:09:08.590 --> 00:09:11.360 that details all the documented risks. A very 00:09:11.360 --> 00:09:14.259 different tone, usually. Exactly. And the investigative 00:09:14.259 --> 00:09:16.419 prompt for Notebook LM broke down into three 00:09:16.419 --> 00:09:19.519 really powerful parts. First, optimism versus 00:09:19.519 --> 00:09:23.019 reported risks. So compare the CEO's talk of, 00:09:23.019 --> 00:09:25.919 say, unprecedented demand directly against specific 00:09:25.919 --> 00:09:28.740 enumerated risks listed in the 10K, things like 00:09:28.740 --> 00:09:31.220 customer concentration or known demand fluctuation 00:09:31.220 --> 00:09:34.419 risks. OK. Second, supply chain nuance. Cross 00:09:34.419 --> 00:09:36.080 -check those general mentions of supply chain 00:09:36.080 --> 00:09:38.139 in the upbeat call against the really detailed 00:09:38.139 --> 00:09:41.139 dependencies, specific suppliers, or maybe geopolitical 00:09:41.139 --> 00:09:44.039 risks that are listed way down in the 10K's disclosures. 00:09:44.620 --> 00:09:47.480 And third, the competitive moat. Compare the 00:09:47.480 --> 00:09:50.059 CEO perhaps claiming a stronger than ever competitive 00:09:50.059 --> 00:09:52.000 moat against the explicit competitors and the 00:09:52.000 --> 00:09:54.460 stated strategy you find in the 10K's official 00:09:54.460 --> 00:09:56.720 competition section. They have to list that stuff. 00:09:56.990 --> 00:10:00.389 Wow. So this method really turns the AI into 00:10:00.389 --> 00:10:03.009 an investigator, doesn't it? Uncovering potentially 00:10:03.009 --> 00:10:05.129 critical differences between the public relations 00:10:05.129 --> 00:10:07.830 narrative for management and the legally required 00:10:07.830 --> 00:10:11.990 filings. It does. So why is that cross -referencing, 00:10:12.289 --> 00:10:15.009 specifically comparing the official 10K against 00:10:15.009 --> 00:10:18.330 the often more upbeat earnings call, why is that 00:10:18.330 --> 00:10:20.950 so essential for proper investor due diligence? 00:10:21.519 --> 00:10:24.000 Well, it helps you identify those potentially 00:10:24.000 --> 00:10:26.480 unaddressed risks. Things that could seriously 00:10:26.480 --> 00:10:28.860 challenge the rosy narrative management might 00:10:28.860 --> 00:10:31.639 be painting. It's about finding the gaps. Okay, 00:10:31.659 --> 00:10:33.279 so we've covered analysis and investigation. 00:10:33.419 --> 00:10:35.159 Now let's talk about automating the tracking 00:10:35.159 --> 00:10:38.360 itself. That's step five, building a live dashboard. 00:10:38.649 --> 00:10:41.029 We can actually use AI to write Google Apps script 00:10:41.029 --> 00:10:43.450 code and build an auto -updating dashboard right 00:10:43.450 --> 00:10:45.950 there in Google Sheets. Yeah, exactly. We asked 00:10:45.950 --> 00:10:48.350 the AI in the example to write a specific function 00:10:48.350 --> 00:10:50.529 called update financials. And this function, 00:10:50.529 --> 00:10:52.429 it basically iterates through a list of stock 00:10:52.429 --> 00:10:54.570 tickers that you provide in your sheet. And it 00:10:54.570 --> 00:10:56.870 automatically populates four key data points 00:10:56.870 --> 00:10:58.769 using the built -in Google Finance function, 00:10:59.169 --> 00:11:01.669 things like current price, P -E ratio, the 52 00:11:01.669 --> 00:11:04.220 -week high, and market cap. That's fantastic. 00:11:04.379 --> 00:11:06.559 Just pure automation, saving you that tedious 00:11:06.559 --> 00:11:09.899 manual data check every single morning. And this 00:11:09.899 --> 00:11:13.139 brings us back nicely to step six, these advanced 00:11:13.139 --> 00:11:15.759 techniques, which involves using different AI 00:11:15.759 --> 00:11:17.919 models for their respective strengths. Yeah, 00:11:17.940 --> 00:11:20.340 it's really essential to choose the right tool 00:11:20.340 --> 00:11:23.029 for the right job. You don't use a hammer for 00:11:23.029 --> 00:11:25.750 a screw, right? Use Notebook LM, as we said, 00:11:25.950 --> 00:11:28.429 for summarizing and pulling exact quotes from 00:11:28.429 --> 00:11:30.730 your uploaded documents because of that grounding 00:11:30.730 --> 00:11:33.669 strength, that low hallucination rate. Then maybe 00:11:33.669 --> 00:11:37.110 use a model like ChatGPT for more creative brainstorming, 00:11:37.230 --> 00:11:40.049 exploring ideas, asking what if scenarios where 00:11:40.049 --> 00:11:42.250 it's broader knowledge base is actually helpful. 00:11:42.409 --> 00:11:44.269 Right. And as we mentioned earlier, you might 00:11:44.269 --> 00:11:46.370 lean towards something like Claude for writing 00:11:46.370 --> 00:11:49.289 and especially debugging complex code, like that 00:11:49.289 --> 00:11:51.929 Pinescript example, where it's stra - than handling 00:11:51.929 --> 00:11:55.970 longer context windows really pays off. And all 00:11:55.970 --> 00:11:58.350 of this just reinforces that fundamental truth, 00:11:58.490 --> 00:12:01.470 our golden rule, which is step seven. Always 00:12:01.470 --> 00:12:03.809 be specific and break down the problem. Break 00:12:03.809 --> 00:12:07.110 it into sequential specific tasks for the AI. 00:12:07.789 --> 00:12:10.370 Avoid asking one huge general question and just 00:12:10.370 --> 00:12:12.610 hoping for the best. That's how you end up with 00:12:12.610 --> 00:12:15.649 generic unhelpful answers. Specificity is key. 00:12:15.820 --> 00:12:18.039 Okay, so beyond just the raw speed increase, 00:12:18.700 --> 00:12:20.720 how does automating all this repetitive data 00:12:20.720 --> 00:12:23.440 collection, how does that really impact an investor's 00:12:23.440 --> 00:12:25.659 allocation of their valuable time and focus? 00:12:25.840 --> 00:12:28.200 It frees up so much mental energy, allowing you 00:12:28.200 --> 00:12:30.440 to focus almost solely on the high level deep 00:12:30.440 --> 00:12:32.620 analysis and crucially the application of your 00:12:32.620 --> 00:12:36.789 own judgment. Less drudgery, more thinking. OK, 00:12:37.029 --> 00:12:38.649 let's try and wrap up with some really practical 00:12:38.649 --> 00:12:41.090 guidance here. We absolutely must emphasize two 00:12:41.090 --> 00:12:43.450 key mistakes that even experienced investors 00:12:43.450 --> 00:12:45.289 can make when they start adopting these powerful 00:12:45.289 --> 00:12:48.450 tools. First, and this is huge, do not ask AI 00:12:48.450 --> 00:12:50.629 for direct investment advice. It simply lacks 00:12:50.629 --> 00:12:52.509 your personal financial context, your goals, 00:12:52.710 --> 00:12:54.970 your risk tolerance. It doesn't know you. Right. 00:12:55.179 --> 00:12:57.379 And second, as we've kept stressing throughout 00:12:57.379 --> 00:12:59.720 this, don't trust the numbers 100%, especially 00:12:59.720 --> 00:13:02.799 initially. Always, always double -check critical 00:13:02.799 --> 00:13:05.279 financial data points like revenue or profit 00:13:05.279 --> 00:13:08.360 or margins against that verified page number 00:13:08.360 --> 00:13:11.129 in the original report. You have to treat the 00:13:11.129 --> 00:13:15.070 AI output as a really powerful, synthesized first 00:13:15.070 --> 00:13:18.309 draft, not the final word. Exactly. What you 00:13:18.309 --> 00:13:20.850 should be doing, though, is using AI to radically 00:13:20.850 --> 00:13:22.789 expand your knowledge base. That's where the 00:13:22.789 --> 00:13:24.750 power lies. Use it to speed up your research 00:13:24.750 --> 00:13:26.809 dramatically, to quickly find correlations you 00:13:26.809 --> 00:13:30.090 might have missed, and to explore dozens, maybe 00:13:30.090 --> 00:13:32.990 hundreds, of companies that you could never humanly 00:13:32.990 --> 00:13:35.529 manage to analyze on your own timeline. It's 00:13:35.529 --> 00:13:38.600 about breadth and speed. Ultimately, AI just 00:13:38.600 --> 00:13:40.980 can't replace your deep, nuanced knowledge of 00:13:40.980 --> 00:13:43.080 a business, its leadership team, its culture, 00:13:43.519 --> 00:13:45.460 or its core competitive advantage in the market. 00:13:45.899 --> 00:13:48.259 Your informed judgment, based on all your experience 00:13:48.259 --> 00:13:51.000 in that AI -assisted research, that remains the 00:13:51.000 --> 00:13:53.179 single most important part of making a truly 00:13:53.179 --> 00:13:55.679 smart investment decision. The big idea here 00:13:55.679 --> 00:13:58.659 is pretty clear, I think. The future of AI -assisted 00:13:58.659 --> 00:14:01.179 investing, it's not about replacing human judgment 00:14:01.179 --> 00:14:03.980 at all. It's about dramatically improving human 00:14:03.980 --> 00:14:06.899 capability. Success is going to belong to those 00:14:06.899 --> 00:14:09.639 investors who learn how to ask the right, really 00:14:09.639 --> 00:14:12.519 structured questions and then rigorously verify 00:14:12.519 --> 00:14:15.019 the results the AI gives back. So here's maybe 00:14:15.019 --> 00:14:17.860 a provocative thought to leave you with. If AI 00:14:17.860 --> 00:14:21.320 can eventually read every single 10k and write 00:14:21.320 --> 00:14:23.879 every conceivable custom indicator basically 00:14:23.879 --> 00:14:27.240 for free, What unique, perhaps non -technical 00:14:27.240 --> 00:14:29.679 edge will the really successful investor need 00:14:29.679 --> 00:14:32.559 to rely on next year or the year after? Something 00:14:32.559 --> 00:14:34.779 to think about. We really encourage you to start 00:14:34.779 --> 00:14:36.899 experimenting with these tools, asking those 00:14:36.899 --> 00:14:39.080 specific questions, and always applying that 00:14:39.080 --> 00:14:41.360 critical human judgment on top. Thank you for 00:14:41.360 --> 00:14:43.139 sharing your sources with us for this deep dive 00:14:43.139 --> 00:14:43.460 today.