How To Use Claude Better Than 99% Of People — Transcript

Full transcript

1. 0:00Give Claude to two people. One will use
2. 0:03it to build a better email. The other
3. 0:06will use it to help build a hundred
4. 0:07million dollar company. Same tool,
5. 0:10different users.
6. 0:12And that gap between people who use AI
7. 0:16and people who wield it is about to
8. 0:19reshape careers and companies and
9. 0:22families and lives over the next decade.
10. 0:26If you're new here, I've been a CEO,
11. 0:29board member, investor in technology
12. 0:32companies worth billions. And today I
13. 0:34want to share with you how to use Claude
14. 0:37better than 99% of people, from beginner
15. 0:40to advanced. And it's way easier than
16. 0:42you think. Claude is not just one app,
17. 0:45it's a whole stack of tools built for a
18. 0:48very different kind of work environment.
19. 0:50[music]
20. 0:51For businesses, there's chat for
21. 0:53thinking, Claude co-work to get work
22. 0:56done, Claude code to build things,
23. 0:59Claude Chrome to browse with a brain,
24. 1:01Claude skills to make it all repeatable,
25. 1:04[music] and lots of tools. Once you
26. 1:06understand how this team of agents work
27. 1:09together, your output can change
28. 1:11dramatically. The best way to visualize
29. 1:13it is this: five activities, think,
30. 1:16remember, execute, build, and browse.
31. 1:19Let's go through each one of them. First
32. 1:21is think. You go to Claude chat. And
33. 1:23this is what most of us do when we think
34. 1:25about chatbots. We input a prompt, we
35. 1:28get a response. You bring a messy
36. 1:30problem, a fuzzy idea, or a rough draft
37. 1:34and work through it. But many of us just
38. 1:37use AI to do all the work so they can
39. 1:40copy and paste. We have all seen those
40. 1:43LinkedIn posts. So, if you go to Claude
41. 1:45chat and type write me a LinkedIn post
42. 1:49about leadership, you'll get a very
43. 1:51polished paragraph that says nothing.
44. 1:55And we've talked about how to prompt
45. 1:57better with things like aim before
46. 1:59[music] actor input mission. Try this
47. 2:01prompt instead. You are a business
48. 2:04school professor. I am trying to explain
49. 2:07why some leaders sound credible and
50. 2:10others don't. I'm attaching my notion
51. 2:13page with a rough idea that talks about
52. 2:16leadership and authenticity, leadership
53. 2:19and humility. Your job is to push back
54. 2:22on what's weak and sharpen what's real.
55. 2:25Find me research that helps me create a
56. 2:28strong insight and then we can write
57. 2:31[music]
58. 2:32a LinkedIn post. You make Claude your
59. 2:35thinking partner now. Of course, like
60. 2:37all AI, it will not do what you tell it
61. 2:40to do, not the first time around at
62. 2:42least. And sometimes it's going to be
63. 2:45super frustrating because you will give
64. 2:46this detailed prompt and just crap will
65. 2:50come back. So, there are three things
66. 2:52that you can do [music]
67. 2:53that can become your power moves when
68. 2:55that happens. First, give it rich
69. 2:58context as much as you can. Connect the
70. 3:00chatbot to your Google Drive, your
71. 3:03email, your calendar, your notion app,
72. 3:05anything else you can. That way you can
73. 3:08attach files to that chat session. It
74. 3:10will read it, it'll get context [music]
75. 3:12from it. Second, ground your insights
76. 3:15with research. You can ask Claude to go
77. 3:17out on the web and do deep research. It
78. 3:20has already learned the entire internet
79. 3:22during its training, so why not use it?
80. 3:25But when you get the research back,
81. 3:27always ask one more time, is this
82. 3:30verified? And third, use AI to build
83. 3:34your AI skill in real time. If you are
84. 3:37patient and curious, you can literally
85. 3:39build your engine while you're flying
86. 3:41the plane. For instance, let's say you
87. 3:43have no background in corporate finance
88. 3:45and you're building a financial model.
89. 3:48You get stuck on how to use a specific
90. 3:51spreadsheet [music] feature or even a
91. 3:53financial concept like IRR or free cash
92. 3:57flow.
93. 3:58No problem. Just start another chat in
94. 4:01another tab and ask away. So, you can
95. 4:03learn and do. Do and learn. The cycle
96. 4:07just continues. Now, let's go meet our
97. 4:09second member of your team.
98. 4:12This one helps you remember and it's
99. 4:14called Claude Projects. We all know when
100. 4:18we're creating something new or solving
101. 4:21a complicated problem, the work doesn't
102. 4:23end [music] in one session or even in
103. 4:27one chat. And that's why all three AI
104. 4:29platforms nowadays have gotten better at
105. 4:32maintaining memory across different chat
106. 4:35sessions. Claude remembers a world
107. 4:37through projects.
108. 4:39A project is where you give the chatbot
109. 4:42your files, your instructions, your
110. 4:45tone, your ongoing work. You're not
111. 4:48starting from zero every time you open
112. 4:50it. And of course,
113. 4:52>> [music]
114. 4:52>> I would be careful about sharing
115. 4:54confidential documents from your office,
116. 4:57but let's say you're actively or
117. 4:59passively looking for a career change or
118. 5:02a new job opportunity. You can create a
119. 5:05project. You drop in your resume, your
120. 5:07LinkedIn profile, the two or three job
121. 5:10descriptions that you like or the
122. 5:12company you're targeting, your writing
123. 5:14samples, your professional bio, the
124. 5:17other resumes that you like, anything
125. 5:19you can think of. And you can customize
126. 5:21this project with exactly what you need.
127. 5:24So, you can add something like, "You're
128. 5:27the world's most badass [music]
129. 5:29recruiter and resume writer who has
130. 5:32placed candidates in top companies in my
131. 5:35industry. I am a brand marketer with 5
132. 5:38years of experience at a large agency
133. 5:40called WPP. Your job is to help me
134. 5:44create a great story for my next
135. 5:47adventure, for my next company, and make
136. 5:49me desirable for the next role. I write
137. 5:52in a direct, confident tone. I don't
138. 5:55like corporate fillers. I hate
139. 5:57buzzwords. When I ask you to tailor
140. 5:59something, match the job description's
141. 6:02language without losing my voice. When I
142. 6:04ask you to do deep research, make sure
143. 6:07you deliver only the relevant and
144. 6:10[music] verified data. Don't make it up.
145. 6:12And when I ask you for feedback, be
146. 6:15honest. Tell me what's weak [music] and
147. 6:17what I need to work on. Be my
148. 6:20constructive partner and strengthen
149. 6:23those areas. This is one way to provide
150. 6:25instructions, but you can see that there
151. 6:28are 100 ways to do it. And that's the
152. 6:29kind of instruction that makes any
153. 6:31project your own. Not a prompt that
154. 6:34you're copying from some YouTuber,
155. 6:36[music]
156. 6:37but a brief that you write for someone
157. 6:40you actually know very well,
158. 6:42>> [music]
159. 6:42>> yourself.
160. 6:44From that point on, every time you come
161. 6:46back, Claude already knows what you want
162. 6:49and who you are.
163. 6:50>> [music]
164. 6:50>> You can ask it to tailor a cover letter,
165. 6:53craft a specific bullet of your resume
166. 6:55for a specific role. A project turns
167. 6:59Claude from a one-off conversation
168. 7:02into an ongoing relationship between
169. 7:05[music] you and your AI. This is also
170. 7:08why I don't customize Claude at the
171. 7:11account level. I like customizing it at
172. 7:14the [music] project level, because that
173. 7:15way I can give it specific context.
174. 7:18Because using AI to plan your vacation
175. 7:21is very different from using it to plan
176. 7:24your retirement. Now, let's go to
177. 7:26execute. And for that, we have our third
178. 7:29team member, Claude co-work. Chat is
179. 7:32where the work begins. Co-work [music]
180. 7:35is where it gets done. You start in a
181. 7:38prompt window, but finish it on your
182. 7:40desktop. Co-work [music] is a desktop
183. 7:43application for Mac or Windows.
184. 7:45Unfortunately, it's available only on
185. 7:48Claude's paid plans right now, but with
186. 7:51Claude Work, instead of asking a
187. 7:52question, you give it a task. When do
188. 7:55you use Claude Work versus chat? Well,
189. 7:57you use Claude Work when you have a lot
190. 7:59of assets already on your computer's
191. 8:01hard drive, when the task needs multiple
192. 8:04steps, and when you need that output to
193. 8:06be on your local drive as a local file.
194. 8:08So, for example, you can tell Claude
195. 8:10Work to do the [music] following. Take
196. 8:13all of these customer interview
197. 8:14transcripts from this folder, these
198. 8:16spreadsheets of accounts, and notes from
199. 8:20last week's product meeting, find top
200. 8:23three reasons on why the customers are
201. 8:25churning. Why are they leaving? Provide
202. 8:28all the supporting data, create a very
203. 8:30clean slide deck that I can use
204. 8:32tomorrow. Put the slide deck in this
205. 8:34folder. [music]
206. 8:35And here's what makes Claude Work even
207. 8:37more interesting. It can reach out to
208. 8:40other AI tools through a connector
209. 8:43called MCP, the Model Context Protocol.
210. 8:47Think of MCP as giving Claude a set of
211. 8:51keys [music]
212. 8:51to other AI tools. It can knock on their
213. 8:54door, unlock the capabilities of those
214. 8:56tools, act with them, act through them
215. 8:59to give you results. [music]
216. 9:01Let me show you what that means in a
217. 9:03real example. Imagine you run a business
218. 9:05selling handcrafted candles. You have a
219. 9:09small business. You have a terrific
220. 9:11product, but you're bootstrapping your
221. 9:12business, right? And you don't have a
222. 9:14huge marketing budget. In the old world,
223. 9:17that would mean that you couldn't create
224. 9:20any ads that looked super polished or
225. 9:22professional. But now, look at what you
226. 9:24can do when you can plug Claude into a
227. 9:27creative AI tool called Hicksfield. They
228. 9:30are the sponsor for this video. Now,
229. 9:32Hicksfield is used by 18 million the
230. 9:35backed by tier one VCs like Excel and
231. 9:38Menlo Ventures, and they've already
232. 9:40crossed the billion-dollar valuation.
233. 9:42And they were the first creative
234. 9:44platform to ship an official MCP
235. 9:47connector for Claude. Here's how you
236. 9:49connect it. You have to do it once, and
237. 9:52that's it. Go to settings, connectors,
238. 9:54plus button, paste the Hexfield URL, and
239. 9:58you're done. That's it. Now you can go
240. 10:00into Co-work, and you tell it to do
241. 10:02whatever you need to sell more products.
242. 10:05For instance, you could say, "Generate
243. 10:0720 beautiful ad creatives for my new
244. 10:11lavender candle line. Three ideas for
245. 10:13Instagram, three editorial close-ups for
246. 10:16the website, and a vertical video for
247. 10:19the launch. Write a tagline for each."
248. 10:21And that's it.
249. 10:23It just runs.
250. 10:24>> [music]
251. 10:24>> Claude Co-work writes the prompt and
252. 10:27passes it on to Hexfield. Hexfield
253. 10:29generates the images and the video, and
254. 10:32the entire set of files land in your
255. 10:35local folder without you needing to hire
256. 10:37a creative agency and pay them thousands
257. 10:40of dollars. That would have taken you
258. 10:42[music] 3 weeks of a professional shoot,
259. 10:44a photographer, a stylist, and weeks of
260. 10:48post-production, and of course thousands
261. 10:49of dollars. But now, you can do it in a
262. 10:52single session with Claude and Hexfield.
263. 10:56This is how AI is leveling the playing
264. 10:59field. A solo founder with the right set
265. 11:02of tools today can now produce the kind
266. 11:04of creative output that used to require
267. 11:07a large team and a large budget. There's
268. 11:10a link in the description below. They
269. 11:12have a free tier. Take it for a spin.
270. 11:14Try it. And you could build something
271. 11:17very cool and interesting this weekend.
272. 11:19Two things to keep in mind. First, the
273. 11:22output will be a starting point. You'll
274. 11:24still have to go through each and every
275. 11:26slide and make sure it's good. But a lot
276. 11:30of heavy lifting is done by Claude
277. 11:32co-worker already. Second, give explicit
278. 11:35permission for Claude co-work to access
279. 11:38only specific directory and specific
280. 11:41files. Claude is very conscious about
281. 11:43security, but please be very mindful
282. 11:46about what you give access to. And I
283. 11:48usually create a subfolder and move all
284. 11:51the files and data into that folder, and
285. 11:54that's all Claude has access to. It's
286. 11:56called sandboxing, so the rest of your
287. 11:58hard drive remains secure. Claude can't
288. 12:01get to it. Fourth
289. 12:03>> [music]
290. 12:03>> is build, and this is where Claude code
291. 12:06comes into play. This is the part that
292. 12:08makes a lot of us nervous. It made me
293. 12:11nervous. The moment we hear code, we're
294. 12:15like,
295. 12:16"Ah, I don't know."
296. 12:17We assume that all that stuff is only
297. 12:20for engineers, but that assumption is
298. 12:23one of the biggest missed opportunities
299. 12:26in AI right now. If you can type in
300. 12:28English, you don't need a computer
301. 12:30science background to code and build
302. 12:33something cool and useful. Let's say you
303. 12:36have a consulting business or a side
304. 12:39hustle.
305. 12:40Simple example, just go to Claude code
306. 12:43and prompt, "I want a dashboard where I
307. 12:46can drop in my sales pipeline and
308. 12:49meeting notes and instantly see which
309. 12:52deals are at risk, what's stuck, and
310. 12:55what my next move should be."
311. 12:58That's all you need. You don't need to
312. 13:00think like an engineer to start using
313. 13:03Claude code. You just need to be clear
314. 13:05about the problem you want to solve and
315. 13:08the thing you want to build. And of
316. 13:11course, [music]
317. 13:11you have to have the right data so that
318. 13:14dashboard you build is useful.
319. 13:17And finally, the fifth team member is
320. 13:20for browsing the internet. And for that,
321. 13:23you can use Claude Chrome. And you're
322. 13:25going to go, "Wait, what?
323. 13:27Don't worry. Claude has this little
324. 13:29extension that you can install in your
325. 13:32Chrome browser. Google already has
326. 13:34something like ask Gemini button baked
327. 13:36into the browser now, but now you can
328. 13:38let Claude see what you're looking at as
329. 13:40well. Help [music] you process it and
330. 13:43act inside your workflow in real time.
331. 13:45And I think about this
332. 13:47>> [music]
333. 13:47>> a lot nowadays, because most of my work
334. 13:51happens inside the browser, right? My
335. 13:53files are on Google Drive or a Microsoft
336. 13:56OneDrive, my emails, my scheduling, my
337. 13:59team meetings, my research, filling out
338. 14:01forms, making purchases, planning
339. 14:04business trips, all of it is [music]
340. 14:06taking place using the browser. And now
341. 14:09Claude is right there with you for all
342. 14:11of it. You're on a job listing site,
343. 14:14Claude can read it and help you tailor
344. 14:16your outreach. You're reading a 40-page
345. 14:19industry report, Claude can pull the
346. 14:21three insights that matter the most in
347. 14:23real time. So, that is your team. There
348. 14:26are other team members, but [music]
349. 14:27these five
350. 14:29are the most interesting and useful.
351. 14:31This is your stack.
352. 14:33But even with the full team working with
353. 14:36you, one gap can make all the difference
354. 14:39between the [music] average and the
355. 14:42elite mode. That's where we go next. Any
356. 14:45AI's results [music] depend on you. Even
357. 14:48with the full team of agents working for
358. 14:50you, Claude is only as good as the input
359. 14:54[music]
360. 14:54you give it. But the superusers who get
361. 14:57exceptional output from Claude, [music]
362. 14:59they direct Claude the way they would
363. 15:02direct a smart person they've just
364. 15:04hired. [music] And the easiest way to do
365. 15:06that is a framework I call Prime.
366. 15:09>> [music]
367. 15:09>> P is purpose. When you prompt Claude,
368. 15:11give it a precise goal.
369. 15:13>> [music]
370. 15:13>> Tell it what you need it to do, like
371. 15:16help me turn this messy proposal into a
372. 15:19five-part organized document that looks
373. 15:22[music] like a sharp client memo. R is
374. 15:25research. We talked about it before. In
375. 15:27many cases, you may need external
376. 15:30research to ground the response. Ask it
377. 15:33to go out and fetch it for you. Ask it
378. 15:35to verify it. And sometimes the raw
379. 15:37material may come from you, your notes,
380. 15:40transcripts, files, [music] examples,
381. 15:43background, constraints,
382. 15:45anything that Claude needs so that
383. 15:47[music] it can stop hallucinating or
384. 15:49guessing. I is for interview. This is
385. 15:53the hidden gem. I really like how Claude
386. 15:55does it.
387. 15:56>> [music]
388. 15:56>> In a lot of cases, you'll see that it
389. 15:59will start asking you multiple-choice
390. 16:01questions. And [music] you pick the
391. 16:03right choice, and based on what you
392. 16:05clicked, it refines its approach [music]
393. 16:08to a specific way of responding to you.
394. 16:11M is for mechanics. Here, you tell
395. 16:13Claude how the output should look like.
396. 16:17>> [music]
397. 16:18>> Do you want bullets or a paragraph? Do
398. 16:21you want a document or a table? Do you
399. 16:23[music] want it concise or you want
400. 16:25detail? Strategic or conversational?
401. 16:28Anything you choose. And finally,
402. 16:30[music] E is for examples. Show Claude
403. 16:33what good looks like, a format you liked
404. 16:36or a tone reference or an outline that
405. 16:39[music] worked before. Examples are one
406. 16:42of the fastest ways to quality. Let's go
407. 16:45through an example. You have an
408. 16:47important presentation coming up
409. 16:48tomorrow.
410. 16:50Run it through prime. Purpose. [music]
411. 16:52What is this meeting really for? Do you
412. 16:55need an approval, alignment? [music]
413. 16:57Are you asking for resources? What is
414. 16:59the outcome you really want? Research.
415. 17:02Give Claude the context. What's the
416. 17:04audience like, the risks, [music] the
417. 17:06data, what your team cares about,
418. 17:09external resources, internal docs,
419. 17:11anything you can [music] get. Interview.
420. 17:13So, tell Claude, "Interview me before
421. 17:15you respond." Now, it will ask you
422. 17:17questions
423. 17:18>> [music]
424. 17:18>> that sharpen its own scope and your
425. 17:21thinking mechanics. Now you know the
426. 17:23output of that deliverable. [music]
427. 17:25You want a 10 slide deck and talking
428. 17:28points. And finally, examples. [music]
429. 17:30Give it a past deck that you liked or
430. 17:32style you admire. So that is prime.
431. 17:35[music] Once you start doing that, AI
432. 17:38stops being a tool and starts [music]
433. 17:40becoming your edge. And by the way, if
434. 17:43you want the frameworks like these in
435. 17:45your inbox, subscribe [music]
436. 17:46to my newsletter. The link is below
437. 17:48somewhere. I mean it's free.
438. 17:51The real edge is the human using it.
439. 17:55Everyone watching [music] this feels
440. 17:56like someone else is getting ahead
441. 18:00faster.
442. 18:01That they are smarter or they are richer
443. 18:04or they have better resources.
444. 18:06>> [music]
445. 18:06>> I know I feel like that sometimes. That
446. 18:09is the default condition of ambition.
447. 18:12The question is not whether you were
448. 18:14outgunned. The question is how you
449. 18:17outshine when you are. You have been
450. 18:19carrying labels for years. I am not
451. 18:22technical.
452. 18:24I'm not a designer. I don't have an MBA.
453. 18:28But those are just labels. For $20 a
454. 18:31month, you can hire something that is
455. 18:35already better than most MBAs and PhDs
456. 18:38now. You can think more sharply with it.
457. 18:40[music] You can design better. You can
458. 18:42code. You can build. But do you have the
459. 18:45curiosity to keep learning?
460. 18:47>> [music]
461. 18:48>> The the tenacity to get through that
462. 18:51machine friction.
463. 18:53The patience [music]
464. 18:54to build real capabilities. Or will you
465. 18:58slowly replace [music]
466. 18:59real effort with convenience
467. 19:02and lose your instinct to wonder?
468. 19:06>> [music]
469. 19:06>> AI will give you a lot of things. But
470. 19:09what AI can't give you
471. 19:12>> [music]
472. 19:12>> is that innate curiosity that made you
473. 19:15press play
474. 19:17on this video.
475. 19:19And you already have it.
476. 19:21As one of my mentors used to say,
477. 19:24"You are the one you've been waiting
478. 19:26for."
479. 19:27Thank you.
480. 19:28And I love you.