YouTube2Text

Baue & automatisiere dein Marketing mit Claude Code — Transcript

by Ismajl Pukaj | KI Agentur · 3,686 words · 550 segments · language en · Watch on YouTube

Full transcript

  1. 0:00This here is my marketing hub, and right
  2. 0:02now three agents are working
  3. 0:03simultaneously. One is building a page,
  4. 0:06one is checking it against 53 rules, and
  5. 0:08one is currently designing new ads.
  6. 0:10Every morning at 9:00 AM, I get a report
  7. 0:12on every channel.
  8. 0:14It shows how many clicks and leads there
  9. 0:16were yesterday and which search terms
  10. 0:18burned money, just to continuously
  11. 0:19optimize the ads and see what is working
  12. 0:21well and what needs changing.
  13. 0:24Every lead lands here with the channel,
  14. 0:26the campaign, and the search term it
  15. 0:27ultimately came from.
  16. 0:30And if I reject any of this work, the
  17. 0:32agent team turns it into a fixed rule so
  18. 0:33that this mistake doesn't happen again.
  19. 0:36Behind all of this are 1,404 sessions
  20. 0:38with Claude and Codex, and in this video
  21. 0:40, I'll first show you what this hub
  22. 0:42actually is and what it can do.
  23. 0:44After that, we'll build it together,
  24. 0:46prompt by prompt, so that in the end,
  25. 0:48you have your own marketing hub tailored
  26. 0:50to your company.
  27. 0:51Let's start with the overview, because
  28. 0:53here I see at a glance every morning
  29. 0:55where my marketing stands. As an example
  30. 0:57, the company Blauwerk Pools is running
  31. 0:59here, a pool builder from Hanover.
  32. 1:01I completely made this company up for
  33. 1:03this video, but I have applied the
  34. 1:04principles and everything you see here
  35. 1:06multiple times exactly like this and
  36. 1:07used it for myself as well.
  37. 1:09And the numbers here are sample figures
  38. 1:11, but they aren't completely pulled out
  39. 1:12of thin air either. At the very top are
  40. 1:14the four figures that really count for a
  41. 1:16business: visitors, leads, the
  42. 1:17conversion rate between them, and the
  43. 1:19money currently outstanding in quotes.
  44. 1:21And below that, you see each channel
  45. 1:23individually: Google Ads, Meta, and
  46. 1:25organic search, each with visitors,
  47. 1:27leads, and the respective conversion
  48. 1:29rate. This way, I see immediately which
  49. 1:31channel brings leads and which is
  50. 1:32currently only bringing clicks. On the
  51. 1:34right is everything that is waiting for
  52. 1:35my decision. Right now, that’s a new
  53. 1:37landing page. The check has already
  54. 1:39verified it against three checklists,
  55. 1:41and I can now accept it, reject it, or
  56. 1:43ask questions about what exactly it is
  57. 1:45or whatever else is still on my mind
  58. 1:48that isn't clearly defined. When I click
  59. 1:50on "ask," Claude opens with everything
  60. 1:52belonging to this page, and I can ask
  61. 1:54directly why something is built the way
  62. 1:56it is. For example, I ask why the form
  63. 1:58is on the right and not at the bottom.
  64. 1:59Claude provides the rule behind it and
  65. 2:02the test that proves it. At the bottom,
  66. 2:04you can see which agents are currently
  67. 2:05working and on what. Next to it, the
  68. 2:07notifications come in, like the daily
  69. 2:09Google Ads report or a new inquiry that
  70. 2:11just arrived. This way, you always have
  71. 2:13a live overview, and on the left here is
  72. 2:15my actual activity. Each box is a day,
  73. 2:17and the brighter it is, the more
  74. 2:19sessions I had that day with Claude Code
  75. 2:21or Codex. The most important area for
  76. 2:23the business is the inquiries, because
  77. 2:24that is what marketing is ultimately
  78. 2:26measured by. Every inquiry here arrives
  79. 2:28as a card. It states what the customer
  80. 2:30wants, e.g., an 8 x 4 m concrete pool,
  81. 2:33where they live, and roughly how big the
  82. 2:35project is. Below that, it says where
  83. 2:36the inquiry came from. This is Google
  84. 2:38Ads, and the card even shows the search
  85. 2:40term the customer used to find us. I’m
  86. 2:42now sending a test inquiry via the form
  87. 2:44on the landing page. A few seconds later
  88. 2:46, it’s there. The office has already
  89. 2:48sent the customer a confirmation, and
  90. 2:50management has scheduled a callback.
  91. 2:52Once I’ve spoken with the customer, I
  92. 2:54simply drag the card to the next column
  93. 2:55. From the site visit, it moves to the
  94. 2:57quote and then to the order.
  95. 3:00With every won order, the hub reports
  96. 3:02back to Google Ads which search term
  97. 3:04ultimately brought in revenue, allowing
  98. 3:05Google's system and its AI to further
  99. 3:07optimize.
  100. 3:09Now for the reports, because this is
  101. 3:11where the hub works without me having to
  102. 3:13do anything. Every morning at 9:00 AM, a
  103. 3:15daily check runs for every channel. For
  104. 3:17Google Ads, it reads all of yesterday’s
  105. 3:19metrics—impressions, clicks, costs, and
  106. 3:21inquiries. Then it goes through every
  107. 3:23search term that someone clicked on.
  108. 3:25Terms that are clearly unrelated to the
  109. 3:27offer are immediately excluded. This
  110. 3:29provides further optimization, such as
  111. 3:31job searches, DIY instructions, or the
  112. 3:33names of other companies—competitors—if
  113. 3:35you don't want them. Anything that isn't
  114. 3:38clear is not blocked. That ends up here
  115. 3:40under review, where I decide myself, and
  116. 3:41at the end, there is a recommendation
  117. 3:43with the relevant figures.
  118. 3:45Here, it suggests directing the search
  119. 3:46term "concrete pool price" to its own
  120. 3:48page because price-seekers aren't
  121. 3:49inquiring on the pool construction page
  122. 3:51.
  123. 3:51This control is not just a gimmick. A
  124. 3:53daily check like this has been running
  125. 3:55for the Google Ads account every morning
  126. 3:57at 9:00 AM since October. Budgets, bids
  127. 3:59, and ads are never changed; that stays
  128. 4:01the same, and on Mondays, the weekly
  129. 4:03analysis follows. An agent compares all
  130. 4:05channels with one another and writes up
  131. 4:06recommendations for me based on that.
  132. 4:08For instance, it found that the hero
  133. 4:10section with the form on the right
  134. 4:11brings in significantly more inquiries
  135. 4:13than the old version. Additionally, it
  136. 4:15reports that a creative on Meta is
  137. 4:17fatiguing because the target audience
  138. 4:18has already seen it over four times. I
  139. 4:20can assign any recommendation to the
  140. 4:22appropriate agent with a single click.
  141. 4:24Here, the task goes to the ads team, and
  142. 4:26they immediately start on three new
  143. 4:27motifs.
  144. 4:28To ensure the agents don't just produce
  145. 4:30anything, the headquarters knows the
  146. 4:32entire funnel.
  147. 4:33Here you can see every stage from first
  148. 4:35contact to the order. Between the stages
  149. 4:37, it shows how many people move forward
  150. 4:39, i.e., the percentage conversion rate.
  151. 4:41And the headquarters directly marks the
  152. 4:43bottleneck, showing exactly where the
  153. 4:45funnel is stalling. Here it is the step
  154. 4:47from visitor to inquiry, which turns
  155. 4:48directly into a task for the website.
  156. 4:50Below that is the strategy for each
  157. 4:52stage. Those who haven't even thought
  158. 4:54about a pool yet get a before-and-after
  159. 4:55Reel. Those who are already comparing
  160. 4:57prices get a guide on the costs. And
  161. 4:59those who want to make an inquiry now
  162. 5:01land on the landing page with the form.
  163. 5:03To do this, the headquarters knows the
  164. 5:04target groups. For each group, it lists
  165. 5:06what they desire, what they are looking
  166. 5:08for, and what makes them hesitate. For
  167. 5:11every objection, there is practically a
  168. 5:12matching response. And at the bottom,
  169. 5:14you can see immediately where content is
  170. 5:15still missing. For couples over 50, for
  171. 5:17example, there is nothing yet for the
  172. 5:18initial contact. This also goes out as a
  173. 5:20task with one click. Everything the
  174. 5:22agents need to know about the company is
  175. 5:23in the company profile. This lists the
  176. 5:25services, the region, and the goal.
  177. 5:27Below that, it states for each channel
  178. 5:28who is allowed to do what. On the
  179. 5:29website, the team may create drafts but
  180. 5:31cannot publish anything. For Google Ads
  181. 5:33and Meta, every change ultimately needs
  182. 5:36my approval, as does every budget. And
  183. 5:38here it shows which claims are
  184. 5:39substantiated and which are still
  185. 5:40missing. Blauwerk Pools does not have
  186. 5:42reviews, reference projects, or
  187. 5:44certificates yet. That is why no agent
  188. 5:46writes such things on a page. It places
  189. 5:48a visible placeholder there until real
  190. 5:50data is available. At the very bottom,
  191. 5:52you see the integrations, meaning where
  192. 5:53the numbers come from: Google Ads,
  193. 5:54Search Console, Meta, the form, and
  194. 5:56keyword data. Now to an area that saves
  195. 5:58me a lot of work: the Reel Lab. For
  196. 6:00example, I simply enter an Instagram
  197. 6:02profile here, such as a strong profile
  198. 6:04from the industry, and click analyze.
  199. 6:06The headquarters pulls the last 30 Reels
  200. 6:08, analyzes every image and sound, and
  201. 6:10rates the profile. Here you can see the
  202. 6:12strongest Reels, and for each one, it
  203. 6:13shows how far it exceeds the average of
  204. 6:15this profile. On the right, it shows
  205. 6:17what works: before and after in the
  206. 6:18first few seconds, a length of 9 to 14
  207. 6:20seconds, and the price in the title.
  208. 6:22From this, the hub creates a video idea
  209. 6:24for your company, already broken down
  210. 6:26into scenes with seconds. With one click
  211. 6:28, it goes to the video agent. In the
  212. 6:29YouTube section, things continue in the
  213. 6:31same way. Each video idea moves here
  214. 6:33from the concept through the script and
  215. 6:35filming to the finished video. For every
  216. 6:37video, it shows how often it was viewed
  217. 6:38, how many people clicked on the landing
  218. 6:40page afterward, and how many inquiries
  219. 6:42resulted from it. The ideas come from
  220. 6:44three sources: from search terms, from
  221. 6:46the Real Lab, and from questions
  222. 6:47customers ask in their inquiries. In the
  223. 6:49SEO section, I can see how the company
  224. 6:51ranks on Google and in AI search. Here
  225. 6:53are the search terms with search volume
  226. 6:54and position, and next to them, the
  227. 6:56trend over the last few weeks. Below
  228. 6:57that, the hub checks whether the company
  229. 6:59is mentioned in AI answers. To do this,
  230. 7:01it asks ten typical questions in ChatGPT
  231. 7:03, Gemini, and Google's AI overview, and
  232. 7:05the SEO reports are kept here. Each
  233. 7:07report has action items, and each item
  234. 7:09goes directly to an agent. This is the
  235. 7:10actual audit report for the landing page
  236. 7:12. 34 out of 52 rules are met, and the
  237. 7:14open points can only be completed at
  238. 7:16launch. For example, HTTPS and the
  239. 7:19sitemap. Lighthouse shows 98 on mobile,
  240. 7:21so that's the performance. Now I'll show
  241. 7:23you how the team processes a task from
  242. 7:25start to finish. This is a real order
  243. 7:28that I just placed in the hub. Under
  244. 7:30websites, you'll find the landing page
  245. 7:31for Blauwerk Pulse. Here you can see all
  246. 7:33three versions side-by-side, and here is
  247. 7:35the entire process. Management accepts
  248. 7:37the order and plans the steps. The
  249. 7:39research team estimates the costs for
  250. 7:41the keyword query and pauses until I
  251. 7:43approve it. After that, we start with
  252. 7:44SEO according to our specifications. The
  253. 7:46copy for the page is written, and the
  254. 7:48website is built. The audit finds nine
  255. 7:50errors in the first round and sends the
  256. 7:51page back. I then rejected this version
  257. 7:53. My reason was a white veil over the
  258. 7:55photo, which made the whole page look
  259. 7:57pale. The team immediately turned that
  260. 7:59into a rule. The second version was
  261. 8:00better, but it still felt like an
  262. 8:02AI-generated page to me, with too much
  263. 8:04text and the form only at the very
  264. 8:05bottom. So I rejected that one as well.
  265. 8:07Three new rules emerged from this, and I
  266. 8:09just accepted the third version. A short
  267. 8:12hero, a large photo, and the form right
  268. 8:14on the right in the first frame. These
  269. 8:16rules end up here under "learned," and
  270. 8:18that is the part that makes the hub
  271. 8:19better every week. A new rule does not
  272. 8:21apply immediately. It waits for my
  273. 8:23go-ahead first. After that, it is on
  274. 8:24probation and must prove itself through
  275. 8:26three tasks. Only when the same mistake
  276. 8:27hasn't occurred three times in a row
  277. 8:29does the rule become fixed for every new
  278. 8:31task. And here is the team that does all
  279. 8:33the work. Ten agents, each with a clear
  280. 8:35task. For each one, you can see what
  281. 8:37it's currently working on, how often it
  282. 8:39has run, how many of its results I've
  283. 8:40accepted, and what rules it has learned
  284. 8:41. And every agent has a limit it never
  285. 8:44crosses. The ads never change a budget,
  286. 8:46the websites never go live without my
  287. 8:48go-ahead, and the research tools don't
  288. 8:50pull anything that costs money before
  289. 8:52I've at least approved it. Under
  290. 8:54workflows are the fixed paths through
  291. 8:56the team. Every workflow ends with me,
  292. 8:58and next to it are the templates,
  293. 8:59meaning how a finished result must look
  294. 9:02. Next to that are the skills, the
  295. 9:03knowledge base, and the checklists. The
  296. 9:05skills are the agents 'tools, for
  297. 9:06example, the SEO audit or the skill for
  298. 9:08sales-optimized websites. The knowledge
  299. 9:10base contains every rule with its source
  300. 9:12, for example, directly from Google's
  301. 9:13documentation. And from this knowledge,
  302. 9:15the checklists are built, against which
  303. 9:17the audit checks every page. For the
  304. 9:19landing page, you can see here for every
  305. 9:20point whether it is met, only partially
  306. 9:22met, or still completely missing. So,
  307. 9:24now let's build this headquarters
  308. 9:26together. I will now show you the prompt
  309. 9:28I enter into Claude Code and the result
  310. 9:29it produces. For this, you only need
  311. 9:31Claude Code and an empty folder on your
  312. 9:33computer. In the first step, we set up
  313. 9:34the basic structure. I open Claude Code
  314. 9:36in an empty folder and enter this prompt
  315. 9:38. Create a marketing headquarters for my
  316. 9:40company in this folder.
  317. 9:42Every role, every piece of knowledge,
  318. 9:43every checklist, and every result is its
  319. 9:45own text file.
  320. 9:46Write a Claude.md that explains how the
  321. 9:48team works. Nothing is published and no
  322. 9:50money is spent before I have approved it
  323. 9:52.
  324. 9:53Claude is now creating the folders for
  325. 9:55approvals, requests, agents, skills,
  326. 9:57knowledge, checklists, reports, and also
  327. 9:59lessons learned. And it writes the
  328. 10:01Claude.md.
  329. 10:02Claude reads this file first at every
  330. 10:04session, and it contains how the team
  331. 10:05works and which rules always apply.
  332. 10:07Let's take a quick look inside. It
  333. 10:08states that every assignment first goes
  334. 10:10to management. Management plans the
  335. 10:11steps and distributes them to the agents
  336. 10:13. Below that, it states that every
  337. 10:15result is checked before it is submitted
  338. 10:17for approval and that every step ends up
  339. 10:19in the log with a timestamp, agent, and
  340. 10:21file. That is basically the operating
  341. 10:22manual for your entire team, and you can
  342. 10:24change it yourself at any time. In the
  343. 10:26second step, the headquarters gets
  344. 10:27everything about the company. For this,
  345. 10:28I simply enter the company data directly
  346. 10:30into Claude. Here is an example of my
  347. 10:32company data: Blau Werk Pools, pool
  348. 10:33construction in Hanover and the
  349. 10:34surrounding area. We build concrete
  350. 10:36pools and prefabricated basins, renovate
  351. 10:38old pools, and handle roofing and
  352. 10:40equipment. The goal is more inquiries
  353. 10:42for new pools; create the company
  354. 10:44profile and target groups from this and
  355. 10:46ask me what is still missing. Claude
  356. 10:48creates the company profile and derives
  357. 10:49three target groups from it, each with
  358. 10:51wishes, search terms, and objections.
  359. 10:53And then it asks about what it cannot
  360. 10:55know. For example, whether there are
  361. 10:56genuine reviews, what reference projects
  362. 10:58exist, and how long the company has been
  363. 11:00in business. I answer what I know
  364. 11:01directly in the chat. For example, that
  365. 11:03most customers are families with gardens
  366. 11:05and that many ask about the costs first
  367. 11:06. Claude immediately enters this into
  368. 11:08the company profile and the appropriate
  369. 11:09target group. Everything I have not yet
  370. 11:11answered remains empty. No agent will
  371. 11:14later write a review or a figure on a
  372. 11:16page that is not verified. In the third
  373. 11:18step, we set up the team. Create ten
  374. 11:20agents: Management, Research, SEO,
  375. 11:22Websites, Copywriting, Ads, Video,
  376. 11:23Quality Assurance, Secretariat, and
  377. 11:25Learning.
  378. 11:27Everyone gets a task, their skills,
  379. 11:28their knowledge, who they hand over to,
  380. 11:30and a clear boundary they never cross.
  381. 11:33A separate file is created for each
  382. 11:35agent. Here you can see the SEO agent's
  383. 11:37file. At the top it says what they do,
  384. 11:38below that which skills they use, which
  385. 11:40knowledge they read, who they hand over
  386. 11:42to, and what they never do. Now the
  387. 11:43agents get their tools, the skills. A
  388. 11:45skill is a finished set of instructions
  389. 11:47according to which Claude always
  390. 11:49completes a specific job in the same way
  391. 11:50. Connect the agents with my skills. The
  392. 11:53SEO agent uses the skill for SEO audits
  393. 11:55and keyword research. The website agent
  394. 11:57uses the skill for sales-optimized pages
  395. 11:59, and the ad agent uses the skill for ad
  396. 12:01creatives.
  397. 12:02Claude notes for each agent which skills
  398. 12:04they use and creates a short card for
  399. 12:05each skill.
  400. 12:07On it is the command, what the result
  401. 12:08should be, and whether the skill costs
  402. 12:10anything. If you don't have your own
  403. 12:12skills yet, that's no problem. You can
  404. 12:13simply ask Claude to build a skill from
  405. 12:16a task you already do often anyway. So
  406. 12:18Claude can analyze and build that for
  407. 12:19you. The next time, it will then run
  408. 12:21according to the same instructions. In
  409. 12:22the fourth step, the team gets its
  410. 12:24knowledge; build the knowledge base for
  411. 12:25SEO, AI search, local search, landing
  412. 12:27pages, and sales psychology. Every rule
  413. 12:29needs a source, preferably directly from
  414. 12:31Google. Create checklists from this,
  415. 12:32against which the Quality Assurance
  416. 12:33agent checks every result. Claude now
  417. 12:35researches official sources and writes
  418. 12:37down every rule along with its source.
  419. 12:39Behind it is the link to the source at
  420. 12:41Google. You can verify every rule
  421. 12:42yourself and the team doesn't invent
  422. 12:44their own SEO tips. This is where the
  423. 12:46checklists come from. If the audit
  424. 12:47rejects a page later, it can always say
  425. 12:50exactly which point wasn't met and where
  426. 12:52the rule comes from. In the fifth step,
  427. 12:54we build in the two rules that make the
  428. 12:55whole thing secure and adaptive. Every
  429. 12:57result first lands under approvals and
  430. 13:00waits for my "yes." Every paid query
  431. 13:02shows the estimated costs beforehand.
  432. 13:04When I reject something, I write a
  433. 13:06learning note and turn it into a rule
  434. 13:08that only becomes permanent after my
  435. 13:09approval and three tasks without relapse
  436. 13:11. Claude writes these rules into the
  437. 13:13cloud and creates the "learned" folder.
  438. 13:15From now on, every rejection
  439. 13:16automatically becomes a rule. This is
  440. 13:18how it looked for the landing page. I
  441. 13:19wrote that the page feels like an AI
  442. 13:21page. The team turned this into a
  443. 13:23learning note with my reason, the error
  444. 13:25, and the new rule. The rule is now:
  445. 13:27short hero section, large photo, and the
  446. 13:29form on the right. And because it is
  447. 13:31registered under websites, the agency
  448. 13:32reads it before every new page. In the
  449. 13:34sixth step, we connect the headquarters
  450. 13:36with the channels. Connect my inquiry
  451. 13:38form, Google Ads, and the Search Console
  452. 13:40. Every new inquiry arrives with channel
  453. 13:42, campaign, and search term in the
  454. 13:43inquiries folder. Only read Google Ads
  455. 13:45and Search Console and never change
  456. 13:47anything there without my approval.
  457. 13:49Claude writes small scripts for this.
  458. 13:50The form now sends every inquiry as a
  459. 13:52file to the folder and the ad data is
  460. 13:54only read daily. This is what an inquiry
  461. 13:56file looks like. At the top are name,
  462. 13:58location, and project, below that the
  463. 14:00channel, the campaign, and the search
  464. 14:02term. The board turns this into the card
  465. 14:04you saw earlier in the Kanban. For the
  466. 14:06connection to Google, you need your
  467. 14:07access credentials once. Claude explains
  468. 14:09every step to you, and the keys are then
  469. 14:10stored securely in your keychain and not
  470. 14:12in the folder. In the seventh step, we
  471. 14:14let the headquarters work by itself. Set
  472. 14:16up a daily report for every channel at
  473. 14:189:00 AM each morning. You may exclude
  474. 14:20clearly incorrect search terms.
  475. 14:22Everything else you present to me for
  476. 14:24review, and every Monday, a weekly
  477. 14:25analysis should compare all channels and
  478. 14:27give its recommendation as an order to
  479. 14:29the agents. Claude creates the orders
  480. 14:31for you and schedules them. Each order
  481. 14:33states exactly what the controller is
  482. 14:34allowed to do and what not. It is only
  483. 14:36allowed to exclude terms that clearly do
  484. 14:38not fit, such as job hunting, DIY guides
  485. 14:40, material purchases without tradesmen,
  486. 14:42or third-party company names. General
  487. 14:44terms like pool construction, car
  488. 14:45dealership in the catchment area, and
  489. 14:47anything that has generated an inquiry
  490. 14:48before, it never blocks. These
  491. 14:50boundaries are important, otherwise an
  492. 14:52agent might eventually block a term that
  493. 14:53actually brings in orders. Let's build
  494. 14:55the board, meaning the interface you've
  495. 14:57been seeing the whole time. Let's build
  496. 14:59a board that reads this folder: the
  497. 15:00areas on the left, key figures at the
  498. 15:02top, approvals with accept, reject,
  499. 15:04query options, and inquiries as a Kanban
  500. 15:06board, in black, silver, and metallic.
  501. 15:08Claude is now building a small website
  502. 15:10that runs directly on your computer. It
  503. 15:12reads the folder, and you can
  504. 15:13immediately see every file that an agent
  505. 15:14creates or modifies here. Of course,
  506. 15:16your design won't look exactly the same
  507. 15:18right away; that's a matter of taste,
  508. 15:20naturally. You can design the whole
  509. 15:21thing however you want it to be. This is
  510. 15:23my design for now. When you click on "
  511. 15:25Accept" here, the board writes the
  512. 15:26decision back to the folder and the team
  513. 15:28continues working. The board wasn't
  514. 15:30finished like this on my first try
  515. 15:32either. I told Claude what I didn't like
  516. 15:34, for example, that the numbers were too
  517. 15:35small or that I wanted the channels as a
  518. 15:37table. Claude then changes it directly,
  519. 15:39and after a few hours, it looks like it
  520. 15:41does here for me, tailored to your taste
  521. 15:43and your design principles, of course.
  522. 15:45And now comes the best part about the
  523. 15:46entire hub. Behind everything you've
  524. 15:48seen in this video is just a folder with
  525. 15:50text files.
  526. 15:51The agents, the knowledge, the
  527. 15:53checklists, the inquiries, and the rules
  528. 15:54are all normal files that you can read
  529. 15:56and change.
  530. 15:57So, you don't need to install anything
  531. 15:59or buy expensive software.
  532. 16:01And because all the knowledge is in the
  533. 16:03folder and not in the model, you can
  534. 16:04also give the same folder to Codex later
  535. 16:06, for example, or any other arbitrary
  536. 16:08LLM.
  537. 16:08It reads the same files and works
  538. 16:10according to the same rules. The longer
  539. 16:12you work with it, the better the hub
  540. 16:13becomes. Every rejection turns into a
  541. 16:15rule, and every week new figures and new
  542. 16:17recommendations are added.
  543. 16:19If you want to build a hub like this, or
  544. 16:21simply download it, and want to learn
  545. 16:22more about AI marketing, feel free to
  546. 16:24join our AI Marketer community.
  547. 16:26You can find all further details there.
  548. 16:28The link is in the video description,
  549. 16:30and you can also simply scan the QR code
  550. 16:32in the top left or right.

About this transcript

This page contains the full transcript of Baue & automatisiere dein Marketing mit Claude Code by Ismajl Pukaj | KI Agentur, generated from the public captions YouTube serves with the video. The transcript has 3,686 words across 550 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.