You Need To Learn Claude Code In 2026 — Transcript

Full transcript

1. 0:00Most people still think building
2. 0:02software is too hard. That was true
3. 0:04until last year. I am not a programmer
4. 0:08and I wanted a claw code tutorial that
5. 0:11normal people could actually use. So I
6. 0:14made this. If you've never written a
7. 0:17single line of code, you will still go
8. 0:19from a blank screen to something real
9. 0:21[music] in one afternoon in plain
10. 0:24English. Whatever role or industry
11. 0:26you're in, you can now build the exact
12. 0:29tool your work was missing. Once you
13. 0:31understand this simple four-step system
14. 0:34in this video, [music] you can build
15. 0:35almost anything. So, let's get started.
16. 0:38If you look at claude code, it will feel
17. 0:41intimidating at first. Everyone thinks
18. 0:43you have to be a techie to build cool
19. 0:46apps. They're wrong. Imagine tomorrow
20. 0:49you have to negotiate an important deal
21. 0:52in Beijing, but you don't speak a word
22. 0:55of Mandarin, but you have a worldclass
23. 0:58translator [music] who converts it into
24. 1:00Mandarin to your partners in Beijing.
25. 1:03Now you can sit there, have a
26. 1:05conversation, you can negotiate and
27. 1:07close the deal without ever touching a
28. 1:10single word of that language. software
29. 1:13programming and building apps in Python
30. 1:16or JavaScript. You know, they all feel
31. 1:19like Mandarin if you've never seen them
32. 1:22before. You can start building products,
33. 1:25projects, [music] and internal software
34. 1:27tools without any coding today because
35. 1:30Claude Code is your translator. If you
36. 1:34can speak in English, you can get the
37. 1:36deal done without knowing a single word
38. 1:39of any programming language. Claude
39. 1:41[music] chat helps you think. Claude
40. 1:43code work helps you do and claude code
41. 1:47helps you build. So if you think that ah
42. 1:50man coding is not for people like me, I
43. 1:54want you to test that assumption in the
44. 1:56next few minutes. What would you need to
45. 1:58build your first simple app? Not a
46. 2:00single line of code. Now this is the
47. 2:03desktop version of clawed app. At the
48. 2:06top you can see claude chat and code.
49. 2:10Select code. By the way, this is the
50. 2:12universal icon for code everywhere. Down
51. 2:14here is where I tell Claude what I want.
52. 2:17You'll see words like repositories
53. 2:20sometimes or GitHub. Ignore them for
54. 2:24now. For this first build, let's use
55. 2:26clot code locally. Pick a folder where
56. 2:30you want the project to live and then
57. 2:32type build me a beautiful 25inut focus
58. 2:38timer with start, pause, reset, and a
59. 2:43progress ring. That's it. And watch what
60. 2:47happens. Cloud code starts figuring out
61. 2:50what it needs. Creating the files,
62. 2:53writing the code, putting the pieces
63. 2:55together. And just in a few minutes,
64. 2:58there will be an actual application
65. 3:01built for you. Let's open it. Here's the
66. 3:06timer. Looks good. Start, pause, reset.
67. 3:09Okay, all the buttons are there. So,
68. 3:11that's the first moment when you start
69. 3:14feeling, wait a minute, I built that. I
70. 3:18can build this thing. You don't need to
71. 3:20know what programming language Claude
72. 3:23used or what all those strange messages
73. 3:26that scrolled on the screen meant. Now
74. 3:29look at the final product. Do you like
75. 3:32the way it looks? If not, you can change
76. 3:35it by just prompting. It's really that
77. 3:37simple. That's the first big shift of
78. 3:41how you use clawed code. You don't have
79. 3:44to manage the code. You can manage the
80. 3:47outcome. Now, we cheated a little bit
81. 3:49here. A timer is deliberately a simple
82. 3:53app. What happens when the thing you
83. 3:56want to build is slightly more
84. 3:57complicated? Well, let's go there next.
85. 4:00We'll take our first bite. The best way
86. 4:03to build software tools and apps is to
87. 4:05stop thinking about software. Imagine
88. 4:08you're the director of a $200 million
89. 4:12Hollywood blockbuster. You show up on
90. 4:15the set and while the entire crew is
91. 4:18waiting outside, you tear up the script
92. 4:21and spend 3 hours rewriting the entire
93. 4:24scene on a paper napkin. That would be
94. 4:28so chaotic. But that's exactly what
95. 4:31happened during a movie called Joker 2.
96. 4:33And the movie was a mess according to
97. 4:36many people. Now contrast that with Mad
98. 4:39Max Fury Road. The entire movie is like
99. 4:43one big car chase through the desert.
100. 4:46So, the director, George Miller, and his
101. 4:49team map the entire movie through
102. 4:53roughly about 3,500
103. 4:56storyboard panels. Each scene, each
104. 4:59stunt was meticulously planned. And the
105. 5:02funny part is, it's one of the most
106. 5:04chaotic action movies [music] ever made,
107. 5:07and ironically, one of the best planned
108. 5:09ones. So before you start building on
109. 5:12any of these vibe coding tools, plan
110. 5:15something, build something small, test
111. 5:18it, and improve it. I like a simpler
112. 5:22version that I made. I call it bite. You
113. 5:26don't eat the entire watermelon at once,
114. 5:29right? [music] You take one small bite
115. 5:31at a time. Bite is your map. B is
116. 5:34briefing. What are you trying to build?
117. 5:36[music] For whom and why. I is
118. 5:39implement. Build the smallest useful
119. 5:42version. T is [music] test. Use it and
120. 5:45see where it breaks. How do you improve
121. 5:47it? And E is [music] evolve. Good
122. 5:50products evolve over time. And so do
123. 5:53good builders. And by the way, if you
124. 5:54like to get such tools and frameworks
125. 5:56[music] delivered to your inbox, please
126. 5:59subscribe to my email. The link is below
127. 6:02and it's totally free. I love writing
128. 6:04it. All right, let's use this framework
129. 6:06to build something slightly more useful.
130. 6:09Let's build a simple relationship
131. 6:12manager for people who matter the
132. 6:14[music] most to me professionally,
133. 6:17personally, former colleagues, my
134. 6:19mentors, my partners, people I want to
135. 6:22stay close to, but I can't because I'm
136. 6:25running around. So, why not build
137. 6:28something that's your own that's focused
138. 6:30on just one question? Who matters to me?
139. 6:33and how am I staying in touch with them?
140. 6:36So, let's get started. We'll start with
141. 6:38B for briefing. AI can build faster than
142. 6:41any human team, but at that speed, it
143. 6:44needs a lot of direction. In your
144. 6:46briefing, answer four questions. Who is
145. 6:50this for? What problem are
146. 6:52[clears throat] we solving? What does
147. 6:54the first version need to do? And just
148. 6:57as importantly, what are we not building
149. 7:00in this product? Let's do it for real. I
150. 7:04open clot code and give it a simple
151. 7:07prompt. I want to build a personal
152. 7:12relationship manager called Orbit. I
153. 7:15just named it Orbit. Here's the problem
154. 7:17I want to solve. I want to track how I
155. 7:19stay in touch with people who matter to
156. 7:21me. For version one, use the sample data
157. 7:26I provided. Show each person's name,
158. 7:29company, when we spoke last, and notes.
159. 7:32Make it look professional, sparse, and
160. 7:36Applelike.
161. 7:38Use a muted color palette.
162. 7:41Don't build any email or LinkedIn
163. 7:44integrations yet. Before writing code,
164. 7:47make a plan and ask me anything
165. 7:50important that I have missed. And this
166. 7:52is the most important part of the
167. 7:54prompt. I am not a techie or a software
168. 7:57engineer. Don't ask me to approve or
169. 8:00review any technical commands. Make all
170. 8:04the technical decisions yourself and
171. 8:07only ask me questions about what the app
172. 8:09should do in plain language. That's your
173. 8:13example prompt. But you get my point,
174. 8:16right? You can write your own prompt
175. 8:18that is better than the one you just
176. 8:20saw. First, it'll ask you a bunch of
177. 8:22questions about what you want to build.
178. 8:24It'll clarify if there are any tradeoffs
179. 8:27you need to make. And depending on what
180. 8:29you were building, it'll ask you
181. 8:31different questions. But it won't ask
182. 8:33you questions about what kind of
183. 8:35programming language it should write in
184. 8:38or what kind of database you want to
185. 8:40design. You can decide what you want and
186. 8:43it decides how to build it. If you don't
187. 8:46understand a bunch of things in the
188. 8:48plan, you can ask questions. You can
189. 8:51make suggestions to change things in the
190. 8:53plan. Claude can update the plan anytime
191. 8:56you give it feedback. In my case, after
192. 8:59I read the plan, I realized I had left
193. 9:02out one very important detail in the
194. 9:06original prompt. The actual user
195. 9:09experience. My prompt said nothing about
196. 9:13what the layout would look like and what
197. 9:15to show where. So I type in plain
198. 9:18English. It should look professional and
199. 9:21slick as if it's designed by Apple. Make
200. 9:24it a threepane layout. Left sidebar with
201. 9:28my categories. Middle column with list
202. 9:32of names. When I click on someone, their
203. 9:35profile opens on the right side. And
204. 9:38after adding a little bit of that, I
205. 9:41approve the plan. And now cloud code
206. 9:45starts building. To see what happens
207. 9:47next, we move to I for implement. The
208. 9:51biggest mistake trying to build
209. 9:53everything at once. We'll dig into that
210. 9:55important concept once Claude finishes
211. 9:58building the first version. But you'll
212. 10:00see that right now it starts creating
213. 10:02and editing files, running commands,
214. 10:06installing whatever it needs. If none of
215. 10:09that means anything to you, great. It
216. 10:13doesn't have to. Now look at the bottom
217. 10:16and see if auto mode is available on
218. 10:18your account. And if it is, you can turn
219. 10:21it on and it will keep working without
220. 10:23stopping for every permission that it
221. 10:26needs to ask from you. For anything
222. 10:28sensitive, you can slow down and review
223. 10:31what it's doing, but usually you can sit
224. 10:34back and enjoy your coffee. In my case,
225. 10:36it ran for about 10 to 15 minutes. It
226. 10:40wrote the entire code. It was a
227. 10:42JavaScript code. It checked its own
228. 10:44work. It made its own database and then
229. 10:47came back with its first version. Now,
230. 10:50here's what it looks like. The search
231. 10:54works. I can type and see the people.
232. 10:57That's good. Three PES just like uh we
233. 11:01asked. All good. Let's see. The
234. 11:03categories on the left are all accurate.
235. 11:06It's good. It's genuinely usable. That's
236. 11:09version one of Orbit. And this is what
237. 11:12you should notice about our version one.
238. 11:14It is the simplest thing that is still
239. 11:18useful in tech. There is a concept
240. 11:20called MVP.
241. 11:23not most valuable player but minimum
242. 11:26viable product and I love this concept
243. 11:30because it can be applied to any
244. 11:31business any career any output all
245. 11:35technology startups are obsessed with
246. 11:38this one question what is the smallest
247. 11:41version you can put in someone's hands
248. 11:43that actually solves their problem
249. 11:45because then you can observe how they
250. 11:47use it what works what breaks what parts
251. 11:50they like what parts they actually use
252. 11:54that feedback tells you what to build
253. 11:57next. And one sidebar, but a useful
254. 12:00trick, I like using the strongest
255. 12:04reasoning model for the planning and a
256. 12:07cheaper, faster model for the
257. 12:09implementation. It's like having a
258. 12:11senior person help you strategize and a
259. 12:15junior person handling the leg work.
260. 12:18Let's go to T for test. We have our
261. 12:22minimum viable product. Now, let's test
262. 12:25it and make it even more useful. Your
263. 12:27first version works. Congrats. Great.
264. 12:31Once you start using it, though, you'll
265. 12:33see what's missing. For instance, in
266. 12:36this app that we just built live, you
267. 12:39realize the the names are there, the
268. 12:42data is all there, the dates are all
269. 12:44there, but there are no real insights
270. 12:46that are obvious or actionable. So,
271. 12:49let's go back to the prompt. Put three
272. 12:52cards at the top. How many of my
273. 12:55connections need attention?
274. 12:58How many of them have gone cold? And how
275. 13:02many follow up this week? That's a nice
276. 13:05prompt. A few minutes later, there they
277. 13:08are. Now, I open the same app and I have
278. 13:13some kind of a dashboard at the top, not
279. 13:16just data. When you're using clot code
280. 13:18or any vibe coding tools, your goal is
281. 13:23to get to the answer and make whatever
282. 13:27you're building more insightful and more
283. 13:29useful to you or the actual user. Then
284. 13:33after a little more testing, you might
285. 13:36find other gaps. For instance, when I
286. 13:39reach out to someone, there's no way to
287. 13:41tell, at least in the app, how I did it
288. 13:44and when I did it. So you can go back to
289. 13:47the prompt and add this add a button
290. 13:50called log contact to each person's
291. 13:53profile. When I click it, let me pick
292. 13:57how we talked. Email, call, text,
293. 14:01coffee. Then let me pick the date and
294. 14:05let me add a quick note. And now Claude
295. 14:07Code will go back and change the code.
296. 14:10It'll rebuild some of the parts. When
297. 14:12it's ready, I click on it. I check it
298. 14:14out. Nice. It's working. Now you have
299. 14:17started inching toward your own
300. 14:21relationship management system. It's not
301. 14:24a spreadsheet anymore. It's how you keep
302. 14:26in touch with those who are close to
303. 14:29you, those in your closest orbit. And
304. 14:33now we go to the final step. E for
305. 14:35evolve. The final step is not just to
306. 14:38evolve what you build, but to make sure
307. 14:41you evolve with it. Anthropic did a
308. 14:44study that looked at roughly 400,000
309. 14:48real sessions of people building
310. 14:50software apps with AI. And the results
311. 14:53were surprising. They found that people
312. 14:56who had never programmed before
313. 14:59succeeded almost as often as trained
314. 15:03engineers and managers actually scored
315. 15:06slightly higher than even software
316. 15:08engineers. Now, it was just a survey and
317. 15:11not a super rigorous study, but it does
318. 15:15point to an interesting fact. The whole
319. 15:17skill of vibe coding boils down to
320. 15:20knowing what you want and describing it
321. 15:24clearly to your staff, which is your AI
322. 15:28coder in this case. So, the best people
323. 15:31managers become the best AI managers. So
324. 15:36from here you can evolve in three
325. 15:39different directions. First, the app you
326. 15:42build can evolve into a a system. You
327. 15:45can reuse the code to make a CRM for
328. 15:49your side hustle, a tracker for your
329. 15:52kids soccer team, a dashboard for your
330. 15:55goals, whatever comes to mind. Second,
331. 15:59if you feel like it, you can go a bit
332. 16:02deeper. You can learn about APIs and how
333. 16:05to connect with other apps, how to use
334. 16:08GitHub, and what those Python libraries
335. 16:12actually do. And every time you get
336. 16:14stuck, don't be afraid to ask clarifying
337. 16:17questions. Open Claw Chat or any chat
338. 16:20and ask it to explain it like you are a
339. 16:2310year-old. And the third direction is
340. 16:26try to learn something about coding.
341. 16:29Even if your career right now has almost
342. 16:33nothing to do with software because of
343. 16:36all the things AI will change, software
344. 16:39building is one of the areas where AI
345. 16:41will improve incredibly quickly in the
346. 16:44coming years. And there are reasons for
347. 16:46that. But that also means that anyone
348. 16:49with the right tools can create absolute
349. 16:52magic. The world is still going to need
350. 16:55great software engineers to understand
351. 16:57complex systems and architectures and
352. 17:00security and much more. But this new way
353. 17:05of working and building your digital
354. 17:07life is coming at you at 100 miles an
355. 17:10hour. So spend a weekend tinkering with
356. 17:14claude code or lovable or replet or
357. 17:18maybe take an online course because the
358. 17:21real goal is to become comfortable
359. 17:24building, breaking and rebuilding
360. 17:27things. What you build today will tell
361. 17:31you what to build next. So go ahead and
362. 17:35make a bad version.
363. 17:39That's the only way you'll know how to
364. 17:41make the best one.
365. 17:43Thank you and I love you.