YouTube2Text

I Built a Jarvis That Runs My YouTube Channel — 2M Views in 2 Weeks — Transcript

by Lume Media · 1,374 words · 212 segments · language en · Watch on YouTube

Full transcript

  1. 0:00You've seen these videos. They're
  2. 0:01everywhere right now. Every feed, every
  3. 0:04day, someone building their own Jarvis.
  4. 0:07Now, one guy built a Jarvis that reads
  5. 0:08his emails, and another one built a
  6. 0:10Jarvis that turns on the lights. And
  7. 0:12some people, just like saying Jarvis,
  8. 0:15good morning and feeling like, you know,
  9. 0:17Tony Stark for 5 seconds. So, respect.
  10. 0:20Look, I get it.
  11. 0:22But here's the thing that nobody is
  12. 0:24asking. How does your Jarvis make you
  13. 0:27money?
  14. 0:29Huh?
  15. 0:29Because a robot that reads your emails
  16. 0:31is cool. A robot that pays your rent,
  17. 0:34that's business.
  18. 0:35So, we asked the same question, and
  19. 0:38[music] we answered it. And we built a
  20. 0:40Jarvis that runs a YouTube channel. It
  21. 0:42finds the stories, it writes the
  22. 0:44scripts, it edits the videos, it names
  23. 0:47them, and it checks its own work.
  24. 0:50And
  25. 0:51in a little over 2 weeks, that channel
  26. 0:54passed 2 million views. Yeah. So, in
  27. 0:57this video, I'm going to show you what
  28. 0:59this Jarvis actually does and how you
  29. 1:01can build one as well. This is Arvin, a
  30. 1:04history channel, true stories of
  31. 1:07ordinary people in the Second World War,
  32. 1:10and a marine with two machine guns, a
  33. 1:1314-year-old girl with a bicycle, and
  34. 1:15then a soldier who laughed when he was
  35. 1:18captured.
  36. 1:19The first video
  37. 1:21went up on October 25th.
  38. 1:24Zero subscribers, zero views, there was
  39. 1:27zero anything.
  40. 1:2810 days later, this one crossed [music]
  41. 1:30a million views. Then 3 days after that,
  42. 1:33this one started climbing too, and it
  43. 1:35passed 800,000
  44. 1:38now.
  45. 1:39So, together those two did almost 2
  46. 1:41million views in about 10 days, and the
  47. 1:43whole channel over 2 million. Now, I am
  48. 1:47not going to lie. I did not expect that
  49. 1:49either. I mean, we expected maybe there
  50. 1:51would be a few thousand views and a and
  51. 1:53a lesson in humility.
  52. 1:55And by the way,
  53. 1:57the full step-by-step lessons on how we
  54. 2:00these Arben videos are already inside
  55. 2:02our community, schooled by Loom. So,
  56. 2:05just go to the link in the description.
  57. 2:07So, today I will teach you how to edit a
  58. 2:09war documentary, but today I'm going to
  59. 2:11show you the thing that makes them. So,
  60. 2:14here's a mistake that I think most
  61. 2:15people make with their Jarvis.
  62. 2:19They built it like a toy. They can ask
  63. 2:21what cool things can it do? And then the
  64. 2:24answer is a lot of cool things. It can
  65. 2:26open apps. It can talk in British
  66. 2:28accent. It can tell you the weather. You
  67. 2:30know, which to be fair, your phone was
  68. 2:32already doing at that time.
  69. 2:34But we asked a different question, not
  70. 2:37what can it do,
  71. 2:39but what job can it take?
  72. 2:43Because a YouTube channel isn't one job.
  73. 2:46It's like five jobs wearing a trench
  74. 2:48coat. You know, somebody has to find the
  75. 2:49story. Someone has to write it. Someone
  76. 2:51has to edit it. Somebody has to name it
  77. 2:54so people will actually click. And then
  78. 2:56somebody has to check all of it before
  79. 2:58it actually goes out. So, normally
  80. 3:01there's a team or it's one very, you
  81. 3:04know, tired team.
  82. 3:05So, we gave those jobs to Jarvis.
  83. 3:09So, let me show you what that looks
  84. 3:10like. So, one video from nothing to
  85. 3:13ready. So, it starts with a topic and I
  86. 3:15didn't pick it. Jarvis does. So, every
  87. 3:18day it goes through the history niche,
  88. 3:21what's getting views, what's getting
  89. 3:23ignored, what everybody has already done
  90. 3:2610 times. And then it comes back with a
  91. 3:29story that nobody has told properly yet.
  92. 3:31And then the reason that it thinks
  93. 3:32people will click on it. Then it writes
  94. 3:35the script. Not a Wikipedia article, a
  95. 3:38story, scenes, tension, a question in
  96. 3:42the first 30 seconds that you only get
  97. 3:44the answer to at the end. And every
  98. 3:46single fact in that script has a source
  99. 3:49next to it. No source, the fact doesn't
  100. 3:52go in.
  101. 3:53Then it edits the video.
  102. 3:56So, the maps, the motion, the archive
  103. 3:59photos, the cults, all of it inside
  104. 4:03After Effects.
  105. 4:04Then it packages it. So titles, not one,
  106. 4:09a whole batch each built on a formula
  107. 4:12that already worked on the channel, the
  108. 4:14description, the chapters, the tags,
  109. 4:17even the question for the comments.
  110. 4:20And then my favorite part, it checks
  111. 4:24itself.
  112. 4:26Because there's not one Jarvis in there.
  113. 4:28There are two. One makes the video, the
  114. 4:31other one, whole video is to find out
  115. 4:33what's wrong with it. So wrong date, a
  116. 4:36fact without a source, a cut that's just
  117. 4:37too long, a length that's dead, a title
  118. 4:40that has already been taken. And it does
  119. 4:43find things. For one of our videos, we
  120. 4:45came up with this
  121. 4:47title conversation that we loved it.
  122. 4:50We were sure that it was original. The
  123. 4:51checker went through the competition and
  124. 4:53came back with this. Our original title
  125. 4:56was word for word another channel's
  126. 4:58title with 648,000
  127. 5:00views.
  128. 5:02So we didn't copy it on purpose. You
  129. 5:04know, one of us had seen it once, it got
  130. 5:06stuck in their head and it came out as
  131. 5:08our idea. So that's how human brains
  132. 5:11work. Jarvis just doesn't care whose
  133. 5:13idea it was. All it does is checks.
  134. 5:16And when everything passes on my screen,
  135. 5:20there is this.
  136. 5:23A finished video with a title, a
  137. 5:25description, everything. And the only
  138. 5:27thing left for me to do, the only thing
  139. 5:31is to press a button.
  140. 5:33That's my job now. Yeah. I'm the guy who
  141. 5:37presses the button.
  142. 5:39Now, how do you get a bunch of AI tools
  143. 5:41to act like one employee instead of five
  144. 5:44strangers? And I'm telling you, it's not
  145. 5:45the voice. It's not the glowing orb.
  146. 5:48It is memory.
  147. 5:51So every time Jarvis makes a video, it
  148. 5:54remembers what worked. Every time it
  149. 5:56gets something wrong, the fix becomes a
  150. 5:59rule. So it doesn't make the same
  151. 6:02mistake twice, which honestly is better
  152. 6:04than most people that I know. So it
  153. 6:06including me. So that is how a brand new
  154. 6:09channel hit a million views on just its
  155. 6:12third video.
  156. 6:14So we didn't get smarter. Jarvis did.
  157. 6:18So how do you build one? Well, it's
  158. 6:21simpler than the Iron Man movies make it
  159. 6:24look. It it it's four layers. Layer one,
  160. 6:27the brain. We use Claude code. It's the
  161. 6:30thing that actually thinks, it writes,
  162. 6:31and it runs the tools.
  163. 6:33Layer two, the memory. This is the part
  164. 6:36everyone skips, and it's the most
  165. 6:37important.
  166. 6:38Files with your rules, what your channel
  167. 6:41is about, what worked, what failed, your
  168. 6:43style, every mistake became a new rule.
  169. 6:47Layer three, the hands. These are the
  170. 6:49connectors. They let your Jarvis
  171. 6:51actually do things instead of just
  172. 6:52talking to them about it.
  173. 6:54Open the browser, work in the After
  174. 6:56Effects, work in Blender, read a
  175. 6:59document. Adobe even has its own
  176. 7:01official connector now for Photoshop,
  177. 7:03Premiere, and the rest of Creative
  178. 7:05Cloud. Layer four, and this one's
  179. 7:07optional, the voice. So if you want the
  180. 7:10full movie experience with the glowing
  181. 7:13orb and the will do, sir, there are open
  182. 7:16source projects on GitHub that put a
  183. 7:18voice on top of Claude code. And some of
  184. 7:20them even have the holographic Iron Man
  185. 7:23interface.
  186. 7:24But here's the thing that I really want
  187. 7:26you to take from this video.
  188. 7:29The voice isn't the point. [music] The
  189. 7:31orb isn't the point. The point is the
  190. 7:34job.
  191. 7:35So find one thing that makes you money
  192. 7:37or saves you money and give that job to
  193. 7:40Jarvis, a channel, editing for clients,
  194. 7:43content for a small business, research
  195. 7:45somebody should pay for. Start with one
  196. 7:47job.
  197. 7:48Teach it your rules. Let him make
  198. 7:50mistakes and then turn every mistake
  199. 7:53into a rule. And that's it.
  200. 7:55That's the whole secret. Ours learned to
  201. 7:59run a channel that did 2 million views
  202. 8:00in the first week. If ours could do it,
  203. 8:03I'm telling you yours could too. So
  204. 8:06right, Tony?
  205. 8:07So all the prompts that we used to build
  206. 8:09our Jarvis, the brain, the memory files,
  207. 8:11the setup, they're already in our
  208. 8:13community. So just go ahead, link in the
  209. 8:16description.
  210. 8:17And if you want to see what we build
  211. 8:20next, go ahead and subscribe. Jarvis is
  212. 8:23watching the numbers, and so am I.

About this transcript

This page contains the full transcript of I Built a Jarvis That Runs My YouTube Channel — 2M Views in 2 Weeks by Lume Media, generated from the public captions YouTube serves with the video. The transcript has 1,374 words across 212 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.