WEBVTT 00:00:00.040 --> 00:00:04.940 70 % of Y Combinator's top 100 companies, I mean 00:00:04.940 --> 00:00:07.780 the absolute giants, they began with organic 00:00:07.780 --> 00:00:10.619 ideas. Right. Problems the founders were personally 00:00:10.619 --> 00:00:12.380 living through. That's an overwhelming number 00:00:12.380 --> 00:00:14.939 for just accidental discovery. It is, but that 00:00:14.939 --> 00:00:18.480 leaves a really critical 30 % that were manufactured. 00:00:18.899 --> 00:00:21.339 These were concepts born from deliberate systematic 00:00:21.339 --> 00:00:24.359 searching. They engineered it. So can you really 00:00:24.359 --> 00:00:27.789 force? a billion -dollar concept. Can you just 00:00:27.789 --> 00:00:30.649 sit down at a blank whiteboard and almost mechanically 00:00:30.649 --> 00:00:33.170 generate an idea that changes things? Welcome 00:00:33.170 --> 00:00:34.750 to the Deep Dive, where we're going to tackle 00:00:34.750 --> 00:00:37.950 exactly that. Today we are equipping you, the 00:00:37.950 --> 00:00:40.390 learner, with the second half of the YC Idea 00:00:40.390 --> 00:00:43.070 Generation Framework. Our mission is pretty simple. 00:00:43.469 --> 00:00:45.909 Give you seven proven recipes for when you're 00:00:45.909 --> 00:00:48.210 facing that blank page. So we're moving beyond 00:00:48.210 --> 00:00:50.299 just... waiting for inspiration. We're going 00:00:50.299 --> 00:00:52.100 to look at leveraging your unique strengths, 00:00:52.340 --> 00:00:55.579 how to spot opportunities others miss, and the 00:00:55.579 --> 00:00:57.679 meta strategies for positioning yourself. Exactly. 00:00:57.700 --> 00:00:59.700 So that when those organic ideas do show up, 00:00:59.700 --> 00:01:01.359 you're actually ready to pounce. Let's get into 00:01:01.359 --> 00:01:04.219 it. The distinction between organic and deliberate 00:01:04.219 --> 00:01:08.180 is just, it's so crucial. That 70 % statistic 00:01:08.180 --> 00:01:11.040 is powerful, but it kind of makes success feel 00:01:11.040 --> 00:01:13.000 like a lottery, you know? That you just have 00:01:13.000 --> 00:01:16.180 to stumble into a problem. But the sources confirm. 00:01:17.000 --> 00:01:19.439 That systematic searching, that accounts for 00:01:19.439 --> 00:01:22.500 that other 30%. That's where we can get an edge. 00:01:22.659 --> 00:01:24.819 And if we're talking deliberate searching, then 00:01:24.819 --> 00:01:28.060 we have to talk about schlep blindness. Oh, absolutely. 00:01:28.400 --> 00:01:30.879 This is such a critical insight for that 30%. 00:01:30.879 --> 00:01:34.060 Great opportunities are so often ignored, not 00:01:34.060 --> 00:01:36.640 because they're impossible, but because the solutions 00:01:36.640 --> 00:01:41.069 require, well, tedious, unglamorous work. Tedious 00:01:41.069 --> 00:01:44.150 work like navigating complex regulations or figuring 00:01:44.150 --> 00:01:47.549 out archaic logistics. People have schlep blindness 00:01:47.549 --> 00:01:49.390 because they just don't want to get their hands 00:01:49.390 --> 00:01:51.849 dirty. Exactly. It might involve manual processing 00:01:51.849 --> 00:01:54.469 or coordinating physical stuff or dealing with 00:01:54.469 --> 00:01:56.590 boring regulations. It doesn't sound like a cool 00:01:56.590 --> 00:01:59.109 startup pitch. But that's precisely why it's 00:01:59.109 --> 00:02:01.730 so ripe for disruption. And the big mistake founders 00:02:01.730 --> 00:02:03.629 make is just assuming that wishing something 00:02:03.629 --> 00:02:07.260 existed is enough. That kind of thinking leads 00:02:07.260 --> 00:02:09.659 straight into a tar pit idea, right? Totally. 00:02:09.860 --> 00:02:13.580 You fail to investigate why every other attempt 00:02:13.580 --> 00:02:17.280 to solve that problem has failed. Spectacular. 00:02:17.300 --> 00:02:19.639 You just run straight at it without understanding 00:02:19.639 --> 00:02:22.199 the history or the structural reasons. It's still 00:02:22.199 --> 00:02:23.960 a problem. It's the difference between saying 00:02:23.960 --> 00:02:26.780 this should be easier and actually asking why 00:02:26.780 --> 00:02:30.479 it's so hard. So if the best ideas often come 00:02:30.479 --> 00:02:33.550 from the schlep. from boring, intense labor, 00:02:33.750 --> 00:02:36.250 how do we find those opportunities that other 00:02:36.250 --> 00:02:38.270 people are just missing? We have to strategically 00:02:38.270 --> 00:02:40.810 look where others are just fundamentally unwilling 00:02:40.810 --> 00:02:43.449 to go. Okay, so let's start that systematic journey 00:02:43.449 --> 00:02:46.250 by looking inward. The first recipe is foundational. 00:02:46.610 --> 00:02:49.150 Start with your unfair advantage. This is the 00:02:49.150 --> 00:02:51.949 founder market fit recipe baked right in. What 00:02:51.949 --> 00:02:56.150 specific expertise or relationships or non -obvious 00:02:56.150 --> 00:02:58.289 insights does your team have that competitors 00:02:58.289 --> 00:03:00.659 just couldn't replicate overnight? This takes 00:03:00.659 --> 00:03:02.759 some real self -reflection. I mean, think about 00:03:02.759 --> 00:03:05.719 Rezzy, the real estate fintech company. Their 00:03:05.719 --> 00:03:07.939 founders didn't just have one specialty. They 00:03:07.939 --> 00:03:10.520 had deep backgrounds in both real estate and 00:03:10.520 --> 00:03:14.000 financial tech. That overlap is key. It is. It 00:03:14.000 --> 00:03:17.159 let them see these deep structural problems that 00:03:17.159 --> 00:03:19.580 the average Silicon Valley outsider couldn't 00:03:19.580 --> 00:03:22.680 even perceive, let alone solve. They started 00:03:22.680 --> 00:03:25.560 with their own asymmetrical knowledge. The practical 00:03:25.560 --> 00:03:28.099 exercise here is so important. List your previous 00:03:28.099 --> 00:03:30.759 jobs, your internships, all of it. For each one, 00:03:30.879 --> 00:03:33.900 you have to ask, what problem did I see that 00:03:33.900 --> 00:03:36.520 no one on the outside could possibly see? And 00:03:36.520 --> 00:03:38.520 that leads perfectly into the second recipe, 00:03:38.699 --> 00:03:40.479 which is kind of an evolution of the classic 00:03:40.479 --> 00:03:45.110 advice. And this is critical. Scratch it while 00:03:45.110 --> 00:03:47.789 you are in an unusual position. Right. Don't 00:03:47.789 --> 00:03:50.349 scratch a generic itch that every college student 00:03:50.349 --> 00:03:52.949 has. The market is just saturated with those 00:03:52.949 --> 00:03:54.990 attempts. The qualification is everything. It's 00:03:54.990 --> 00:03:57.389 not about wishing college apps were easier. It 00:03:57.389 --> 00:04:00.849 needs to be unique. Look at Vetcov. The founder's 00:04:00.849 --> 00:04:04.430 father was a veterinarian. That gave them a window 00:04:04.430 --> 00:04:07.409 into this messy world that tech founders almost 00:04:07.409 --> 00:04:10.810 never see. They saw that vet practices ordered 00:04:10.810 --> 00:04:13.789 supplies with, like, phone calls and paper catalogs. 00:04:13.789 --> 00:04:16.670 Wow. Yeah. And because there was no modern e 00:04:16.670 --> 00:04:19.589 -commerce platform, they were constantly overpaying 00:04:19.589 --> 00:04:22.069 and wasting time. Silicon Valley founders are 00:04:22.069 --> 00:04:24.009 not looking at veterinary supply chains. It's 00:04:24.009 --> 00:04:26.449 the perfect example, an unusual position giving 00:04:26.449 --> 00:04:28.990 you insight into a market that is just desperate 00:04:28.990 --> 00:04:31.949 for help, a market with maximum schlep. Exactly. 00:04:32.149 --> 00:04:34.370 So we've seen how internal insight can spark 00:04:34.370 --> 00:04:37.089 an idea. But how do we navigate that line between 00:04:37.089 --> 00:04:40.129 a genuinely unique itch and, you know, falling 00:04:40.129 --> 00:04:43.170 headfirst into a dangerous, overtried tar pit? 00:04:43.310 --> 00:04:45.319 That's the challenge, right? A problem existing 00:04:45.319 --> 00:04:48.000 doesn't prove it's a good business. It just proves 00:04:48.000 --> 00:04:50.180 the problem is hard. And you need to perform 00:04:50.180 --> 00:04:52.519 what the sources call the check. Okay. Which 00:04:52.519 --> 00:04:54.439 brings us to the most common and I'd argue the 00:04:54.439 --> 00:04:57.139 most dangerous path. The third recipe is the 00:04:57.139 --> 00:04:59.220 most human, but we have to approach it with maximum 00:04:59.220 --> 00:05:01.779 caution. Things you wish existed. DoorDash is 00:05:01.779 --> 00:05:05.060 the great example here. The undeniable one. Stanford 00:05:05.060 --> 00:05:07.439 students wanted better local delivery and they 00:05:07.439 --> 00:05:10.360 built it. And it worked because they were willing 00:05:10.360 --> 00:05:12.589 to do the schlep. Going door to door at first, 00:05:12.730 --> 00:05:15.509 but for every DoorDash, there are, what, hundreds 00:05:15.509 --> 00:05:18.009 of ideas that just become ghost town apps because 00:05:18.009 --> 00:05:21.050 the founder didn't do that diligence? Precisely. 00:05:21.170 --> 00:05:24.629 You have to conduct the check. If you wish something 00:05:24.629 --> 00:05:27.470 existed and it doesn't, you must relentlessly 00:05:27.470 --> 00:05:30.509 ask why. Did the market try it five years ago 00:05:30.509 --> 00:05:33.629 and fail? If so, what's your genuine quantifiable 00:05:33.629 --> 00:05:36.689 insight into why you will succeed now? It could 00:05:36.689 --> 00:05:38.850 be the tech wasn't ready or consumer behavior 00:05:38.850 --> 00:05:41.129 wasn't there yet. Right. But if you don't ask, 00:05:41.170 --> 00:05:43.990 you're just running on an assumption. And honestly, 00:05:44.149 --> 00:05:45.790 when I try to validate these kinds of wishes 00:05:45.790 --> 00:05:48.089 myself, I still wrestle with confirmation bias 00:05:48.089 --> 00:05:50.509 in that early phase. You know, I get pulled toward 00:05:50.509 --> 00:05:52.610 what I want to believe will work. It's a very 00:05:52.610 --> 00:05:55.509 human thing. That bias is so real. You have to 00:05:55.509 --> 00:05:57.829 fight it with facts, which is why Recipe 4 is 00:05:57.829 --> 00:06:00.170 so powerful. It moves away from personal wishes 00:06:00.170 --> 00:06:03.170 and focuses on the environment. Look for recent 00:06:03.170 --> 00:06:06.610 changes. Yes. Changes create these unique opportunity 00:06:06.610 --> 00:06:09.769 windows where a previously failed concept is 00:06:09.769 --> 00:06:12.410 suddenly viable. It's all about timing. Gather 00:06:12.410 --> 00:06:14.990 Town is the perfect case study. Totally. Pre 00:06:14.990 --> 00:06:17.550 -COVID, demand for their virtual hangout space 00:06:17.550 --> 00:06:21.160 was low. But the pandemic lockdowns created this 00:06:21.160 --> 00:06:24.120 massive immediate behavior change. Suddenly, 00:06:24.160 --> 00:06:26.759 the whole world needed a casual digital space. 00:06:27.000 --> 00:06:29.279 They took advantage of that shift. A world event 00:06:29.279 --> 00:06:31.990 opened the window. The systematic approach here 00:06:31.990 --> 00:06:35.610 is to map shifts in four areas. Technology, like 00:06:35.610 --> 00:06:39.329 new GPT models. Regulation, like GDPR. Behavior, 00:06:39.529 --> 00:06:43.329 like remote work. And world events, like supply 00:06:43.329 --> 00:06:46.329 chain disruptions. The sweet spot is where that 00:06:46.329 --> 00:06:48.649 change makes something newly necessary that was 00:06:48.649 --> 00:06:51.689 maybe just optional a year or two ago. Timing 00:06:51.689 --> 00:06:54.000 is everything. So if we spot a working model 00:06:54.000 --> 00:06:56.560 that's now viable because of a change, is it 00:06:56.560 --> 00:06:58.439 enough to just adapt it for a new geographic 00:06:58.439 --> 00:07:01.360 market? Is that a safe bet? Not entirely. Local 00:07:01.360 --> 00:07:03.740 pain points and regulations have to be rigorously 00:07:03.740 --> 00:07:06.680 validated first. Okay, so recipe five leans into 00:07:06.680 --> 00:07:09.360 that idea of adaptation. Find successful companies 00:07:09.360 --> 00:07:12.339 and create variants. Right. The idea is you take 00:07:12.339 --> 00:07:14.620 a proven business model from market A and you 00:07:14.620 --> 00:07:16.579 just execute it exceptionally well in market 00:07:16.579 --> 00:07:19.300 B. You're focusing 90 % on execution, 10 % on 00:07:19.300 --> 00:07:21.970 proving the concept. NuvoCargo is a great example. 00:07:22.269 --> 00:07:24.629 They looked at Flexport's model for digital freight 00:07:24.629 --> 00:07:27.189 forwarding and applied it specifically to Latin 00:07:27.189 --> 00:07:29.910 America. Exactly. Flexport already de -risked 00:07:29.910 --> 00:07:32.790 the underlying model. NuvoCargo's job was to 00:07:32.790 --> 00:07:35.529 focus on execution excellence in a new, really 00:07:35.529 --> 00:07:38.170 difficult market. But isn't cloning a model risky? 00:07:38.370 --> 00:07:41.709 How do you avoid just looking like a cheap knockoff? 00:07:41.949 --> 00:07:44.410 The key is true variance. It has to be more than 00:07:44.410 --> 00:07:47.149 just a language translation. Good variants are... 00:07:47.670 --> 00:07:51.149 True geographic transfers or rebuilding a horizontal 00:07:51.149 --> 00:07:54.189 product to be deeply industry specific like Viva 00:07:54.189 --> 00:07:56.589 did for Pharma. Got it. You can also vary the 00:07:56.589 --> 00:07:58.509 customer size or even the business model itself. 00:07:59.209 --> 00:08:02.250 Which brings us to recipe six, which is maybe 00:08:02.250 --> 00:08:04.949 the most systematic hunting method of all. Talk 00:08:04.949 --> 00:08:07.689 to people. through structured interviews. This 00:08:07.689 --> 00:08:10.810 is how you deliberately find hidden, messy pain 00:08:10.810 --> 00:08:13.269 points. It requires months of what sounds like 00:08:13.269 --> 00:08:15.949 agonizingly boring legwork, but it must pay off 00:08:15.949 --> 00:08:19.360 with clear demand validation. It does. Look at 00:08:19.360 --> 00:08:21.519 Adobe, which built a modern fuel card for the 00:08:21.519 --> 00:08:23.560 trucking industry. The founders didn't start 00:08:23.560 --> 00:08:25.300 with a solution. They went to truck stops for 00:08:25.300 --> 00:08:28.680 months, interviewed 20 to 30 drivers, and just 00:08:28.680 --> 00:08:31.639 meticulously mapped out their pain points with 00:08:31.639 --> 00:08:34.460 the existing messy systems. So the boring work 00:08:34.460 --> 00:08:36.940 generated a tested idea before they even wrote 00:08:36.940 --> 00:08:40.059 a line of code. Exactly. The procedure is simple 00:08:40.059 --> 00:08:43.860 but hard. Pick a space, find 20 to 30 people, 00:08:43.919 --> 00:08:46.559 and just ask open -ended questions about their 00:08:46.559 --> 00:08:49.620 frustrations. Don't pitch anything. Just listen 00:08:49.620 --> 00:08:51.860 to the pain. That structured listening is how 00:08:51.860 --> 00:08:54.100 you find the schlep blindness, the stuff that's 00:08:54.100 --> 00:08:55.899 so tedious people just assume it's part of the 00:08:55.899 --> 00:08:58.399 job. Right. So what if we want to tackle something 00:08:58.399 --> 00:09:00.639 really massive like a foundational industry, 00:09:00.919 --> 00:09:03.100 but we don't have that deep domain knowledge 00:09:03.100 --> 00:09:05.879 right away? Where do we start? Go straight to 00:09:05.879 --> 00:09:08.580 the largest, most obviously broken sectors. Okay, 00:09:08.639 --> 00:09:11.620 recipe seven. Look for large, broken industries. 00:09:11.759 --> 00:09:14.679 We're talking multi -trillion dollar sectors 00:09:14.679 --> 00:09:17.220 that everyone complains about. Healthcare, real 00:09:17.220 --> 00:09:19.659 estate, legal services. They're just begging 00:09:19.659 --> 00:09:21.679 for disruption, often because they're running 00:09:21.679 --> 00:09:25.019 on 20th century tech. Flexport and freight forwarding. 00:09:25.259 --> 00:09:28.220 LegalZoom. These are the classic examples of 00:09:28.220 --> 00:09:31.360 attacking those giants. But the catch is... These 00:09:31.360 --> 00:09:33.519 industries are broken for a reason. They are. 00:09:33.659 --> 00:09:35.840 Flexport had to deal with customs regulations 00:09:35.840 --> 00:09:38.899 and systems built decades ago. There are structural 00:09:38.899 --> 00:09:42.259 obstacles regulation, entrenched interests. You 00:09:42.259 --> 00:09:45.240 need insight into how you'll overcome those specific 00:09:45.240 --> 00:09:47.960 hurdles, not just build a better landing page. 00:09:48.360 --> 00:09:50.740 The source also gave us a bonus recipe, which 00:09:50.740 --> 00:09:53.500 feels very practical. Find a co -founder who 00:09:53.500 --> 00:09:56.100 already has the idea or the domain expertise. 00:09:56.340 --> 00:09:58.539 Yeah, if you lack the regulatory knowledge for 00:09:58.539 --> 00:10:01.299 Recipe 7, find the partner who has lived that 00:10:01.299 --> 00:10:04.669 pain. It bridges the skill gap and validates 00:10:04.669 --> 00:10:06.809 the problem in one move. It's founder market 00:10:06.809 --> 00:10:09.190 fit and technical execution all at once. Okay, 00:10:09.289 --> 00:10:11.590 let's zoom out to the meta strategy. Because 00:10:11.590 --> 00:10:14.429 if 70 % of great ideas are organic, we need to 00:10:14.429 --> 00:10:16.629 maximize our odds of noticing them. And the number 00:10:16.629 --> 00:10:19.029 one rule is to become an expert on something 00:10:19.029 --> 00:10:21.409 valuable. You have to deliberately put yourself 00:10:21.409 --> 00:10:23.990 in a fertile environment, go work at a high growth 00:10:23.990 --> 00:10:26.690 startup, or in a domain right next to a changing 00:10:26.690 --> 00:10:29.570 technology. We see this over and over. Stripe 00:10:29.570 --> 00:10:31.250 came from the Collison brothers' frustration 00:10:31.250 --> 00:10:33.889 with payments for their previous projects. You 00:10:33.889 --> 00:10:36.690 can't notice massive structural problems in spaces 00:10:36.690 --> 00:10:39.509 you don't actually live in. The second meta strategy 00:10:39.509 --> 00:10:43.100 is deceptively simple. Build things. Side projects 00:10:43.100 --> 00:10:45.080 turn into companies all the time because you 00:10:45.080 --> 00:10:46.879 scratch an itch that turns out to be universal. 00:10:47.179 --> 00:10:49.379 Right. So many developer tools start as internal 00:10:49.379 --> 00:10:51.600 projects that get open source because founders 00:10:51.600 --> 00:10:54.940 realize, wait, other people need this too. Replica, 00:10:55.000 --> 00:10:57.539 the AI companion company, started as a personal 00:10:57.539 --> 00:11:00.840 project to build an AI version of a friend. Whoa. 00:11:01.620 --> 00:11:04.620 Imagine scaling a simple personal side project 00:11:04.620 --> 00:11:07.759 like that to field a billion queries a day. The 00:11:07.759 --> 00:11:09.799 compounding effect of just continuously building 00:11:09.799 --> 00:11:12.399 is undeniable. So we've covered seven recipes, 00:11:12.639 --> 00:11:15.159 two meta strategies. But after all that framework, 00:11:15.259 --> 00:11:17.799 we get to the final unavoidable rule. When in 00:11:17.799 --> 00:11:20.659 doubt, just launch. The market test is everything. 00:11:21.310 --> 00:11:24.649 Over -analysis is a huge opportunity cost, especially 00:11:24.649 --> 00:11:27.090 in fast -moving sectors. You do the diligence 00:11:27.090 --> 00:11:29.450 to avoid the obvious tar pits, but if there's 00:11:29.450 --> 00:11:31.649 still uncertainty, you have to commit to building 00:11:31.649 --> 00:11:34.149 something minimal and testing it with real users. 00:11:34.450 --> 00:11:36.870 Market feedback is just, it's so much higher 00:11:36.870 --> 00:11:39.590 quality than theory, always. Customers either 00:11:39.590 --> 00:11:42.490 pay or they don't. That signal is clear. And 00:11:42.490 --> 00:11:45.289 most successful startups pivot anyway. The key 00:11:45.289 --> 00:11:47.909 is starting in a good idea space, getting that 00:11:47.909 --> 00:11:50.700 feedback, and iterating quickly. The practical 00:11:50.700 --> 00:11:53.919 advice is firm, set a deadline, two weeks for 00:11:53.919 --> 00:11:56.120 diligence, then four weeks to build a minimal 00:11:56.120 --> 00:11:59.059 version. The meta lesson from YC's top companies 00:11:59.059 --> 00:12:01.620 is that the founders didn't have perfect, fully 00:12:01.620 --> 00:12:04.240 formed ideas. They positioned themselves in fertile 00:12:04.240 --> 00:12:06.879 spaces, they recognized real problems, and they 00:12:06.879 --> 00:12:08.960 acted immediately. The framework gets you 90 00:12:08.960 --> 00:12:11.840 % of the way there. That last 10 % is action. 00:12:12.039 --> 00:12:14.240 As one source concluded, and this is really the 00:12:14.240 --> 00:12:16.559 core takeaway, often the only way to know if 00:12:16.559 --> 00:12:18.759 a startup idea is good is to launch it and find 00:12:18.759 --> 00:12:21.009 out. So before you go, we'd encourage you to 00:12:21.009 --> 00:12:23.529 actually apply that unfair advantage exercise 00:12:23.529 --> 00:12:26.649 this week. Seriously, list your unique expertise, 00:12:26.850 --> 00:12:29.210 your jobs, the problems only you could have seen. 00:12:29.350 --> 00:12:31.610 What's your asymmetrical knowledge? And here's 00:12:31.610 --> 00:12:34.429 a final thought. If 70 % of great ideas start 00:12:34.429 --> 00:12:37.230 organically, how much time should we really spend 00:12:37.230 --> 00:12:39.990 on systematic brainstorming versus simply building 00:12:39.990 --> 00:12:42.590 deep, defensible expertise in a high -growth, 00:12:42.710 --> 00:12:45.870 fertile domain? The job is ultimately to notice 00:12:45.870 --> 00:12:48.429 the problem, not necessarily to invent it.