YouTube2Text

GPT-Live-1 Just Dropped. Voice Agent AGI is HERE! — Transcript

by Brendan Jowett · 4,715 words · 653 segments · language en · Watch on YouTube

Full transcript

  1. 0:00Hey Sophie, we are looking for a new
  2. 0:02accountant.
  3. 0:02>> Okay, I'll check what next steps might
  4. 0:04be helpful here. And what kind of help
  5. 0:06would you need from the team?
  6. 0:08>> Um, really everything. Obviously the
  7. 0:09bookkeeping for our business, any sort
  8. 0:11of tax advisory would be good as well.
  9. 0:13>> Yep, we can help with your bookkeeping
  10. 0:15and both business and individual tax
  11. 0:17returns. And an accountant can review
  12. 0:19your situation. Could I get your name,
  13. 0:20please?
  14. 0:21>> That's OpenAI's new GPT live one model
  15. 0:24released this week in the API running on
  16. 0:26a real phone number. If you're not in
  17. 0:28the voice AI space, this might not mean
  18. 0:30much to you. But I spent three years
  19. 0:32selling AI phone agents to real
  20. 0:34businesses with receptionists and
  21. 0:36customer support systems for over
  22. 0:38$12,000 each. And this right here is an
  23. 0:41AGI moment for voice AI. To compare
  24. 0:44this, here's the previous best voice AI
  25. 0:46setup that we could get. Hey Alex, we
  26. 0:48just recently stopped working with our
  27. 0:50recent accountant and we are looking to
  28. 0:52get a new accountant.
  29. 0:53>> Yeah, we definitely help out with that.
  30. 0:55>> Would you be able to give me a quote at
  31. 0:56all for bookkeeping?
  32. 0:57>> The main thing is that first chat is a
  33. 0:59free console where they look at your
  34. 1:01books and give you a quote based on
  35. 1:03that.
  36. 1:03>> Would we be able to book an appointment
  37. 1:05to learn more?
  38. 1:05>> Yeah, we can set that up. Can I get your
  39. 1:07name?
  40. 1:08>> So there we have it. Doesn't sound too
  41. 1:10bad, responds relatively quick, but it
  42. 1:12still has that robotic quality. And that
  43. 1:14agent is three separate models glued
  44. 1:16together. Speech to text, which turns
  45. 1:18your voice into words, then an AI model
  46. 1:20that writes the reply, then text to
  47. 1:22speech, which reads it out aloud. Now,
  48. 1:24meanwhile, the big AI labs have been
  49. 1:26working on something called speech to
  50. 1:28speech. ChatGPT voice mode, which you
  51. 1:30may have already used, is already speech
  52. 1:32to speech, which is why it sounds so
  53. 1:34human and so good. The model itself is
  54. 1:36actually listening to your voice instead
  55. 1:38of reading a transcript. So, for
  56. 1:40example, when you laugh, cough, or
  57. 1:42pause, it knows exactly what's happening
  58. 1:44and just acts and responds like a real
  59. 1:46person would. The problem forever though
  60. 1:48is that speech to speech has never
  61. 1:49worked properly through the API, and
  62. 1:51especially not on a real phone number
  63. 1:53like we just saw. The latency was always
  64. 1:55really bad, calls kept breaking, and it
  65. 1:58was not something that you could ever
  66. 1:59put in front of a real customer. But now
  67. 2:01that has finally changed with the GPT
  68. 2:03live one API. Let's take another listen
  69. 2:06so you can directly compare it to that
  70. 2:07previous call. Hey Sophie, um what we're
  71. 2:09just looking for a new accountant.
  72. 2:11>> Great, thanks Brendan. What's the best
  73. 2:12number for the team to reach you on?
  74. 2:14>> Um can I just learn more about the
  75. 2:15services? You guys do do bookkeeping
  76. 2:17just before we lock anything in?
  77. 2:19>> Yeah, of course. We're a Melbourne
  78. 2:20accounting firm and we do offer
  79. 2:22bookkeeping as well as tax, BAS, and
  80. 2:26payroll.
  81. 2:27What sort of help would you like to
  82. 2:29start with?
  83. 2:29>> Okay, so there we have it. Today I'm
  84. 2:31going to go through step-by-step how to
  85. 2:33set up this brand new model yourself on
  86. 2:35a real phone number that you or anyone
  87. 2:37else can call and talk to. This is an
  88. 2:39incredible service to sell to other
  89. 2:41companies. These systems can take on
  90. 2:43complex receptionist roles, answer calls
  91. 2:4524/7, never miss a single call, and
  92. 2:47instantly push the data from every call
  93. 2:49into all of the systems that a company
  94. 2:51already uses. And the best part is that
  95. 2:52I'm going to use the new GPT-6 Astro
  96. 2:55model to build the entire voice system
  97. 2:57for us. So you don't need any technical
  98. 2:59knowledge at all to follow along and get
  99. 3:01this fully built. Let's get started.
  100. 3:02Okay, so getting started in building out
  101. 3:04one of these phone agents, the first
  102. 3:05thing that I want you to download is
  103. 3:06going to be Codex. So Codex is OpenAI's
  104. 3:09coding agent and this is what we're
  105. 3:11going to be using to fully create
  106. 3:13everything to do with this voice agent.
  107. 3:14So once again, to reiterate, you don't
  108. 3:16have to be technical at all to follow
  109. 3:17along with this. We're not going to be
  110. 3:18doing any coding or technical setups
  111. 3:20ourselves. We are very simply going to
  112. 3:22be prompting this AI to build the entire
  113. 3:24system for us. It's going to be able to
  114. 3:26literally put the agent on the phone
  115. 3:27number and then all we need to do is
  116. 3:29call it and we'll have the entire thing
  117. 3:30set up for us. So just make sure that
  118. 3:32you've got Codex installed on your
  119. 3:33computer. It is a desktop application
  120. 3:35and as for the pricing of Codex, you can
  121. 3:37get started for completely free on their
  122. 3:39free plan, which will give you limited
  123. 3:41Codex access. But just be aware that you
  124. 3:43might hit the usage limits if you start
  125. 3:45talking to the agent too much and then
  126. 3:46you might be forced to upgrade to their
  127. 3:48go or plus accounts. And also keep in
  128. 3:50mind that these are Australian prices
  129. 3:52and not US dollars, so just keep that in
  130. 3:53mind. Next up, we now need to create an
  131. 3:55account with LiveKit. So, LiveKit is a
  132. 3:57voice agent development platform and
  133. 3:59it's what we're going to be using to
  134. 4:00connect this new OpenAI GPT Live One
  135. 4:03model two. And it's also going to help
  136. 4:04us to connect our phone number into the
  137. 4:06system as well. So, we are going to be
  138. 4:08using a separate platform for the phone
  139. 4:09number, but we do need a platform in the
  140. 4:11middle to connect our model and our
  141. 4:13phone number together and that's going
  142. 4:14to be happening on LiveKit. And this is
  143. 4:16also the exact same stack that OpenAI
  144. 4:18themselves are using. see that they are
  145. 4:20a customer of LiveKit and the
  146. 4:22speech-to-speech AI that is running on
  147. 4:24that ChatGPT app, which you may be
  148. 4:26familiar with, is a system running on
  149. 4:28LiveKit. So, we are getting the best
  150. 4:29system possible right here. And one
  151. 4:31really cool thing about LiveKit is that
  152. 4:33we can actually self-host it as well.
  153. 4:35So, because LiveKit is fully open
  154. 4:36source, we can self-host these voice
  155. 4:38agents and run them locally on our
  156. 4:40computer. But, of course, the downside
  157. 4:41of that is if you turn your computer
  158. 4:43off, then nobody's going to be able to
  159. 4:44call your agent. So, LiveKit does have a
  160. 4:46LiveKit cloud option as well. So, we can
  161. 4:48very easily use this and this is going
  162. 4:50to be my recommended way for doing this
  163. 4:51so that we can turn off our computers
  164. 4:53and these voice agents will still run
  165. 4:55continuously. And this is the setup that
  166. 4:57I would recommend if you are going to be
  167. 4:58building these out for other companies
  168. 5:00to have these systems run in more of a
  169. 5:01secure cloud that you can be confident
  170. 5:03that it's going to be running reliably
  171. 5:05as much as possible. Now, getting
  172. 5:06started with LiveKit, there are a few
  173. 5:08different plans. You can get started
  174. 5:09with the completely free plan and this
  175. 5:11allows you to build one agent and
  176. 5:13connect a phone number as well. So, you
  177. 5:15can get started in building out and
  178. 5:17following along with this entire video
  179. 5:18just by running on this free plan. But,
  180. 5:20of course, if you do scale this up and
  181. 5:22start to build other agents for other
  182. 5:23companies, but of course, they do have
  183. 5:25these other plans. So, if you do have
  184. 5:26multiple agents and multiple needs, you
  185. 5:28will be able to upgrade and handle all
  186. 5:30of that all on LiveKit's cloud. And once
  187. 5:32again, to be clear, these plans are for
  188. 5:34LiveKit cloud. Of course, if you are
  189. 5:36going to be self-hosting it, then you
  190. 5:37will not need to pay a monthly
  191. 5:39subscription for the cloud service
  192. 5:40because you're not going to be using it.
  193. 5:42And just to make things easier, I'll
  194. 5:43have CodeX and LiveKit linked it below
  195. 5:45in the description and pinned at the top
  196. 5:47of the comments down below. So, you do
  197. 5:48have easy access to these platforms.
  198. 5:50But, once you are on this page, the only
  199. 5:52thing that you need to do is essentially
  200. 5:53just create an account. So, you can
  201. 5:55click on this start building button, and
  202. 5:56then you'll just go through a typical
  203. 5:58sign-up process to create an account,
  204. 6:00and then you'll land on a dashboard like
  205. 6:01this. Now, don't get too overwhelmed.
  206. 6:03We're not actually going to be doing too
  207. 6:04much work in this interface itself,
  208. 6:06because we're going to be getting CodeX
  209. 6:07to essentially manage and do all of the
  210. 6:09work here for us. But, we'll come back
  211. 6:10to this in just a minute. The last thing
  212. 6:12that we do need is the actual phone
  213. 6:14number itself, and one of my preferred
  214. 6:16providers for this is Telnyx. They
  215. 6:18essentially just have a big database of
  216. 6:19a bunch of phone numbers in which we can
  217. 6:21purchase. And once we've purchased the
  218. 6:23phone number through Telnyx, we're going
  219. 6:24to be able to get CodeX, and finally, we
  220. 6:26should be able to call this Telnyx phone
  221. 6:28number, and it will connect directly to
  222. 6:30our LiveKit Build voice agent. I also
  223. 6:32just wanted to quickly add that if
  224. 6:33you're just looking to get a US phone
  225. 6:35number, you can actually purchase that
  226. 6:36directly through LiveKit itself. If you
  227. 6:38want to get any international numbers
  228. 6:40outside of the US, then I would
  229. 6:42recommend using Telnyx. In my case, I
  230. 6:44bought an Australian phone number, and
  231. 6:45so that's why I'm using Telnyx in this
  232. 6:47use case. But, if you're just looking to
  233. 6:49get a US phone number, I would say it's
  234. 6:51probably going to be the easiest just to
  235. 6:52purchase it directly through LiveKit,
  236. 6:54and you can skip the entire Telnyx
  237. 6:56setup. So, to get started on Telnyx, you
  238. 6:58will just need to hit the login button,
  239. 6:59sign up for an account, and we'll come
  240. 7:01back in just a minute on actually
  241. 7:02getting this all connected to CodeX so
  242. 7:04it does everything for us. Okay, so we
  243. 7:05are now in the CodeX desktop app. Make
  244. 7:07sure that you have signed in to your
  245. 7:09account, and make sure you are in the
  246. 7:11CodeX option. If you look at the top
  247. 7:12right, make sure you're in CodeX and not
  248. 7:14the chat GPT
  249. 7:15mode. Once we are in this CodeX mode, we
  250. 7:17are now going to create a folder in
  251. 7:19which we're going to be doing all of
  252. 7:20this work in. Now, it should prompt you
  253. 7:21on this main part of the screen to
  254. 7:23create a folder, and if it doesn't, you
  255. 7:25should be able to click into this text
  256. 7:26and select a new folder or create a new
  257. 7:28project, in which will automatically
  258. 7:29create a new folder. In my case, I've
  259. 7:31named the folder LiveKit Demo. And now,
  260. 7:33as simple as it sounds, I am literally
  261. 7:36just going to talk to CodeX with my
  262. 7:37voice and tell it exactly what I'm
  263. 7:39looking for. I'm going to tell it that I
  264. 7:40want a voice agent. I'm going to give it
  265. 7:42the use case for the voice agent to be a
  266. 7:43receptionist, maybe for an accounting
  267. 7:45company or something else. And I'm going
  268. 7:46to tell it that I want the stack to be
  269. 7:48an agent that's built on LiveKit. I want
  270. 7:50that agent to be using the GPT Live 1
  271. 7:52API, and I want to get a phone number
  272. 7:54from Telnyx plugged into this as well.
  273. 7:57And literally by me speaking to the
  274. 7:58computer and telling it to build that,
  275. 7:59it's going to figure out how to build
  276. 8:00everything for us, and I can sit back,
  277. 8:02relax, and just watch it work. Hey
  278. 8:03Codex, I would like you to build a voice
  279. 8:05agent. I would like to build a
  280. 8:07receptionist system for a plumbing
  281. 8:09company. If you could build this system
  282. 8:11out using LiveKit, using the new GPT
  283. 8:14Live 1 API to power this, that would be
  284. 8:16great. I would also like you to connect
  285. 8:18into Telnyx. I've got a phone number on
  286. 8:19Telnyx. Uh if you can connect into that
  287. 8:21account as well, uh to build the entire
  288. 8:23setup. And so, literally from that
  289. 8:25incredibly basic prompt, laying out step
  290. 8:27by step what I want it to build, Codex
  291. 8:29is going to be able to go on the
  292. 8:30internet, find LiveKit, it's going to be
  293. 8:32able to look through that documentation
  294. 8:33to look at all of the LiveKit APIs, and
  295. 8:35it's going to be able to look at that
  296. 8:36code from LiveKit and understand exactly
  297. 8:38what it needs to do to actually build
  298. 8:40this voice agent on my LiveKit account.
  299. 8:42One thing that I will quickly add to my
  300. 8:43prompt is that I want it to be using my
  301. 8:45LiveKit cloud account. We don't want it
  302. 8:46to accidentally build it and run it
  303. 8:48locally on my computer. So, I will just
  304. 8:50tell it that I want it to build the
  305. 8:51agent into my LiveKit cloud account. And
  306. 8:53just to make sure, can you put this
  307. 8:55agent in my LiveKit cloud account? I
  308. 8:57don't want this to be running locally. I
  309. 8:59want you to connect into my LiveKit
  310. 9:01cloud account and build the agent there.
  311. 9:02>> Just another quick note, in your prompt,
  312. 9:04I'd actually recommend specifically
  313. 9:06typing LiveKit CLI, just telling it in
  314. 9:09the prompt to install the LiveKit CLI
  315. 9:11specifically. In my prompt, I just said
  316. 9:13LiveKit, but if you want it to run as
  317. 9:15smoothly as possible, just type LiveKit
  318. 9:17CLI. That will then pick up on exactly
  319. 9:20what you're looking to install. In my
  320. 9:22case, mine was already sort of
  321. 9:23configured, um so it was able to figure
  322. 9:25that out. But to ensure that it runs as
  323. 9:27smoothly as possible, just type LiveKit
  324. 9:29CLI. It's going to be able to install
  325. 9:32and understand exactly what we need to
  326. 9:33do just by doing that.
  327. 9:34>> Okay, I have now sent that message off
  328. 9:36to Codex. We're going to about to sit
  329. 9:37back, relax, watch it work, and it is
  330. 9:40just going to tell us step-by-step
  331. 9:41everything that we need to do. And if we
  332. 9:43ever run into any issues, if you are
  333. 9:45following along with this, you can just
  334. 9:46tell CodeX what the issue that you're
  335. 9:48having is. You can just ask it how to
  336. 9:50find something that it told you to find.
  337. 9:52We can see now that CodeX is asking us
  338. 9:53about the plumbing company that we're
  339. 9:55building the voice agent for. You can
  340. 9:56just make something up. This is just for
  341. 9:58a demo. Okay, now just quickly, we can
  342. 9:59see that the agent actually found a
  343. 10:01LiveKit project from one of my other
  344. 10:03folders that it was able to connect
  345. 10:04into. So, in your case, obviously, if
  346. 10:06you're setting up brand new, this is not
  347. 10:07going to happen for you. But in your
  348. 10:08case, what should happen is that it
  349. 10:10should tell you that it requires your
  350. 10:12account to be authenticated. And it is
  351. 10:14more than likely going to ask you to
  352. 10:16provide some kind of an API key from
  353. 10:18LiveKit. And we can find these keys by
  354. 10:20going into the LiveKit web dashboard,
  355. 10:22heading into the settings section, which
  356. 10:24is just if we click out of here at the
  357. 10:26bottom in settings, and then we click on
  358. 10:28API keys. This is where we can create a
  359. 10:30key in the top right. And we can create
  360. 10:31this key in the top right with the
  361. 10:33create key button. We can name this
  362. 10:35whatever and once it's created, I can
  363. 10:37click into this one that I just created.
  364. 10:39We will then get back quite a few
  365. 10:40details, and it's more than likely that
  366. 10:42CodeX will ask us for some of these
  367. 10:44details. And you just need to follow
  368. 10:45along, take the API keys and the secret
  369. 10:48and all the variables that it might be
  370. 10:49asking for, and provide it directly to
  371. 10:51the agent. Now, I don't recommend
  372. 10:53directly copying and pasting these API
  373. 10:55keys into the chat itself, and CodeX
  374. 10:58will more than likely not recommend you
  375. 10:59to do that either. But it will give you
  376. 11:01a step-by-step on how to do it properly.
  377. 11:03Essentially, we just have to paste these
  378. 11:05keys into a secrets file that CodeX has
  379. 11:07created for us. This way that the keys
  380. 11:09don't actually go through any AI model,
  381. 11:11and we're just putting the keys in where
  382. 11:12they should go. And once we've done
  383. 11:13this, you should get that authentication
  384. 11:15into your account, and you should see it
  385. 11:17come up with your account name as well
  386. 11:19just to confirm that it is properly
  387. 11:21connected. Now, another thing to note,
  388. 11:22whilst this is building, you can see
  389. 11:24here it says your OpenAI key has access
  390. 11:26to GPT live one. Now, CodeX might
  391. 11:29already have access to your API key if
  392. 11:31you've used it in any previous projects.
  393. 11:33But, if it doesn't have access to that
  394. 11:35API key yet, or it just seems like it
  395. 11:37hasn't found that API key and is at any
  396. 11:38point asking you for that OpenAI API
  397. 11:41key, you will, of course, just need to
  398. 11:43head to OpenAI's API console. So, you
  399. 11:45will, of course, need an account with
  400. 11:46OpenAI, the same one that you're using
  401. 11:48for Codex is fine. So, on the openai.com
  402. 11:51page, you can just go to login and then
  403. 11:52click into the API platform. And then,
  404. 11:54you should see at the bottom left of
  405. 11:55this dashboard, there's an API keys
  406. 11:57section. Inside of here, you can then
  407. 11:59very easily create the API key, copy it,
  408. 12:01and then provide it to the project where
  409. 12:03needed. Okay, so our Codex agent is
  410. 12:05almost complete at building out our
  411. 12:07entire voice agent. And once it's done,
  412. 12:09I'm going to give it a live call, so we
  413. 12:10can listen to exactly how good it
  414. 12:11sounds. But, before we do that, if you
  415. 12:13could just quickly check that you are
  416. 12:14subscribed by scrolling down below and
  417. 12:16clicking on the subscribe button. I
  418. 12:18recently discovered that 90% of you
  419. 12:19watching right now are not subscribed to
  420. 12:21the channel at all, which is pretty
  421. 12:23insane, and I would really appreciate it
  422. 12:25if you were subscribed. And make sure to
  423. 12:26keep watching as we're now going to
  424. 12:27finish up building this voice agent.
  425. 12:29Then, I'm going to give it the live call
  426. 12:30and really stress test it, so we can see
  427. 12:32exactly how well it sounds from the very
  428. 12:34little work that we've put into actually
  429. 12:36building it. And then, after this, I'm
  430. 12:37also going to get Codex to build an
  431. 12:38actual tool call into our voice agent,
  432. 12:41so it isn't just a conversation back and
  433. 12:43forth, but we can get our voice agent to
  434. 12:44actually do something for us. So, that
  435. 12:47might be in the form of booking
  436. 12:48appointments or sending information to
  437. 12:50some kind of a CRM system. Okay, so
  438. 12:52Codex is almost complete at building out
  439. 12:54the voice agent. The one The last thing
  440. 12:55that we do need to do is give Codex
  441. 12:57access to our Telnyx account. So, once
  442. 12:59again, Telnyx is just used for our phone
  443. 13:01numbers. You will just need to click the
  444. 13:02login button at the top right and create
  445. 13:04an account if you don't have one
  446. 13:05already. Once you've created that
  447. 13:07account, you should land on a dashboard
  448. 13:08that looks similar to this. All that we
  449. 13:10need to do is go to left-hand side and
  450. 13:12click on the voice suite. Inside of this
  451. 13:14section, underneath the numbers tab on
  452. 13:15the left-hand side, we can then click
  453. 13:17into phone numbers. Now, this is where
  454. 13:19all of our existing phone numbers are
  455. 13:21going to be stored. You can see one of
  456. 13:22my Australian phone numbers, in which is
  457. 13:25on my dashboard. And so, all you need to
  458. 13:27do is to click on buy number in the top
  459. 13:29right corner. And then, you're going to
  460. 13:30be able to buy any phone number that you
  461. 13:32want through Telnyx, and it will just
  462. 13:34store itself on our account. And then,
  463. 13:36we're just going to connect Code X to
  464. 13:37our Telnyx account, and it's going to
  465. 13:38set up everything for us. So, if you are
  466. 13:40looking for a US phone number, you can
  467. 13:42just click on search numbers, and just
  468. 13:44make sure that the US is selected. And
  469. 13:46you'll see that there are hundreds of
  470. 13:48phone numbers that we can get,
  471. 13:49especially the US ones. There is a lot
  472. 13:51of them, and they're about a dollar a
  473. 13:52month to use. Once you do have your
  474. 13:54phone number, there isn't anything else
  475. 13:56to do. Just make sure to follow through
  476. 13:57with any instructions that it gives you
  477. 13:59to set up that number. But otherwise,
  478. 14:00just go to the top right corner in this
  479. 14:02sort of profile icon. You will, of
  480. 14:04course, need to add some balance into
  481. 14:06your account. It does cost money to make
  482. 14:08these calls, of course. And so, you can
  483. 14:10add in 5, 10, 15 dollars just to make
  484. 14:12sure that you can do some demo calls.
  485. 14:14But for reference, it will cost a couple
  486. 14:16of cents for every single call that you
  487. 14:18make. So, this will last quite a while.
  488. 14:20And also, in this section, just make
  489. 14:21sure to click into API keys. And then,
  490. 14:24we just need to create a brand new API
  491. 14:25key. And once again, Code X should be
  492. 14:27giving us the step-by-step instructions
  493. 14:29on where to place this API key in the
  494. 14:31files it's given us. And once you've
  495. 14:33done that, that is literally all that we
  496. 14:34need to do. Code X will now have full
  497. 14:36access to our Telnyx account, and it's
  498. 14:38going to be able to get that phone
  499. 14:40number and fully connect it to Live Kit
  500. 14:42without us having to do any of the
  501. 14:43technical work ourselves. Okay, so this
  502. 14:45is now done. We can see built and
  503. 14:48deployed Harbor Plumbing's AI
  504. 14:49receptionist to your Live Kit Cloud
  505. 14:51account powered by GPT Live 1. So, this
  506. 14:54is all ready to go, and it's even
  507. 14:56telling us that this is the phone
  508. 14:57number, in which is the one that we just
  509. 14:59saw on my Telnyx account. And it is
  510. 15:01supposedly all set up and ready to go.
  511. 15:03>> Hi there. This is Mia, Harbor Plumbing's
  512. 15:05AI demo receptionist. This call may be
  513. 15:07recorded. How can I help with your
  514. 15:09plumbing issue today?
  515. 15:11>> Hey Mia. Um, we've got actually a
  516. 15:13leaking shower tap, uh, leaking shower
  517. 15:15head. I was wondering if that's
  518. 15:17something that you guys, uh, do any
  519. 15:19services for, and if you could provide
  520. 15:20us with a rough price estimate for
  521. 15:22something like that, that would be good
  522. 15:24as well.
  523. 15:27>> Yep, a leaking shower head is something
  524. 15:29we help with. We cover leaking taps,
  525. 15:32shower heads, and pipe leaks. Pricing
  526. 15:34can vary depending on the job and what's
  527. 15:36found on site though. So, I can't
  528. 15:38promise a quote. What's happening with
  529. 15:40the leak? Like is it steady dripping or
  530. 15:42more of a constant flow?
  531. 15:44>> Yeah, [snorts] it's more of a constant
  532. 15:45flow. There is sort of a flood that's
  533. 15:47building up at the bottom, so it is a
  534. 15:48bit concerning.
  535. 15:52>> Mhm, I see. Thanks for telling me.
  536. 15:55Is there any active flooding spreading
  537. 15:56beyond the shower area or any sewage
  538. 15:58coming up?
  539. 16:00>> Uh yeah, it's sort of flowing out of the
  540. 16:02bathroom now and almost down the stairs.
  541. 16:04So, we do need to get this checked on
  542. 16:05pretty quickly.
  543. 16:10>> Yeah, that's understandable. Thanks for
  544. 16:12letting me know. I'm just checking on
  545. 16:14the next step.
  546. 16:15Please move everyone away from the water
  547. 16:17and stairs and contact local emergency
  548. 16:19services on triple zero if there's any
  549. 16:21immediate danger. Don't enter unsafe
  550. 16:23areas. Once you're safe, I can take down
  551. 16:26the details for this demo inquiry. Is
  552. 16:28anyone in immediate danger right now?
  553. 16:31>> [snorts]
  554. 16:31>> No, everyone's safe. Uh we just wanted
  555. 16:32to get somebody to uh come out and help
  556. 16:34us.
  557. 16:36Okay, so there we have it. Obviously, it
  558. 16:37now works and it's swapped over to that
  559. 16:39plumbing type receptionist system. Of
  560. 16:40course, there will always be refinements
  561. 16:42when it comes to the actual
  562. 16:43conversational aspect in terms of in
  563. 16:45terms of what the agent is or isn't
  564. 16:47capable of doing. So, you may have heard
  565. 16:49that it was telling me that it was a
  566. 16:50demo agent and it wasn't able to do
  567. 16:51certain things. So, Kore AI has
  568. 16:53obviously prompted it to be a demo
  569. 16:54agent. What I could do is just tell it
  570. 16:56to remove the demo agent part of it,
  571. 16:57obviously. But, hopefully, you get the
  572. 16:59idea. In just very little work, we've
  573. 17:00been able to build that full-on voice
  574. 17:02system. And if we spend some actual time
  575. 17:04into this, giving it some proper
  576. 17:06requirements for a legitimate business's
  577. 17:08series of call reasons, we're going to
  578. 17:09be able to build the system pretty good
  579. 17:11to be able to handle a wide range of
  580. 17:13calls. But, of course, these agents are
  581. 17:14useless if we are just talking back and
  582. 17:16forth with them and we're not doing any
  583. 17:18actual work. And so this is where tool
  584. 17:20calls come into place. So a tool call
  585. 17:22might be booking an appointment or
  586. 17:23sending some information into a CRM
  587. 17:25system. This could also be pulling
  588. 17:27information out of a CRM system into the
  589. 17:29voice agent as well. And so I'll quickly
  590. 17:31showcase exactly how we can set up tool
  591. 17:33calls very, very easily through this as
  592. 17:35well. It is as simple as me telling
  593. 17:37Codex to set up a function call within
  594. 17:39this voice agent. We don't have to use
  595. 17:41any secondary or third-party platforms
  596. 17:43to build any automation workflows
  597. 17:45anymore. We just need to tell Codex to
  598. 17:46literally build out the entire workflow.
  599. 17:48It's going to be able to connect into
  600. 17:50any external apps that have APIs that
  601. 17:52are readily available and build
  602. 17:53everything out for us. Hey Codex, I
  603. 17:55would now like you to build a function
  604. 17:57call into this voice agent in which is
  605. 17:59able to look up the live price of gold.
  606. 18:02So if you could if somebody is asking
  607. 18:03for the live price of gold, this should
  608. 18:05trigger the function call to hit a
  609. 18:07public API endpoint for the live price
  610. 18:09of gold, pull that price back into the
  611. 18:12conversation and read it out to the user
  612. 18:14so that they are getting the real
  613. 18:15up-to-date price and it's not some
  614. 18:17outdated price. And now I'm going to
  615. 18:18send that off and that's all I need to
  616. 18:20do. Okay, Codex is now telling us that
  617. 18:22this is now complete. The tool call has
  618. 18:23been added to our voice agent. I'm going
  619. 18:25to give it a call and see if this works.
  620. 18:27>> Hi there. You're through to Mia, Harbor
  621. 18:29Plumbing's AI demo receptionist. This
  622. 18:31call may be recorded. How can I help
  623. 18:33today?
  624. 18:34>> Hey Mia, I was just wondering if you'd
  625. 18:36be able to give me the live price of
  626. 18:38gold at the moment.
  627. 18:42>> Sure thing. I'll check the latest gold
  628. 18:44price for you.
  629. 18:48Okay, the latest spot quote from Gold
  630. 18:50API is 4,314
  631. 18:53US dollars and 50 cents per troy ounce,
  632. 18:56updated about 30 seconds ago.
  633. 18:58Is there anything else I can help with
  634. 18:59today?
  635. 19:00>> Perfect, that's it. Okay, and there we
  636. 19:02go. You can see that it was able to
  637. 19:03obviously do that function call to get
  638. 19:05the live price of gold. It actually said
  639. 19:07the name of the API on there. Obviously
  640. 19:08you'd fine tune that. It's not really a
  641. 19:10use case for a plumbing business. But
  642. 19:11just to prove the point, we can do tool
  643. 19:13calls on this speech-to-speech model, as
  644. 19:15well. I hope you've enjoyed the video.
  645. 19:16Let me know down below in the comments
  646. 19:17what you think of this new voice model.
  647. 19:20And, if you want to push GPT-6 Astra
  648. 19:22even further, I did a full-on video
  649. 19:24building out websites with this new
  650. 19:26model. And, honestly, the results have
  651. 19:27blown my mind as to how good it is at
  652. 19:29design now. So, I'd highly recommend
  653. 19:31checking this video out.

About this transcript

This page contains the full transcript of GPT-Live-1 Just Dropped. Voice Agent AGI is HERE! by Brendan Jowett, generated from the public captions YouTube serves with the video. The transcript has 4,715 words across 653 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 translate the transcript into 100+ languages.

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, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.