0:00Jordan Metzner opens with a Veo 3.1 fast clip, and Sam Nadler says the episode covers an internal tool built on the new Sora API, the workflow automation tool n8n and the week's AI news.
2:19Jordan Metzner says Sora 2 is both OpenAI's video social app and its video model available by API. His tool makes horizontal or vertical videos, sets length, takes a reference image of an exact size and can remove the watermark.
4:54Jordan Metzner runs the same capybara prompt through Sora 2 and Google's Veo 3.1 fast, which he calls pretty good and inexpensive, and says the Veo 3.1 capybaras look more real. Sam Nadler agrees Veo 3.1 looked better.
9:25Jordan Metzner says you can send one prompt to Sora and Veo several times and combine the best clips into trailers or commercials. He calls it a huge threat to Hollywood but says it democratizes content creation.
10:43Jordan Metzner says n8n is a platform for agentic workflows, not that different from Zapier and Make, that connects tools like Gmail, Google Sheets and Airtable with APIs like OpenAI and Perplexity. Sam Nadler says it has a learning curve.
13:08Jordan Metzner's workflow researches Russell 2000 companies with Perplexity, gets contact emails from Prospeo, sends structured data to ChatGPT, logs it in Google Sheets and saves a custom email to the CEO as a Gmail draft.
15:34Sam Nadler says prompting Claude produces workflow JSON you import into n8n, leaving the last 10% of connecting. Jordan Metzner imports a Claude-built podcast outreach workflow and calls ChatGPT's agent workflow tool very rudimentary by comparison.
17:54Jordan Metzner says chat is becoming the new search interface, and if shopping searches move to ChatGPT, Amazon could lose conversion and may have to pay ChatGPT for traffic the way it has paid Google.
21:41Jordan Metzner says he has never seen software companies grow revenue this fast and his business could not operate without OpenAI or Anthropic, but 2.5 times in a year may be hard given data center buildout.
24:52Jordan Metzner says Apple has not announced premium M5 chips, so the top RAM MacBook Pro still needs an M4 Pro, his team does not run LLMs locally, and he calls it a so so launch.
Sora 2 API vs Veo 3.1, and n8n Workflows Built by Claude: the key questions from this episode, answered with quotes and timestamps.
Jordan Metzner [0:00]: This is Veo 3.1 fast. We can watch this now.
Sam Nadler [0:21]: You built this week. Hey, everyone. And welcome to another episode of Built this week, the podcast where we share what we're building, how we're building it, and what it means for the world of AI and startups. I'm Sam Nadler, cofounder here at Ryz Labs. And each and every week, I'm joined by my friend, cohost, business partner, Jordan Metzner. What's up, Jordan? How are you doing today?
Jordan Metzner [0:45]: Yo, Sam. Happy to be back. Episode 17. Can you believe we've made it this far? Lots going on in AI. Nothing has slowed down. Each week has gone, seems to be compounding on the week previous. Lots of AI news, lots of new AI technology, and, yeah, lots to talk about today. So super excited to get started.
Sam Nadler [1:03]: Yeah. We have a full docket, but before I jump into it, please don't forget to like and subscribe. We have new episodes of Built This Week released every single Friday, and you can find them on your favorite platform, Apple Podcasts, YouTube, and Spotify. And this week, we're gonna cover a tool we built using the new Sora API to give to our teams to, have the latest and greatest, AI video generation tool at their fingertips. We're gonna jump into a kind of hot automation tool that, people have been using to, AI based automation tool called n8n that people have been using to automate all different types of tasks, whether it's in sales or marketing, IT, etcetera. And then finally, the latest and greatest, AI news of the week. There's always something new and, valuations and growth seem bigger and every single week. So, very exciting docket. So let's start with, number one, this new Sora tool you built using the, newly released API just, I think, less than a week ago.
Jordan Metzner [2:18]: Yeah. So, Sora two has made a quite a storm in regards to some of the viral videos it's made, and, I think there's some common nomenclature, maybe some wording that gets a little people that gets people a little bit confusing, as the name Sora has been used for a lot of different things. So Sora or Sora two is the name of the app by OpenAI, which is a new video generation social network, but it's also the name of their new model for video generation, which you can obtain via their API. And on the app, I saw they released some new features today or yesterday regarding storyboards and a bunch of new features, but today, we're gonna show off the Sora two API. And the reason I think this is a little more interesting is, obviously, you can just access it by building some app yourself and being able to develop any type of content. And this is probably one of the first ways you can start to kinda make movies. They have a lot of continuity opportunities, so they call it like a reference image, but you upload an image and then it'll make content based on the image. The descriptions are really good. Anyway, yeah, let me just jump right into it. So I built a little Sora two tool, and basically, I can generate videos both horizontally and vertically.
Jordan Metzner [3:34]: I can choose kind of my time length. I can add a reference image, and then I can also remove the Sora watermark as well. When I add a reference image, it has to be this exact size. Otherwise, it doesn't work. And sometimes, like, if you make too long of a post of an input, then sometimes the API fails, and I try to put some stop gaps in, but sometimes it still happens. But I like capybara videos. I think they're pretty funny, so I thought maybe I could make some, capybara built this week video. So let's do two capybaras talking on the built this podcast, and they say like and scribe.
Sam Nadler [4:15]: And something unexpected and funny happens at the end.
Jordan Metzner [4:18]: Yeah, that's a great one. So yeah, I think, we've seen that that works like pretty well, and we've gotten some good results. So let's say, and something funny and happens. Alright. So let's put that at twelve seconds. That way, we can kinda get a full length kinda sort of video.
Sam Nadler [4:33]: One small feature that I like is the remove the watermark as an option. Obviously, if you're creating content for your own channels, you wanna remove the Sora watermark, and I know there's, like, a ton of different tools you can do this, but just having it directly built in feels like a great little touch.
Jordan Metzner [4:53]: Yeah. I mean, I don't ever really want a watermark. Exactly. Let's take this post, and what I wanted to show you in just like a little kind of but wait, there's more. Google yesterday released Veo 3.1, and using our standard Ryz Labs video generator, Veo 3.1 fast is is pretty good and pretty inexpensive. So I was thinking I could put the prompt in here of what we just talked about with the two capybaras, and I could generate it with this, and then we can see what the output is here, and then we can compare. So that way through my video history tool, it should generate these videos. Let's see if our Sora one is done. Might take a few minutes here. And then we can kinda compare the results of, the Sora twelve second video versus the new Veo 3.1 video, which also does referential. So overall, I've seen like the impacts went pretty awesome. Like you saw, as we talked about last week, me skydiving and, you doing all these cool fun things. But now we're starting to see how we can use these models to produce, content at scale. So whether that's social media content, whether that's video trailers, the quality of the content is becoming so high quality that you can't really tell that it was AI generated.
Sam Nadler [6:11]: And I know you built this quickly to highlight the Sora two API, but is there any reason you wouldn't just connect it to the kinda wider Ryz Labs video generation tool just as another option?
Jordan Metzner [6:24]: Yeah. I probably will and, like, kind of deprecate the other one. I think when Sora two came out, it was so new, and I didn't expect kind of these other models to follow-up so quickly, and so I just built like a quick sort of tool. But now that I've got it integrated and everything, I can just add it back into this generation tool. Here is the capybara. This is Veo 3.1 fast. We can watch this now. Oh, nice. They got some bananas. Okay. That's a pretty awesome video. I thought it was pretty funny and cute. Okay. Great. We got our ad. Here it is. Okay. Cool. And let's see. Playback Welcome back to Built Ring the Welcome back to Built This Okay. Cool. So it's okay. You know? I mean, I'll download it. We'll add it to our assets, and, we'll share it back. But honestly, I thought the Veo 3.1 was better. Just running back to it. Anyway, pretty awesome. So I'll I'll share all these videos with our editors, so we can chop them in and play them back in full playback, and kind of compare the Sora to 3.1, while we're just kind of killing a lit final time. I'm gonna try the same exact prompt one more time with Veo 3.1, and maybe, something even more funny and unexpected happens. But overall, yeah, I would love to hear your feedback on what you think of the quality of the video. I mean, I don't know.
Jordan Metzner [8:05]: I think the Capybara's look super real. That would take a ton of CGI to make something like that, look as real. On the Sora one, they don't look as real, but still, I mean, pretty good, I would say.
Sam Nadler [8:15]: Yeah. I just think it highlights the, a, the competition, the release velocity. Two weeks ago, all everyone was talking about was Sora. A few months before that, I don't remember the exact timeline, but not too long before, it was all Veo 3. Now Veo 3.1 just came out. We did a quick side by side test. Veo 3.1 looks better, but it's just gonna be nonstop, and these are such great tools getting better and better every week. It's really hard, getting for me, especially with the Sora app and my own likeness, getting closer and closer to being, hard to discern if this is real or fake. So just a really exciting space in the AI video generation vertical.
Jordan Metzner [9:00]: I wanna say kinda one or the last two points here. So we'll, we'll produce one more. Here's one more that capybara is using Veo 3.1. He has narcolepsy. Hilarious. Yeah. Okay.
Sam Nadler [9:20]: It seems oddly accurate for you.
Jordan Metzner [9:25]: Okay. So anyway, we'll share these. We'll get these played back in full, and we can kinda do a Sora, Veo 3.1 comparison. But, you see it just takes a few minutes to generate these clips. And so, if you wanted to make a short film or a movie or a video trailer, you could take the same prompt and send it out to Sora, to send it out to Veo 3, even send it out multiple times and get different kind of feedbacks, and then take the best of all of them and bring them back together again. So I think, like, this is really the beginning of content generation, automation of video at scale. I think we'll start to see, like, a lot better video production tools where Sora and Veo 3 are integrated. Veo 3.1 are integrated, so you can kind of generate, TV commercials and things like that. We're, immediately started to do that ourselves across, like, HipTrain and Ryz Labs and some of our other businesses. Here, like to show the capybara's on the Built This Week podcast.
Jordan Metzner [10:15]: But, yeah, obviously, it's a huge threat to Hollywood and, content creators overall that were filming content historically, but it's just an exciting time because, it democratizes content creation. And, if some kid in his bedroom wanted to make a capybara, TV commercial, it would be almost impossible for them. And now they can just, prompt it out. And so, I think this will generate, a new form of content creatorship and hopefully a new form of content. And I don't know. It's super exciting, I think.
Sam Nadler [10:43]: Absolutely. Okay. Next up, I wanna talk about I think it's been one of the more popular apps for workflow automation, and I think, it's called n8n. And there's several similar businesses, but I would say this is one of the leaders. And I know OpenAI is is has recently released or in early beta are make going to release a similar functionality, for, to automate tasks, whether it's, sales kind of automations for inbound leads or marketing automations for generating content and posting content across channels or potentially, internal IT or back office automations. It's been really a popular tool, I think, over I don't know how long it's been, but I've I've been hearing about it pretty much nonstop for the past year. What I think is interesting, and we've started using it here at Ryz Labs, is, if you're while it's theoretically low code or no code, to get started, there I think there is a learning curve. It is a bit daunting when you're just getting started. But we've found a way to use Claude to build out, 90% of really any automation you could think of and then just kinda tie it all together at the end. But why don't you walk us through a little bit of how we've used n8n, some of the use cases we've applied it to, and just how we've accelerated the development of the process?
Jordan Metzner [12:13]: Yeah. So, just a little bit about n8n. N8n is a platform for, I would call, like, agentic workflows or just like workflows in general. It's probably not that much different from Zapier and Make and some of these other previous kind of like workflow automation tools. I think what n8n has done is kind of built this like agentic workflow that is now being copied by, most of the other players, and like you mentioned, OpenAI now has its own version of this. It connects to pretty much, every major tool, like Gmail, and Google Sheets, and Airtable, and others. You can, bring in third party APIs like OpenAI and ChatGPT, Perplexity, and others. And you can use it to generate all types of content and automated tools. So today, I'm gonna talk about kind of one of the automations that we've built, and I think it'll be pretty cool for you to see. I don't think you've seen this before, Sam. So these are kind of, like, what am I even looking at? And these are kind of two different automation workflows kind of repeated here. Yeah. Let's start with this one down here. So basically, what this n8n automation workflow does is it researches Russell 2,000 companies.
Jordan Metzner [13:24]: It uses Perplexity to do some additional research about the company. We then use a data API specific data, Prospeo, to obtain email data about the points of contact at that company. We then create structured outputs, send it to ChatGPT, merge all the data, put it into a Google Sheet, and then we actually put it into an email, and then we actually send the email. So the whole purpose of this is to do, basically sales outreach. The cool part about n8n is kind of once you connect all these pieces, you can then execute this workflow. So if you can see here, kind of the workflow is set execute every day, but just to show you and execute it by hand, what I can do here, Sam, is I can just go here and click execute workflow, and what you're gonna see happen, hopefully we can see it happen, You're gonna see the logs start to execute. So first, it's getting the companies, then it's getting the questions, then it's setting the topics, setting company questions, splitting out the questions. Now it's waiting, so we have some wait period so that we don't, like, spam any of our APIs or anything like that. And it's gonna continue to just you can see it just jump through these hoops one by one, going through the research, talking to ChatGPT, building up this email data, building all of the structured information. It's running through this process, it's doing all the research right now, and then here it is, it keeps going.
Jordan Metzner [14:46]: You see it here? And then it's gonna create all this structured output. The structured output is actually the customized email that we're gonna write to the company CEO. And so we do the company research, we find the company CEO's email, and then we use the agent to basically do some research together to create this structured email output. And it'll take a few minutes to run this entire task, but I think you can kinda get an idea here. You could see how many tokens, like, each job uses as it calls the different OpenAI APIs. All of this is connected into Google Sheets. And then all the way at the end, you can see here, this is connected to my Gmail, where it'll basically save the email draft, and then go back into the email and ship it out. Yeah. Now we have this whole process automatically done, but right now, we're just doing a manual run just to show you a little bit about how n8n works.
Sam Nadler [15:34]: I think we can show how easy it is using this technique for people to get started on n8n is do a quick prompt in Claude to build a quick n8n workflow. And then, you get the JSON and you import it. And then, yes, there is some and the last 10% of connecting it all, which can be at times challenging, but, you, for the most part, will have the workflow done, or the structure of the workflow done in seconds with, taking it to Claude first.
Jordan Metzner [16:06]: Okay. So I said, build me an n8n workflow that automates outbound outreach to podcast host asking him to be on our podcast built this week. There's a bunch of ways to get kind of lists of podcasts, but we could use perplexity for research and a bunch of different ways. I didn't really give it too many instructions. But it's gonna go create it all, and what it'll create is this kind of JSON file here. I'll just let it kind of automate that. We've built a bunch of different workflows so far. We're just beginning to use this tool. Like I said, there's a lot of other tools. I've checked out the ChatGPT workflow tool, agent workflow tool. It's very, very rudimentary, I would say, now. It definitely doesn't have as many features that n8n does. But, yeah, super cool opportunity where, like, AI starts to do work for you, and this is kind of the beginning of agentic workflows a little bit.
Sam Nadler [16:53]: Okay. Cool. If we can, let's import that into n8n just to see if it's done yet. Yeah. Oh, still working.
Jordan Metzner [16:59]: Oh, these are the instructions. So hopefully, let me see if I can get the JSON. So download the JSON from here, and then I can go to n8n and let's create a new workflow, and then let me how I import it again. There we go. From file. So now I'll just take the downloaded workflow that I just got out of Claude, and boom, there we go. Yeah. And here you can see kind of the scheduling versus manual trigger, getting the podcast leads, having some filtration, preparing the data, fetch the podcast website, generate an email, and then parse the email content, yada updates.
Sam Nadler [17:34]: Wow. Yeah. That's I mean, it's so fast, especially if, like, you don't really know how to get started, you know. I'm not saying this is the final version that will be perfect, but it's pretty good. And this is how we've, developed several different workflows in just a matter of couple weeks. Alright, Jordan. I wanna transition to the news of the day. Couple topics I wanna cover. The first one I think that is really cool given your background, working for many years at the largest online retailer in the world. Walmart has partnered with OpenAI so shoppers can buy things directly in ChatGPT. I know, ChatGPT signaled this, like, ability to purchase directly in app a couple weeks ago, but Walmart being, the largest, I think, physical retailer in the world, it is, I think, a huge opportunity for them heading into Christmas and the largest, buying season every year. And specifically, Amazon does not seem to be a part of this. I don't know. What are your high level thoughts?
Jordan Metzner [18:49]: Yeah. So, I mean, OpenAI just released its new kind of, like, app integration tools. My guess is Amazon and Expedia and all the top major apps are gonna integrate because it's kind of free traffic. Why not? And, chats become the new search interface. Right? It's it's beating Google in many ways, and people are starting their search on ChatGPT. Find me the best this, or, show me this product. I just had a pump for my pool break, and I just took a picture of it, and, I could've just led all the way you know, and said, like, tell me what this is and I need to buy a new one. And I could've led all the through the commerce directly in ChatGPT. I already have an Amazon account. It can know my preferences, kind of one click checkout. At the end of the day, like, it's not hard to go to Amazon and type in that product and search for it and check it out. Like, obviously, that is not a challenging thing. But as, the AI starts to know me better and know my preferences, then, maybe you can build a better shopping experience.
Jordan Metzner [19:47]: Amazon and Google historically have had, a yin and yang type relationship as, Google has been a massive driver of traffic into Amazon. But Amazon has been, skeptical in sharing too much information with Google as they've built things like Google Shopping and, Google Shopping ads compete with Amazon Shopping ads, etcetera. So, I think this is super interesting because, one, users are starting their search on ChatGPT, and so it feels like everyone's gonna have to integrate if you're gonna wanna get that type of affiliate traffic. Two, I think it kinda has a huge impact on, where users start their search, and I think, Amazon's been working over the past twenty years to get customers to start their, commerce searches directly on Amazon. But if inherently behavior changes and those searches start to move back to or start to move to ChatGPT versus, start to move away from Amazon again, that will be costly for Amazon both in, conversion as well as, whether they have to pay ChatGPT for traffic again like they have paid for Google historically.
Jordan Metzner [20:49]: So this is a really interesting time. It seems like the early days of like kind of web two thousands when like everybody was trying to do deals with everybody and nobody knew, what was happening. I think ChatGPT is gonna compete with all these guys too. They've raised a ton of money. They continue to build their stack. What products are they gonna sell? Are they gonna compete against you know, they're already competing against Google. Right? They're already competing against AWS. Are they gonna compete against Walmart and Amazon and ecommerce? Like, maybe so. Yeah. Lots going on. I think this is just the beginning days, and we'll see, if this has an impact. Are customers really gonna go to ChatGPT to buy on Walmart? Maybe. Maybe not. But, maybe it's a great lead gen and makes the shopping experience just a little bit better. So anyway, early days of, chat, AI, and shopping, and I don't know. We'll see.
Jordan Metzner [21:37]: Amazon's gonna continue to have to innovate if they wanna continue to have customers.
Sam Nadler [21:41]: Cool. The next story I wanna cover is about Anthropic and the just massive revenue growth targeting 9,000,000,000 ARR by the end of this year. And supposedly, according to Reuters, I'm 20 to 26,000,000,000, a two more than two x potential three x increase by the 2026. I just think it's again, everyone's talking about a bubble, or many people are talking about a bubble, but, are we is the cloud compute and budgets around spending on these AI agentic AI solutions, even bigger? Are we gonna see even more growth in 2026? I don't know where personally I land. I feel like there's so many signals that there's still a ton of growth on the horizon equally? It just feels like we're in uncharted territory with this amount of growth and hype.
Jordan Metzner [22:34]: Yeah. I mean, look. It's early days. I don't think we've ever seen software companies grow to this type of revenue in such a short period of time. I would say that if you take, us as a business, Ryz Labs, like, we are 100% codependent on OpenAI, less so probably in Anthropic, but, like, if OpenAI didn't exist, we'd probably be a 100% codependent on Anthropic. We are dependent on one of those two existing as it's critical for our ability to, generate code, to build content, to, write emails. I mean, we are 100% dependent on that. And they are dependent on data centers, and those data centers are dependent on NVIDIA chips, and those NVIDIA chips are dependent on TSMC, and, those data centers also depend on energy, and, it kinda goes all the way down. But I would say that we, as a business, could not operate without Anthropic or OpenAI or both. Each week, Anthropic just came out with an a new model, for kind of sub agentic processes.
Jordan Metzner [23:31]: But each week, we're seeing kind of like, each brand lev levitate itself above the others. I think we'll see a Gemini three coming out any day now, as we just saw Veo 3.1 kind of, they did it pretty quietly, but looks like it's a little bit better than Sora two. You know? And so we're just gonna continue to see these things where, to be in the game, you're gonna be, basically paying Google, paying Anthropic, paying OpenAI, and probably a few others, that'll come up in the kind of, like, vertical specialization from a model perspective. So, are they undervalued or overvalued? I don't know. I'm not an investor. If I could buy stock in Anthropic, I probably would. If I could buy stock in OpenAI, I probably would. I don't know at what price would make sense for me, but I think that, like, growth is only up to the upside. Whether they can go from 9,000,000,000 to 26,000,000,000, which would be, two and a half times revenue in a year, That probably is a little bit challenging just because, like, there's probably some, like, operational execution has to happen, like getting data centers up and some other things that, like, physics, like, just may not be able to allow. But I think it just shows that, like, there's an insatiable demand, to use these types of services in business. And, you and I see it every day. There's so many companies that just don't have AI implemented in their organizations. And so when they do, that revenue is gonna come to Anthropic and Gemini and OpenAI and others.
Jordan Metzner [24:51]: Right?
Sam Nadler [24:51]: Yeah. It's crazy. Okay. Last story is Apple has just rolled out m five across their product line, MacBook Pro, iPad, Vision Pro. I think not yet the Air, but supposedly, I think one more AI performance, I think four x AI performance compared to the m four. Also, in my opinion, really reasonable price tag, 14 inch MacBook Pro at about 1,500, $1,600. I don't know. It feels like probably a year ago, maybe it's been more that they released the m four, but just, a ton of more storage for more AI, essentially.
Jordan Metzner [25:33]: Yeah. Obviously I think, Apple's like it's kinda like the old Saturday Night Live thing with the razor. Right? Like one more blade. Right? So, m six is gonna be four times faster than m five, and, seven will be four times faster than six, and, etcetera. But, is this exciting and interesting to software developers or not? So I think that's, like, the real question here. Do you wanna go buy an m five? So a few things I think that are interesting to know. One is, in the previous versions, like m four and m three and m two, etcetera, they've had two additional lines besides kind of the m four line. They had, like, a pro, and I forgot the other line no line what they call it, they had two lines of premium chipsets on top of the standard line. With this m five announcement, they have not announced the premium version. So meaning that if you want the top of the line MacBook Pro with the most amount of RAM, you actually have to buy an m four Pro, I think, is what they call it.
Jordan Metzner [26:32]: So I think that's, like, incredibly interesting because that's it's a little, antithesis to kind of how Apple normally operates. I think the second thing is that, to your point, kind of they price them down, meaning that, like, it doesn't really make sense to buy a base model m four because you can buy a base model m five for the same price now. So that's kind of a replacement device. Obviously, the purpose of this device so you can run LLMs locally on your machine. I don't really know if developers are doing that. I know that we are not really doing that. I think, like, having a port more powerful machine will continue to help developers, like, produce more AI and produce better applications, but we're not really running kind of LLM models locally considering that we're using mostly, managed services from the LLM perspective. But, I guess we'd have to try using the m five and see if it's significantly better. I know when I moved up to m four, I didn't notice the difference, and I have a lot more RAM. So I think it's kind of like a half completed approach from Apple here.
Jordan Metzner [27:26]: And even funnier, like, they updated the Vision Pro. I thought that was, like, I thought that was dead, honestly. I mean, I don't think anybody's even talking about the m five update on the Vision Pro, or even, like, the updates on the iPad. I mean, is anybody even excited about, an iPad that now has m five or m four? I don't know. I don't think so. So and it's mostly because that when they do announce these features, they knew they announced these new chips, there's no features. And people care about features. They don't care about performance. Anyway, that's my wrap up on Apple. I think they're in a weird spot. Think I you just saw yesterday their head of search, who just got promoted to the role, just left to Meta. So it's been brutal for them on the AI front. Looks like they're partnering with everybody. But yeah, I thought that this was a so so launch from Apple, and not a superstar launch.
Jordan Metzner [28:08]: It seems like from the phone perspective, the AirPods threes, those have been doing really well, especially at retail. But this, we'll see. I think it's just a subset launch out of the Apple camp.
Sam Nadler [28:18]: Well, cool. Well, that wraps it up for Built This Week episode 17. Is there any final thoughts before we sign off?
Jordan Metzner [28:27]: Yeah. Sam, great episode. Like we mentioned, we've seen a crazy amount of investments. We didn't even talk about kind of all the OpenAI deals that they've done with, Oracle and NVIDIA. And, we talked about Walmart, but, I mean, it's just nonstop the amount of partnerships that they're doing. Broadcom got announced this week as well with OpenAI. So it's just been a slew of deals happening, ton of capital all over the place. And I'm excited for episode 18 next week because I think we'll see even more AI news and movement. So, thanks everyone for listening. Like and subscribe on YouTube, Spotify, Apple Podcast, and thanks, Sam. Great episode.
Sam Nadler [29:05]: See you next week, everyone.
Jordan Metzner [29:06]: Bye. See you.
