YouTube2Text

Reverse Engineer Anything Does Way More Than You Think — Transcript

by Austin Marchese · 3,881 words · 569 segments · language en · Watch on YouTube

Full transcript

  1. 0:00Reverse engineer anything just went
  2. 0:01viral and people are freaking out
  3. 0:03because it's a free tool that lets you
  4. 0:04rebuild almost any app in minutes. Now,
  5. 0:07I will show you how to use this, but
  6. 0:08that's not the point video. Instead, I'm
  7. 0:10going to give you a step-by-step
  8. 0:11playbook to reverse engineer anything
  9. 0:14including non-technical work. So first
  10. 0:16we'll cover what REA is and use it to
  11. 0:18rebuild an app. Then we'll apply the
  12. 0:20exact same playbook to reverse engineer
  13. 0:21four things, successful businesses,
  14. 0:23specific tools, tasks you want to
  15. 0:25automate, and strategic thinking from
  16. 0:27industry experts. I guarantee whether
  17. 0:29you're technical or not, this will
  18. 0:30change how you use AI every single day.
  19. 0:32So what REA is and how it works. So REA,
  20. 0:35short for reverse engineer anything, is
  21. 0:37an open source tool that anyone can use
  22. 0:39to rebuild a computer app, but it's
  23. 0:41actually nothing new. It's really just a
  24. 0:43combination of several pre-existing
  25. 0:45tools that engineers have used for
  26. 0:47years. So Ghidra, Hopper, and IDA, which
  27. 0:50take apart compiled software, Chrome Dev
  28. 0:52Tools, which show you what a website is
  29. 0:54doing, and app code reader, which walks
  30. 0:56through an app. And these aren't new,
  31. 0:57right? Some of them have actually been
  32. 0:58around since the 1970s. So what makes
  33. 1:01REA as unique is it's a harness. It lets
  34. 1:03any person talk to all of these tools at
  35. 1:05once without knowing how to use any of
  36. 1:07them. On screen you can see a diagram
  37. 1:09directly from REA's own documentation
  38. 1:11and it's essentially just a loop. First
  39. 1:13you give your AI agent a target, for
  40. 1:15example an app. Then REA inspects it and
  41. 1:18traces it with its analysis tools. Third
  42. 1:20the agent then takes that information
  43. 1:21and then starts building the app. It
  44. 1:23then keeps looping until the app is
  45. 1:24entirely recreated. So there is a bigger
  46. 1:26story here about what this means for the
  47. 1:28future of software and I'll get to that
  48. 1:30at the end of the video. But first, how
  49. 1:31do you actually use this thing? So first
  50. 1:33download REA by running this prompt.
  51. 1:35Then you can pick the app that you
  52. 1:36actually want to recreate. On screen I
  53. 1:38chose to reverse engineer Apple Notes
  54. 1:40and after about 40 minutes here is the
  55. 1:41result. On the left is Apple's version
  56. 1:43and on the right is the recreated one.
  57. 1:45They're pretty damn similar and it is
  58. 1:47awesome, but again that is not the point
  59. 1:49of this video. You're likely a business
  60. 1:51owner or a high-level operator watching
  61. 1:52this channel. How can you actually apply
  62. 1:54this to get value from it? Just ripping
  63. 1:56off a notes app is genuinely useless. So
  64. 2:00what you actually take from this? Well,
  65. 2:02REA has put a spotlight on something
  66. 2:04people have been doing forever, and
  67. 2:06that's figuring out exactly how
  68. 2:08something works, then reverse
  69. 2:09engineering it to rebuild it. And in
  70. 2:12every field, not just technical ones,
  71. 2:13people have been doing this. So for
  72. 2:15example, writers. They learn by studying
  73. 2:17authors they admire. They take apart the
  74. 2:19sentences, borrow techniques, and
  75. 2:21eventually they find out their own
  76. 2:22voice. Now, if you're not a writer,
  77. 2:24let's say in business, it's the same
  78. 2:26process. We borrow pieces of successful
  79. 2:28businesses and their processes to make
  80. 2:29them our own. So here is how you can
  81. 2:31actually do this yourself. REA works in
  82. 2:33five steps, and we're going to borrow
  83. 2:35them for the rest of the video. First,
  84. 2:37pick what you're studying. Second,
  85. 2:39gather the evidence. Third, design what
  86. 2:41you'll build from it. Fourth, rebuild
  87. 2:43and grade it against the original. Five,
  88. 2:45iterate yourself and make it yours.
  89. 2:47That's the process to learn anything
  90. 2:49from winners and make yourself more
  91. 2:50productive. So the first thing we're
  92. 2:52going to reverse engineer is businesses.
  93. 2:54There's a famous process Russell Brunson
  94. 2:55popularized called funnel hacking. You
  95. 2:58walk through a competitor's funnel,
  96. 2:59which is how they turn strangers into
  97. 3:00customers, and study the exact steps
  98. 3:02they use. And we're going to systematize
  99. 3:04that using our five-step REA framework
  100. 3:06so you can see what makes them so
  101. 3:08successful. So first, pick the business
  102. 3:10you want to analyze. And a general rule
  103. 3:12of thumb for whatever business you're
  104. 3:14analyzing, the bigger the player, the
  105. 3:16less you should study them. And that's
  106. 3:18because let's say you're looking at
  107. 3:19Nike, their brand props up whatever work
  108. 3:21they're doing, so you can't just redo
  109. 3:23what Nike's doing because you don't have
  110. 3:24that brand authority. So instead, pick
  111. 3:26the small competitor that just broke out
  112. 3:29and see what they're doing right. So use
  113. 3:30this prompt to analyze and identify your
  114. 3:32top competitors. And no, you don't need
  115. 3:34to run a business for this to be
  116. 3:35valuable. For example, if you're in
  117. 3:36marketing, see what your marketing
  118. 3:38competitors are doing. Second step,
  119. 3:40gather data you can leverage. Run this
  120. 3:42prompt to search through a competitor's
  121. 3:44page for info. And you'll find that some
  122. 3:46websites block automated tools, so you
  123. 3:48won't be able to see their subpages in
  124. 3:50the funnel. And so that's where you have
  125. 3:51a couple of options. One, you can have
  126. 3:52Claude walk through it in your browser.
  127. 3:54And the key here, whenever you do this,
  128. 3:56I would tell Claude to take screenshots
  129. 3:58of any key steps. In my eyes, think of
  130. 4:00these screenshots as your evidence pile.
  131. 4:02They remove the interpretation gap
  132. 4:04because you can see what Claude saw
  133. 4:06instead of trusting its summary. But at
  134. 4:08this point, you may still get blocked
  135. 4:10and this is very slow. So if you want a
  136. 4:11faster, more effective way to do it,
  137. 4:13then effectively look through each page
  138. 4:15and get through any blockers. I added a
  139. 4:17competitor research tool to
  140. 4:18buildpartner.ai, which you can use
  141. 4:20entirely for free. And you can run this
  142. 4:21single command and it'll go through the
  143. 4:23funnel and do a deep analysis. Step
  144. 4:25three, design. In this case, that means
  145. 4:27taking what you learned and applying it
  146. 4:29to your work. Here's a prompt that takes
  147. 4:31that information and gives you
  148. 4:32actionable takeaways for your business.
  149. 4:34Four is rebuild and grade. Take that
  150. 4:36list from the previous step and have
  151. 4:37Claude build what it can, then evaluate
  152. 4:39the final product. And finally, step
  153. 4:42five is iterate. See what's working, see
  154. 4:44what didn't, see what other people are
  155. 4:45doing over time. This is a continuous
  156. 4:47learning process. Now, as we go through
  157. 4:49each of these use cases, both REA as
  158. 4:51well as anything that we're going to
  159. 4:52cover, there are limitations. So neither
  160. 4:55this funnel hack nor REA can see the
  161. 4:57back end of how things work. So in the
  162. 4:58business example, it can't see the
  163. 5:00customer support, [music] the email
  164. 5:01follow-ups, etc. And for REA and the
  165. 5:03computer example, it can't see anything
  166. 5:05that it doesn't have direct access to.
  167. 5:06So for example, the back end or the
  168. 5:08database of the tool itself. But it
  169. 5:10doesn't have to be that extreme.
  170. 5:11Sometimes the most valuable thing is a
  171. 5:13single tool that you recreate. But
  172. 5:15before we get to number two, building
  173. 5:16tools, everything in this section was
  174. 5:18about studying how someone else turns
  175. 5:19strangers into customers. What it can't
  176. 5:21show you is where those strangers come
  177. 5:23from the first place, which brings us to
  178. 5:25today's video sponsor, Clay. So everyone
  179. 5:27on this channel is learning to build
  180. 5:28with AI, but nobody really talks about
  181. 5:29who you're actually building it for. And
  182. 5:31finding customers is still the hardest
  183. 5:33part. Now, there are two ways to
  184. 5:34actually get these customers. First, you
  185. 5:36can build an audience, which takes
  186. 5:37years, or go find the right people and
  187. 5:40reach out to them directly. And you can
  188. 5:41start doing that today. And that's what
  189. 5:43Clay is for and that's what I've used to
  190. 5:44scale my own business. So what Clay does
  191. 5:47is it finds the businesses and people
  192. 5:48that need what you sell and tells you
  193. 5:50how to reach them directly. So you
  194. 5:52describe the customer you want in plain
  195. 5:54English like find US companies with 10
  196. 5:56to 100 people hiring an AI automation
  197. 5:58engineer right now and get me the
  198. 6:00founder or head of operations at each
  199. 6:02one with their email. Clay will then
  200. 6:04build a list, finds the decision makers
  201. 6:06and researches each one so you know
  202. 6:08who's worth your time. A company that's
  203. 6:10hiring is telling you it's ready to buy,
  204. 6:12which makes your outreach feel smart,
  205. 6:13not spammy. And Clay's new go-to-market
  206. 6:15agent is about to launch and you can
  207. 6:17access it today directly through Clay's
  208. 6:19CLI in Claude. You tell it something
  209. 6:21like build me an outbound campaign that
  210. 6:23finds dental practices with five to 20
  211. 6:25employees and writes a personalized
  212. 6:27email to each office manager. Then the
  213. 6:29Clay CLI will build the whole thing for
  214. 6:31you. It builds the go-to-market
  215. 6:32workflows for you, the outbound
  216. 6:34campaigns, the scoring flows, the
  217. 6:36sourcing data, finding lists of leads.
  218. 6:38It can do it all, which is pretty wild.
  219. 6:39And the thing that powers all of this is
  220. 6:41that Clay connects to over 250 data
  221. 6:43providers. And by pairing that data from
  222. 6:45all these providers together, Clay has
  223. 6:47access to the best data in the world all
  224. 6:49in a single place. Now, on my channel, I
  225. 6:51talk a lot about problems you should and
  226. 6:53shouldn't solve yourself. And setting up
  227. 6:55250 data contracts is one that you
  228. 6:57should not solve. So Clay is free to
  229. 6:59start and no credit card is needed. And
  230. 7:01if you sign up through the link in the
  231. 7:02description, my community gets $250
  232. 7:05worth of free credits to get started,
  233. 7:07which you can get a lot done with. Some
  234. 7:08problems need one-time solutions and
  235. 7:10some problems are recurring. Finding new
  236. 7:12customers is a recurring problem and
  237. 7:14Clay provides a recurring solution. So
  238. 7:16that's about finding customers. The
  239. 7:17second thing that we're going to reverse
  240. 7:18engineer is about building tools, not
  241. 7:21apps. On this channel, I've stressed the
  242. 7:22importance of building for yourself, not
  243. 7:25for others. The reason is that if you
  244. 7:27build something that you will use, no
  245. 7:29matter what, you get value from it. And
  246. 7:31for me, this is actually where the whole
  247. 7:32buildpartner.ai roadmap comes from. I
  248. 7:34figure out what I need and I build it
  249. 7:36for me. And then it usually works for
  250. 7:38people who also want to get the most out
  251. 7:40of Claude. And so when you're building
  252. 7:41for yourself, the key here is also that
  253. 7:43you don't want to try and do everything
  254. 7:45at once. Every app or product that you
  255. 7:47see is really just a bundle of features
  256. 7:48and they're designed for the masses. So
  257. 7:50some features may be valuable to you,
  258. 7:52some features aren't. So when you're
  259. 7:53reverse engineering existing apps, I
  260. 7:56would avoid rebuilding entire things,
  261. 7:58but instead focus on rebuilding specific
  262. 8:01features or tools. And before you
  263. 8:02actually go ahead and rebuild anything,
  264. 8:04you have to make sure you understand why
  265. 8:06you're actually rebuilding it. It can be
  266. 8:08very tempting to fall into a dopamine
  267. 8:10building loop where it feels productive
  268. 8:12to recreate things, but in reality, it's
  269. 8:14not actually a needle mover. So if
  270. 8:15you're just rebuilding something to
  271. 8:17rebuild it, don't do it. There has to be
  272. 8:19a concrete enhancement that you want to
  273. 8:21do that it can't currently do, and we'll
  274. 8:23cover that a bit more in a sec. But once
  275. 8:25you've got a tool in mind, run it
  276. 8:27through our reverse engineering
  277. 8:29framework. First, pick one feature, not
  278. 8:31the entire app. Here is a prompt that
  279. 8:32can actually help you do this. So REA's
  280. 8:34own example does exactly this. It
  281. 8:36targets a feature, not the whole app.
  282. 8:38Their read me, their documentation says,
  283. 8:40understand how search works in the notes
  284. 8:42app. Show me the evidence and build a
  285. 8:43similar feature for my project. And if
  286. 8:45REA feels like too much, just use
  287. 8:47Claude. Second step, gather information
  288. 8:49about the feature. This is where I love
  289. 8:50either screenshots or a recording of
  290. 8:52using the feature. I've taken thousands
  291. 8:54of screenshots of features that I like
  292. 8:55to inform product decisions, and this is
  293. 8:57extremely effective. The third, design
  294. 8:59and focus on creating a minimum viable
  295. 9:01architecture. Depending on how you want
  296. 9:03to recreate the feature, you do not want
  297. 9:05to overcomplicate it. If it's one
  298. 9:07feature that stands on its own, have
  299. 9:09Claude make it as an artifact. Don't
  300. 9:10spin up a whole app and worry about
  301. 9:12hosting and all the other stuff. Just
  302. 9:14let Claude do it for you in its artifact
  303. 9:16templates. The simplest solution is the
  304. 9:18best solution. At this point, you should
  305. 9:20be ready for step four, which is rebuild
  306. 9:22the feature. Hand over the evidence and
  307. 9:24the design and let Claude rip or use REA
  308. 9:26to rebuild it. Five, iterate from there
  309. 9:28and make it yours. Back to understanding
  310. 9:30why you'd be recreating a feature. In
  311. 9:32this reverse engineering scenario, there
  312. 9:34has to be some enhancement you're
  313. 9:36adding, otherwise it's useless. So this
  314. 9:38step is probably the most important. Now
  315. 9:40building a tool is great, but the
  316. 9:41biggest time savings can come from
  317. 9:43something even simpler, and that's
  318. 9:45getting Claude to do a task for you. The
  319. 9:47third category to reverse engineer,
  320. 9:48automate tasks. And we're going to
  321. 9:50record the best version and then hand it
  322. 9:52off. The goal here is to reverse
  323. 9:53engineer how you complete a task. Then
  324. 9:56take that and systematize Claude to do
  325. 9:58it for you. Here we won't use REA in
  326. 10:00this step, but we're going to reuse the
  327. 10:01framework. And so the simplest way to do
  328. 10:03this is actually just ask Claude to
  329. 10:05review your past conversation history
  330. 10:07and flag any Claude skills you should
  331. 10:09create to systematize what you're doing.
  332. 10:11This will go through your conversation
  333. 10:12history and reverse engineer Claude
  334. 10:15skills based on your past activities.
  335. 10:17And if you run that, you'll have
  336. 10:18repeatable Claude skills that you can
  337. 10:20use for a task going forward. And you'll
  338. 10:21see in the prompt itself an optimization
  339. 10:23trick. When you do mine your chats, stop
  340. 10:25asking what did I use AI for, instead
  341. 10:28ask what did I have to send back. Every
  342. 10:30time you send something back, that's a
  343. 10:31step worth making into a skill, but if
  344. 10:33it got it right the first time, then you
  345. 10:34probably don't need a skill for it. But
  346. 10:36that is just level one of doing this.
  347. 10:38Level two is a lot more interesting. My
  348. 10:40favorite way to reverse engineer a task
  349. 10:42is to record myself doing it. Within
  350. 10:44Claude, you can record your screen and
  351. 10:46Claude will identify ways to automate
  352. 10:48what you're doing on the recording. But
  353. 10:50just clicking through the screens is
  354. 10:51only part of the process. You have to
  355. 10:53make sure to narrate while you work to
  356. 10:56give Claude additional insight into your
  357. 10:57thought process. That combines the
  358. 10:59visual component with the audio
  359. 11:01component. Claude will see what you do
  360. 11:02and then hear what you say. And as you
  361. 11:04do that, make sure to mention what to do
  362. 11:07and why. You want to explain it like
  363. 11:08you're teaching a human on an onboarding
  364. 11:10call. You would explain the logic behind
  365. 11:12your decisions. Then Claude will take
  366. 11:13all this data and it's trained to
  367. 11:15suggest programmatic ways to do the
  368. 11:16task. It won't just think about brute
  369. 11:18forcing it, it'll actually think about
  370. 11:20things like using APIs or MCP
  371. 11:22connections to automate this process.
  372. 11:24Then if you want to take it to the next
  373. 11:26level, you can record a rubric, not just
  374. 11:29the task itself. So you can record
  375. 11:30yourself reviewing the finished outputs,
  376. 11:32saying what's good, what's wrong, and
  377. 11:34why. And then Claude can take that to
  378. 11:36iterate and improve the skill over time.
  379. 11:38This is your way of teaching Claude the
  380. 11:40judgment, not just the clicks. So bring
  381. 11:42all of this through our framework. You
  382. 11:43want to first pick the task you do most,
  383. 11:45second gather the info through the
  384. 11:47recording process, third lean on Claude
  385. 11:49to design the automation or rebuild it
  386. 11:51and then grade it by running it
  387. 11:52yourself, and then five iterate until it
  388. 11:54matches how you do it and provide
  389. 11:56feedback directly. At this point, we
  390. 11:58focused on getting AI to automate tasks,
  391. 12:00build out specific features, and study
  392. 12:01successful businesses. But the biggest
  393. 12:03opportunity is to reverse engineer how
  394. 12:05industry experts think. Before we get to
  395. 12:07that, if this is your first video of
  396. 12:08mine, welcome the channel. But if it's
  397. 12:10your second or more, this is our
  398. 12:11anti-slap agreement. Everything here,
  399. 12:13the hours of research, the hours of
  400. 12:15editing, this is built entirely by
  401. 12:16humans for humans. So all that I ask as
  402. 12:19part of this agreement is you subscribe
  403. 12:20to help this content reach more people.
  404. 12:22And if you already subscribe, you're a
  405. 12:23legend, but make sure to turn on those
  406. 12:25post notifications. And as a thank you
  407. 12:27to everyone who watches, I give away a
  408. 12:28Claude Max subscription every video. So
  409. 12:30this video's winner is YTP Pepper for
  410. 12:33building an MMO RPG. For this video,
  411. 12:36comment below with what you're building
  412. 12:37to enter the next giveaway. Let's get to
  413. 12:39the next process we'll reverse engineer,
  414. 12:40which is strategic thinking. This is how
  415. 12:42you can reverse engineer how an expert
  416. 12:44thinks and works. Of everything in this
  417. 12:46video, this is the most important
  418. 12:47section. So I talked about earlier
  419. 12:49stealing like an artist, and this is the
  420. 12:51same idea at a more professional
  421. 12:52strategic level. And we're actually
  422. 12:54going to follow a Japanese martial arts
  423. 12:56learning practice called Shuhari. And
  424. 12:58here's how the practice breaks down.
  425. 13:00Shu, which means obey. Ha, which means
  426. 13:03break. And Ri, which means move away. To
  427. 13:05explain all three of these steps,
  428. 13:07there's a great example of this in the
  429. 13:09TV show The Bear. Richie spends a week
  430. 13:11working at a top fine dining restaurant
  431. 13:13where he starts by polishing forks.
  432. 13:14That's Shu. That's following someone
  433. 13:16else's system exactly, even when you
  434. 13:18don't understand why yet. Then he starts
  435. 13:20to get what makes that restaurant works,
  436. 13:21right? He sees how the pieces fit
  437. 13:23together, what matters, and how those
  438. 13:24practices could translate to a different
  439. 13:27environment. That's Ha. And eventually
  440. 13:28he brings the experience back to The
  441. 13:30Bear where he and the team build
  442. 13:32something of their own. That's Ri,
  443. 13:34taking everything you've learned and
  444. 13:35developing an approach that works for
  445. 13:37your own goals. First you obey, then you
  446. 13:39break, and finally, once you're ready,
  447. 13:41you move away. So how do you actually do
  448. 13:43this with AI? Well, first pick who you
  449. 13:45want to learn from. And this can be
  450. 13:46people in your own circle or people that
  451. 13:48you respect from afar. Then we want to
  452. 13:50gather. We're going to gather
  453. 13:51information from them so we can download
  454. 13:53their brain directly to AI. For people
  455. 13:55in your circle, just collect information
  456. 13:57you know about them already. And so you
  457. 13:58can use Claude connectors to pull in all
  458. 14:00their context, right? Their calls and
  459. 14:02conversations, their research frameworks
  460. 14:04and written work, their previous
  461. 14:05reports, their emails, scripts,
  462. 14:07presentations, anything else that shows
  463. 14:09how they think. And for people that you
  464. 14:10admire but you don't know, you'll need
  465. 14:12to work with public material. The
  466. 14:14simplest route is to use something like
  467. 14:15buildpartner.ai's expert advice, which
  468. 14:17is pre-trained on top professionals. But
  469. 14:19if the person you want isn't there, you
  470. 14:21can clone them with buildpartner's
  471. 14:22social tools, where you can run this
  472. 14:24prompt to search Instagram, Twitter,
  473. 14:25YouTube, Substack, and more to gather
  474. 14:28their material. And if you don't know,
  475. 14:29Build Partner is my tool, which is
  476. 14:31linked below if you do want to use it.
  477. 14:32Three, you then want to design a Claude
  478. 14:34skill using that context. The goal isn't
  479. 14:37to just give AI a bunch of documents.
  480. 14:39It's to capture how this person thinks,
  481. 14:40what framework they use, and how they
  482. 14:42approach decisions. For someone in your
  483. 14:44circle, you can build a skill around the
  484. 14:45material you've gathered. For someone
  485. 14:47who you admire from afar, you're
  486. 14:48building a model from their public body
  487. 14:50of work. Four, rebuild plus grade. This
  488. 14:53is where you really put it to work. You
  489. 14:54could ask, what would this person think
  490. 14:56about this situation? Give it a real
  491. 14:57problem to work through. See how it
  492. 14:59approaches the question, what it
  493. 15:00prioritizes, and what recommendations it
  494. 15:02makes. Then compare that what you know
  495. 15:04about the expert's actual thinking and
  496. 15:06work. Where does it get it right? Where
  497. 15:07does it miss something? What are the
  498. 15:09assumptions it's making? How can you
  499. 15:10actually improve this? And now this is
  500. 15:12where you start understanding the
  501. 15:13underlying method, not just collecting
  502. 15:15someone else's opinions. [music] And
  503. 15:17then finally, step five, iterate. Keep
  504. 15:19improving it as you find gaps, gather
  505. 15:21more context, correct assumptions, and
  506. 15:23refine the skill. Over time, you're
  507. 15:24moving through the same Shu Ha Ri
  508. 15:26progression. First, learning someone
  509. 15:27else's method, then understanding how to
  510. 15:29adapt it, and eventually developing your
  511. 15:31own. That's the real opportunity here.
  512. 15:33You're not just creating an AI version
  513. 15:34of someone else's expertise. You're
  514. 15:36using it as a way to build out your own.
  515. 15:38Now let's zoom out for a second and
  516. 15:39bring it all together. REA is a tool
  517. 15:41that rebuilds apps, but the real lesson
  518. 15:43is the five-step process behind it.
  519. 15:45Pick, gather, design, rebuild plus
  520. 15:47grade, and then iterate. It's the same
  521. 15:49loop every time, but you want to copy
  522. 15:51the process, not the property. And if
  523. 15:53you're still hesitant, Paul Graham put
  524. 15:55it very well. There's no better way to
  525. 15:56learn how something works than by trying
  526. 15:58to reproduce it. Nor does copying
  527. 16:00necessarily make your work unoriginal.
  528. 16:02Now I will say this tool REA is scary
  529. 16:04because it calls into question the value
  530. 16:06of products in our economy. So what do I
  531. 16:08think of the future of software? If
  532. 16:10anyone can rebuild how an app works in
  533. 16:12minutes, the code stops being the moat.
  534. 16:14So what is actually left that's hard to
  535. 16:15copy? Let's go back to what REA can't
  536. 16:18see. That's the back end, the
  537. 16:19relationships, the trust, and the
  538. 16:20judgment about what to build and why.
  539. 16:22When the cost of building a feature
  540. 16:24drops towards zero, the scarce thing
  541. 16:26isn't the software. It's about knowing
  542. 16:27what you need and having the taste to
  543. 16:28know when it's good. The winners won't
  544. 16:30be the people who build the most.
  545. 16:32They'll be the ones who know what's
  546. 16:33worth building for their specific
  547. 16:34customer. And building on that, a
  548. 16:36mindset I say to my team, we build
  549. 16:37things you can't just five code. So what
  550. 16:39are the things that make you special? So
  551. 16:41if it's Uber, it's not the app, it's the
  552. 16:43driver. If it's Airbnb, it's not the
  553. 16:45app, it's the hosts and the guests.
  554. 16:47There's plenty of things that make you
  555. 16:48valuable that AI just can't copy and
  556. 16:50paste. Now I touched on it earlier, but
  557. 16:52one problem you'll run into is this
  558. 16:54non-productive building dopamine loop
  559. 16:56where you continuously build things that
  560. 16:58have no real value. And the most
  561. 16:59effective AI builders know what to build
  562. 17:01and what not to build. And to get a
  563. 17:03better understanding of this, go check
  564. 17:04out this video where I share the five
  565. 17:06questions you can ask yourself to build
  566. 17:0810 times faster. If you're new here, my
  567. 17:10name's Austin Arkazy. Welcome to the
  568. 17:12channel and I'll see you in the next
  569. 17:13video. Peace.

About this transcript

This page contains the full transcript of Reverse Engineer Anything Does Way More Than You Think by Austin Marchese, generated from the public captions YouTube serves with the video. The transcript has 3,881 words across 569 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.