031 Will Vibe Coding Make Software Free?

[00:00:01] Speaker A: Okay, we're back over there. [00:00:02] Speaker B: We're good. [00:00:03] Speaker A: We're good. [00:00:04] Speaker B: All right, let's do it. Hey, welcome back. [00:00:09] Speaker A: Hello. Welcome back. That's right. [00:00:10] Speaker B: Welcome back. Gregory and Paul show. I'm Gregory. [00:00:13] Speaker A: I'm Paul. [00:00:15] Speaker B: And we break down the latest in SaaS, startups, AI, whatever the Internet's talking about. We love memes way too much. And we broadcast live every Friday at noon. Right. And we upload this to YouTube, where we've been building a great audience. Right. I think it's going really well over there. Yeah. Yeah. [00:00:42] Speaker A: Who knew? We knew. [00:00:43] Speaker B: And then. And we do. We do. We do upload this as a podcast as well, Correct. [00:00:50] Speaker A: Yeah. So podcast. Apple podcast. So Spotify. So listen. Listen wherever you listen to podcasts. [00:00:59] Speaker B: Exactly. Okay, so let's get into it. Obviously, I'm here in San Francisco. Yeah. Spending a couple of days here, and actually, what I wanted to share with some photos from the event I went to last night, which I thought would be pretty interesting for people. So I've been going to this, like, event series called the Llama Lounge. It's run by. Actually, there he is right there in the photo. This guy. Jeremiah. Oh, Yang. Yeah, Jeremiah right there in the middle. He's the host. And then they actually had CEO product hunt last night. Yeah. As the mc. Yeah, no, he was the mc. Yeah. There was, like, a startup pitch, and he just kind of ran the pitch, talked to all the. All the people pitching and stuff. But it was cool event. It's been around for two years. But what I thought was really interesting and why I wanted to talk about, I'd kick off the show with this is that it's an AI startup event. It's called the Llama Lounge, which is like a. Yeah, it's a joke based on Facebook's AI. Is it their model called Llama, Right. Yeah. And so I've been going to it since the very beginning, and I think the first event might have happened the month that ChatGPT 3 1/2 came out. [00:02:42] Speaker A: So it's been a couple years. [00:02:44] Speaker B: Yeah, it kicked off and at that point was pretty dark days when it came to startups in high tech. We just come through all of those, like, massive layoffs by large tech companies. All kinds of startups were laying people off. It was a really difficult time. So it's been like 2022. Is that right? [00:03:06] Speaker A: Yep, Sounds perfect. [00:03:07] Speaker B: And Jeremiah kicked off the series of events and brought a lot of enthusiasm and excitement to startups and to the Bay Area. You know, it was just like, people were Feeling really down and really negative. And of course a bunch of other events emerged around that time. There's a community here in San Francisco called Cerebral Valley who do all kinds of AI events. And what's been interesting is I have been going to these now for the past couple of years and I feel like it's a good barometer for the hype cycle when it comes to AI and startups. So of course like the events first started, they were small, they're at like a pizza shop. Like there wasn't a lot of interest from like large companies and large brands. And of course it took off and it grew. And you have interest now from all kinds of different large tech corporations, banks, you know, this one was hosted at the Amazon Builder loft in San Francisco. Next one will be at Google, it'll be at Snowflake. All the big tech companies are really excited now about AI and AI startups. Right. But I kind of feel like it's not quite as much enthusiasm as there was perhaps a year ago. And I thought that that was the part that we should, we should talk about. [00:04:34] Speaker A: Yeah, yeah, that's very interesting. [00:04:37] Speaker B: So, so I, I went like a year ago, I went to, I guess, I think it was Llama Lounge in Palo Alto and it was just packed to the rafters. Like you couldn't get in. Like it's an hour to get in, there's no. But standing room only. It was just insane. Right. And so last night I just didn't feel the same level of, you know, excitement. It could be that like it's January, people were like kind of, you know, getting into swing of things. But my guess is that a lot of people are looking at the AI space right now and the likelihood that you're going to get started today and build a big AI company I think is pretty low. I think they've all been established. Right. You've had Cursor Lovable, there's like Harvey on the AI side. There is a bunch of leaders who've emerged in maybe every category and they've all received an enormous amount of funding. So to compete with them would be really difficult. [00:05:43] Speaker A: Yeah, people are, I think people are getting used to the fact that big AI companies would just steamroll your product six to 12 months later or even faster. Right. So Anthropic just came out with this task management which essentially lets you run long running AI agents, a training mob and they would be able to parallel process a whole bunch of things. And this killed probably dozens of startups that had. [00:06:21] Speaker B: Yeah, yeah. So like last Night. So that's what I was going to say was like so what did I say? What are people working on that they still think there's an opportunity. So there was like an industrial automation company and they were working with like large scale, let's call it manufacturing, using AI to help optimize and automate some of it. And another company that was like it actually was similar technology perhaps but on circuit boards and electronics fabrication and manufacturing. So similar idea actually to this industrial one was like using AI to read through schematics and documentation. And to build any of these like hardware things requires like an enormous amount of understanding of what the components are and the databases. The way that you track all these different components is actually quite fragmented. It's just a lot of time goes into like is this the right power supply or is this the right chip that I need? And trying to like make sure that all the specs are correct. And so people working on these like AI tools that are able to database and look at all that stuff and help you make the right decision to build like a circuit board. And same thing with like the industrial AI, same idea. There's lots of schematics and like documentation like if you're going to build a large building, you're going to put all the smoke alarms in. There's all kinds of like regulatory things that you need to manage. There's just a lot there. Right. And so trying to use like AI to optimize and automate so AI, some [00:08:03] Speaker A: of that with hardware probably like big government or government industries. Just things that like normal people can't get into I would say. [00:08:14] Speaker B: Right, yeah, I think that's the case right there. So it was interesting. I would lump them both together so exactly like they're exploring how to apply AI to hardware, large scale industrial industries, industries like they're not software. Right. Where it's all self contained, it's all in the cloud. Right. Like lovable, like like auto generating apps, vibe coding, all this stuff. Right. This is taking it into like the real world and like trying to find like what parts they can optimize. And also people are using. I think this part's important actually. Like the guy who had this like it's kind of chatgpt for electronics manufacturing and what he said was that there's a lot of people out there with these tools that like just auto build you a diagram for the circuit board that you want. And I've actually seen a lot of people who are like they've got some tool that creates schematics for Circuit board. But like a lot of AI experiences that I've had with AI where you get like hallucinations or it just doesn't have a great solution. That's what holds a lot of those back is that you look at the circuit board that AI is recommending. It's not the right solution. [00:09:23] Speaker A: Right. [00:09:25] Speaker B: So yeah, so it just says one part. So what he's doing is like using AI actually to just like database all the information. So just throw all the PDFs in there, just all the information, just throw it into like one thing and then you can kind of like query it and just use it as like a tool. Which I thought was like really interesting. Like accepting the fact that like you can't automate the entire thing, just like improve parts of it to like an AI enabled circuit board designer or engineer. [00:09:53] Speaker A: Yeah, that's it exactly. It's like I think people are just trying to do too much with AI really at the end of the day it's just treating it like, you know, a really capable intern. At the end of the day that's what you have to do. So like another field that I think AI still has place to compete is it is actually still dev experience. So dev tools, Cursor, even though that they've raised lots of money, they're super successful, there are still areas where they can't compete. So Open Code is a big competitor to both Clock Code and Cursor. And I mean it's vanity metrics, but they're one of the fastest company ever to reach 80,000 stars on GitHub. [00:10:45] Speaker B: What are they called? [00:10:46] Speaker A: They're called Open Code Open Code. [00:10:49] Speaker B: Oh, okay. [00:10:52] Speaker A: So there are also like a terminal UI for Vibe coding. They plug into all sorts of foundational models like Anthropic OpenAI and they do provide a better wrapper around the foundational models, a little bit better than all of the other companies. [00:11:15] Speaker B: How is it different from Cursor? [00:11:18] Speaker A: So really it's like every single company has their proprietary, let's say customized prompts around the foundational models to make it do behave a little bit differently. So they are more developer focused even than Cursor because like you have to, you have to realize, for example, Lovable is targeting people who has zero dev experience. So their experience of using the tool should be like going as far as possible without asking you for follow up questions. Cursor is somewhere in the middle. Right. It's people who are comfortable with living IDs which is like this coding interface. Maybe it's their first time using ID for open code. They're literally just going to, let's say senior developers who want a lot of control, who wants a lot of back and forth, who knows exactly what they want. So the experience is slightly different. [00:12:20] Speaker B: Interesting. I saw a guy last night who had a tool that, it was a Vibe coding tool, but it worked inside the cloud. So inside your Amazon instance or Google Cloud and it can access all the services and stuff, which like Cursor is not good at or, or I don't even think it can actually do. Yeah, he's like, he's like. Or you can ask it to do it, but it doesn't. What did he. What was one specific feature or challenge I was able to overcome is that even if Cursor was like to give you all the code required to run directly inside Amazon, it doesn't know like what else is in Amazon. And so his tool is able to like look at all of the code, everything you're doing in your cloud account, and then augment all of that. It was pretty interesting use case and I can see how it's very hard, I think for Cursor to do that well. So people are still finding opportunities out there. [00:13:21] Speaker A: Exactly. So here's my prediction. Right? So all of these five coding tools. You know, like at the very beginning of Vibe coding, Rick Rubin was kind of like the face of it. [00:13:31] Speaker B: People made like, he's still. [00:13:34] Speaker A: That's right. It's kind of hilarious, right? Like, so you had that like initial adoption of people that aspire to become like Rick Rubin type. [00:13:45] Speaker B: I, I do, I love. He's a hero. [00:13:47] Speaker A: But, but like not everybody aspires to that. So now you're getting people that's like more senior level developers. They, they're aspired to be a different types of person. So like there's still archetypes that's being unfolded here. So to your, to your story of like the guy that plugs into Amazon, I could totally see a use case there. Right. Sysadmins, for example, or security developers who. [00:14:13] Speaker B: No, I think you're dead on. I think you're, I think it's almost like you could define. [00:14:18] Speaker A: Well. [00:14:19] Speaker B: So a couple of things have emerged I think and this, this ties all nicely back in like Llama Lounge. Right. So like the beginning of the AI revolution, everyone's like experimenting, trying different things and where we're at is. And, and I guess my summary would be the number one use case is writing code and engineering of the current version of AI that we're working with. And it. And that Alone is massive. It's huge. And, like, I think it's well understood and accepted now. Like, I think the, the, the majority of people in the space accept that some form of Vibe coding is the future of engineering. You had a lot of resistance. So, like, my brother's an engineer. Right. In the beginning, he hated it. Right. He's been around a long time. He's like, this stuff is garbage. Even he's, like, using it, and he's, like, reluctantly telling me, like, some cool stuff. He still doesn't tell me stories about what it did wrong, but he finally told me some stories about what it did right. [00:15:14] Speaker A: Right. [00:15:15] Speaker B: So I think that, like, because it's so well accepted now, all these things are emerging where, like, oh, it doesn't work here. We need a better tool to do that. And those are all getting built out now because people have accepted that, like, vibe coding in general is the future of software coding. So, like, all the edge cases and other things you might need, the entire space is going to get built out. [00:15:37] Speaker A: Yeah. So still lots of room to grow, but. Yeah, absolutely. I think I agree with you that the high pass died down just a little bit compared to. Little bit. [00:15:49] Speaker B: Yeah. Okay. Okay, let's. Let's get into our doc here. I know we got some stories we want to talk about. I think this is perfect one to kick it off with. [00:16:01] Speaker A: Yeah. [00:16:02] Speaker B: So the CEO of Anthropic. [00:16:07] Speaker A: Yeah, yeah. [00:16:08] Speaker B: Let me bring this up here. Dario Amodi. Is that how you pronounce his last name? Do you know he. He's, you know, everyone's at Davos this week and talking. And so he made this interesting statement about how not only does he see, as we see all coding being automated by AI, but he's saying that, like, you know, it's very possible that the cost of developing software just goes to zero. Right. That, like. And other people said this too, that like, like it's just going to be so inexpensive to do. That's going to change the nature of, of software engineering and change the nature of SaaS. [00:16:59] Speaker A: Yeah. [00:16:59] Speaker B: So as you're. You're. You're our resident Vibe coding extraordinaire. [00:17:03] Speaker A: Yeah. So. So what do you. [00:17:05] Speaker B: What do you think about this? [00:17:05] Speaker A: Yeah, so there's, well, there's like, there's two to three thoughts here. One is tokens are still expensive. So before. Before I really see that happening, it's still quite expensive to build software in a production. Production way. So it still costs millions and millions of tokens and hours and hours to even begin to put Together a production ready software and tokens right now seems like it's getting more expensive, not cheaper because more people is using it. And then if you listen to the news a little bit more, it seems like it's an energy, it's tied to energy as well. Right. So as more compute is needed to bring more AI capacities online, the tokens are actually getting more expensive. So like yeah, the guy is probably thinking like five to 10 years out, which could potentially be possible. Good. Yeah, which could. The second thought here is just because AI is able to cope for you, it doesn't mean that the quality of software will follow suit. I think just more and more software that nobody ever uses or nobody ever sees will get built and it'll just become more and more doicy. [00:18:35] Speaker B: So yeah, so on the first point on the costs, that's an interesting one I actually hadn't thought. And of course no one asked them at Davos, they're just like hey, you're so amazing. Yeah, right. Like no one asked like hey, token costs and energy because it is important, right? Like we don't have the energet that we need to power this. And like in fact like at DAV this I guess this ties into this discussion because like at Davos, Germany got up and admitted that their green energy strategy was not working for them. That like that they need to actually invest in nuclear and they, they need more, they need more energy. So, so I guess like, and for like the context of here regarding vibe coding, it's bullish for Nvidia for sure and bullish for all the people that are creating the like selling the picks and shovels, right? That like there's always this question like is that going to peter out? Have we built enough? If, if, if the token cost isn't like rapidly going down, it means that like there's, there's not enough capacity out there. [00:19:49] Speaker A: That's right, yeah, there's not enough capacity. Well it's actually not good for Nvidia. Right? [00:19:54] Speaker B: Like why say that? [00:19:56] Speaker A: Well if people suddenly realizes that token cost will not keep going down and there is actually a threshold and there's limited access, then price is going to shoot through the roof and Nvidia's market share is going to go down. Right now everybody is trying to bring up capacity, but if people start realizing that they can't compete, that they're going to stop buying, investing. [00:20:24] Speaker B: No, it doesn't work like that. If there's demand, if there's demand, like they will just keep building until the demand goes away. [00:20:31] Speaker A: But the problem, that's what I think the problem is. Like it's so capital intensive that smaller correctors will not even enter the market. There could potentially be a case. So. Okay, I'll give you an example. Right, okay. [00:20:45] Speaker B: Okay. [00:20:45] Speaker A: Okay. Every single person in the world wants, wants a cell phone. Right? [00:20:50] Speaker B: Okay. [00:20:50] Speaker A: But there's not a million cell phone manufacturers. No, there's actually only two to three maybe. [00:21:00] Speaker B: Well there's some small Chinese ones that do pretty low cost stuff, Right? [00:21:03] Speaker A: Sure, yeah. But like in, let's say America, there's Sam. There's really Samsung and Apple. [00:21:10] Speaker B: Sure, in the United States. [00:21:11] Speaker A: Yeah, in the United States. So. But every single person in the United States has cell phones. But like the reason why there's no like a third or even like there's no dozens of software manufacturers just because like the supply chain and people look at this market and they're going to be like, why would I start a software company? There's no market for me to compete. [00:21:36] Speaker B: Yeah, I think it's, I think it's a different, different. I think it's, I think the dynamics are different. So what I think you're talking about is like a mature market. So I would just, and I could be wrong on this, but I just think of the business cycle. So new technology or a new innovation or a new piece of land, like there's a lot of opportunities where something new happens, discovery. Right. A bunch of people rush to that market and there's usually this massive fragmentation of like players who try all kinds of things and then over time they consolidate and over the long term most markets tend to monopolize. Right. And I think that that's where in the US like you have two players on the mobile phone size because it's just too expensive and too difficult to enter. Enter the market. But like what is that? But what? I, I don't think that's necessarily the case. Oh, go ahead. [00:22:31] Speaker A: Yeah, well like it ties back because like, so everybody's investing into capacity right now because they are expecting the overall, the model quality and their hardware will give them. Well essentially like they, they could potentially deliver more in the future. Right. [00:22:59] Speaker B: Well there's, there's demand. Right. Like there's demand and there's having trouble filling the demand. So that's what's scaling the market. Right. And then. [00:23:08] Speaker A: Right. [00:23:09] Speaker B: You do have compet technologies, GPUs. TPUs. What is, does AMD make GPUs? I don't even know. Right. So I mean there's kind of like a lot of players in the market right now. But there's technological like segmentation or fragmentation. Like the technologies are fundamentally different and that's why they don't work together, which actually does exist in the. It does exist in the cell phone world too. Here's a really nerdy aside. There are different competing cell phone technologies. CDMA and I can't believe I know this. And gsm. Yeah. And CDMA is only United States and Verizon. But at this point it's so obfuscated away that it doesn't matter. So that kind of dynamic exists in chips right now where there's different type chips and they only work with their own kind of technology stack. And so at this moment, that part is important in driving the business. But at the highest level, what I thought was interesting what he said is that there's demand. The price of tokens is not going down because people want this stuff and are using this stuff in such a huge way that all this capacity built isn't enough. Right. Yeah. I think it means that the market's not done. They maybe got another year or two of just infrastructure build out. Maybe, maybe more. [00:24:35] Speaker A: Right. [00:24:36] Speaker B: Like because there's, there's people that keep saying on the one hand, right. That like, oh, it's a, it's a bubble. But I kind of feel like it proves it's not like there is actually enormous demand for the current generation of technology that we have today and we're not able to meet it. [00:24:54] Speaker A: Yeah. So like there's demand because well one, it's new. People don't know where the limit is. So they're always pushing the boundary. Right. Maybe like a way for me to summarize what I'm trying to say is once that boundary has been established, people will lose interest fast. So the demand will go down. And then whether if it's almost free or not, that really depends on or in the cycle we are. Because I know for a fact that these LLMs are not making money. Right. Anthropic is a lot of the usage. [00:25:35] Speaker B: Yeah. It's not profitable at all. [00:25:37] Speaker A: It's not problem. [00:25:37] Speaker B: Well, they got to get. They got to get the cost down. [00:25:39] Speaker A: They got to get cost down and get may retain the users because there's no mo in keeping the users correct. I can just use tomorrow. [00:25:52] Speaker B: Let's move on. You got some other things in the doc here. Do you want to talk about this stuff? There's this quad code thing. Do we. [00:25:59] Speaker A: Yeah, talk about that. I would love to. I'll spend like 30 seconds. But it is incredible what we can do with. So Remotion. Remoji used to be this speech to. Remoji used to be this tool to help you make videos. And they literally just came out with a cloth skill that plugs into their. Their code base. And now you can vibe create. You can vibe create demo videos. I've been playing around with this for the past couple days and it is phenomenal. It's awesome. [00:26:45] Speaker B: Dude, they've got a cool video here. [00:26:47] Speaker A: Yeah. So this video that they're showing you, they vibe coded it so essentially it turns cold. Yeah, that's right. Yeah, it turns cold into videos. It is. [00:27:00] Speaker B: I like the ASCII art in there. [00:27:02] Speaker A: It is totally pretty nerdy. Super nerdy. It's awesome. [00:27:08] Speaker B: Like, so you're using this stuff. [00:27:10] Speaker A: You like this? I love it. [00:27:12] Speaker B: I haven't tried this at all. [00:27:13] Speaker A: Yeah. Yeah. So I'll show you what I'm. I'll give you. Okay. [00:27:16] Speaker B: Yeah, show us something. [00:27:19] Speaker A: Screen showing. [00:27:21] Speaker B: Yeah, we haven't done any like, live vibe coding in a while on the show, so. [00:27:24] Speaker A: Okay. And so first of all, our. [00:27:27] Speaker B: Our low. Our lowest. Our lowest in popularity video ever. But I don't care. That's what we do here. [00:27:34] Speaker A: This is cool. So like, so this basically they give you a live video editor right off out of the box. And then this entire video I vibe coded so far. [00:27:50] Speaker B: What? [00:27:51] Speaker A: Yeah. Okay. [00:27:52] Speaker B: Can you create something from scratch? [00:27:54] Speaker A: Yeah, totally. Here, Olivia. Sure. [00:27:56] Speaker B: So there's like a prompt window. You tell one and then it just creates a video for you. [00:27:59] Speaker A: Yeah, exactly. Yeah. So I'll show you the. I'll show you the entire run and [00:28:03] Speaker B: it's on the timeline. That's interesting. I haven't actually seen anything do it in a timeline. [00:28:10] Speaker A: I'll show you exactly what. [00:28:12] Speaker B: Yeah. Yeah. Okay, cool. [00:28:13] Speaker A: So like, literally at the very start of the thing, I told it that I wanted to create this. This product demo. I'm doing it for this product called Spec hold and it literally just like organized the whole thing. I uploaded a screenshot of like what the product cool page look like and it just writes code and this code turns into. Into the video. It is amazing. This is like the coolest thing ever, dude. So like this all lives inside of plot code and all I have to do is just like the next thing I was going to do is, hey, let's add a message that prompts the user to connect their database. What message do connect the Supabase data. Anyone Just go through and figure this out how to do it. It's just. It's awesome. I am Having so much fun. [00:29:16] Speaker B: All right, what's. What we got next? Okay, I want to talk about the TikTok stuff. Just quickly. [00:29:21] Speaker A: Let's do it. [00:29:23] Speaker B: So let me bring this up here. Yeah, it just was in the news this morning when I was like, looking through my feed and I saw that the, the TikTok deal has just been finalized, so. That's right. Yeah. Yeah, we've talked a lot about this. I was, I was kind of. I thought it was kind of interesting that, like, this didn't get a lot of attention. So bytedance has finalized deal to create a majority owned American joint venture which protects the data, you know, Oracle, and there's a venture firm that are part of it. So it's like a done deal. Like TikTok is now legally [00:30:02] Speaker A: majority. Yes. American joint venture. [00:30:06] Speaker B: Yeah. Yeah. So I thought that was kind of interesting because that was like in the news for a long time. There was a lot of debate about it. [00:30:13] Speaker A: Right? [00:30:14] Speaker B: Yeah. And so it's a, it's a done deal. So I thought that was a. I thought that was kind of cool. [00:30:21] Speaker A: So do you think that Oracle is the big winner in here? It looks like. [00:30:27] Speaker B: Oh, dude, without a doubt, man. Like, there's a lot of interesting stuff on Twitter about this. Right? Like, you know, there were some people saying, like, it's a sweetheart deal for Oracle, but like I said, there's also the other perspective, which is like, hey, this business has been outlawed and it's worth zero. And so like, if anyone gets anything for it, which I was actually the, the position I would take on this. I also think that, like, I'm curious to see how it plays out, like, at a high level. I always, I always believed the rumor that it was just, let's say, bad to have all this data living in China and that they still have some kind of adversarial view of the United States. And I just didn't think that that was like a great setup for us. And so I think pushing this and creating actually solution that's win, win for everybody in the sense that, like, users didn't want to get shut down. A lot of people were on there, of course. [00:31:26] Speaker A: Yeah. [00:31:27] Speaker B: But like, we need to protect our national interests. What I think does matter and that we found a way to solve for that is great. So I think at a high level, I think it's really positive. Yeah, keeps. It. Keeps also like some of the value and it's called value creation in the United States. [00:31:47] Speaker A: Right. [00:31:48] Speaker B: All goes to America, company employees, Americans, and we make money like it. [00:31:52] Speaker A: I Like, it a lot. Maybe the alternative what will. What's. What shut them down and everybody goes to. Goes to Instagram. I don't think that was a good solution at all. [00:32:04] Speaker B: Yeah, well, yeah, it's like a good one for Mark, right? [00:32:07] Speaker A: For. [00:32:07] Speaker B: For Zuckerberg. But, like, I don't think that's a great solution and I don't think it's great for the users. And I feel like it's very un American solution, just kind of ban things. It's like, why not find some legal mechanism where we can bring the business here and then have some more control? I think that's a better solution in China. That's kind of what they did to American companies. Right. Google doesn't operate in China. They just left the market. So that's not my favorite way to do it. I think they kind of saw it in the same. Same way that, like, yeah, Google is definitely like, I have no idea what's going on, but in my opinion, they're. They're like the fourth branch of the U.S. government. Right. There's a lot happening there and they're really influential. And I could see if I was China, like, this isn't great for me. [00:32:57] Speaker A: Right. So. [00:32:58] Speaker B: So I think. I think it's great that we found, like, what I think is a good solution for it. All right, let's do. Oh, I got this one tweet from Sam Altman that I thought was funny. So he. He. This is good. I guess he was kind of feeling left out with, like, Anthropic. You know, they're raising all that money. Anthropic's got sequoia investing at $350 billion valuation. And so he just had to let everyone know that, hey, they're. Where is it? [00:33:39] Speaker A: Million dollar ARR. [00:33:41] Speaker B: Yeah, they've got a. [00:33:43] Speaker A: In the last month from just their API business. [00:33:49] Speaker B: Yeah. [00:33:51] Speaker A: You know that Twitter account, vc? There it is. [00:33:54] Speaker B: One billion. I'm sorry, one billion of Arrow. So, like, there's been a lot of criticism, too, of their B2B strategy and that, you know, there's like, this narrative that they're losing to Anthropic. So I thought this was interesting that. [00:34:09] Speaker A: Sure. [00:34:09] Speaker B: Sam had to kind of chime in. That's all I had. Now I got another kind of quick hit. So this is. This actually might be something we should spend a little more time on. So agentic commerce is real. And I feel like this is kind of under the radar too, that, you know, Amazon announced this Rufus AI shopping tool. This thing's at a $10 billion run. Yeah. Yeah. It's like, I feel like it's just kind of under the radar. Like no one's talking about this. [00:34:47] Speaker A: So I'm. But it's big. Yeah. I'm a big user of Amazon and I have seen this before. How is this different? It's just a new way to discover [00:34:57] Speaker B: products is that it's like a chatbot AI tool where you can go and just like make purchases and stuff. Like there's a part of it, things that like, it's possible that agentic commerce becomes a primary way that people start interacting with like Amazon shop. We've talked about it with Shopify to some degree, but like going onto the website, navigating through the ui, doing all of that. If you could just kind of about talk, talk to a bot, maybe, maybe through voice, maybe through text. And it was really good at figuring out like what you wanted. I think that might become a preferred way to shop. [00:35:32] Speaker A: 100. I personally would love to use this. [00:35:36] Speaker B: And it's not, it's not a weird experiment. And with a small number, I mean like, like, like 10 billion is a big number. It's not like a hundred million, 200 million, 500 million. It's 10 billion. Like that, that number like blew my mind actually, that like Amazon is able to like drive. [00:35:56] Speaker A: Right, right. [00:35:57] Speaker B: They're a big company and like, they've got resources. But like, if you asked me before it launched, like, that's not the number I would have picked. I thought like, oh, maybe it'll catch on. Maybe we'll do 100 million, 50, you know, 500 million. [00:36:09] Speaker A: Right. [00:36:11] Speaker B: I mean they're a big company and I get that. And they have the users, but that they got the adoption, that's the part that's like quite interesting. And it makes me very, very bullish for agentic commerce in general. [00:36:20] Speaker A: I mean, I think they have the distribution. Right. Is this built into the Amazon setup? I guess it's probably exactly. What's up? [00:36:27] Speaker B: I think it's actually I haven't tried it myself, so I don't know exactly what the standalone. Like if it, if it's standalone app with the use case in terms of like how you access it is like I just stumbled, like I literally stumbled on this a few days ago because I was like having fun making fun of the name. I'm not a big fan of Rufus. I love to tweet at Jassy like, dude, you need some help with Burning Market. Like I'm around like you guys can always call me. It's like A dog name. In America, we would name a dog Rufus. We name a dog Ruth. I don't know, maybe I have like a little dog logo or something. [00:37:05] Speaker A: But Alexa to Rufus. [00:37:09] Speaker B: Yeah, yeah. What I also think is interesting is that I think it means that ads inside LLMs, I'm extremely valuable. Bullish on. Yeah. We had Andrea from thread here and everyone knows, like, threads, that's my customer. I'm a 100% sold on the idea that an ad network and advertising inside LLM is going to be a massive, massive business. I think it's the most exciting thing to happen in ad tech in a long time. That's why I look at Rufus and I'm like $10 billion worth of like, commerce is happening in here. It's very obvious that you would put advertising in there. And if I was Amazon, like, I'd be very excited about an ad platform that just massively drives their profitability because they could just, you know, sell ad slots in there. I also think that, like, ads inside LLMs I think will give people like a search type experience, but on steroids. Like, search ads were pretty good. Like people, you typed what you wanted, you clicked on it and you kind of got it right. So I think from a user perspective, they're pretty good. And advertisers, they were great for them. Right. You bought the demand. But I think, I think an ad as an ILM is going to be even better. You're gonna be like, I'm trying to find some shoes. What kind of shoes? Oh, I want sneakers. What kind of sneakers? I want running shoes. Like, okay, here's like all the different Nikes. There's all different ideas. Like the sponsors can buy those things and like, you know, really provide a lot of value to the, to the end user, which I think is like, is really. Yeah, really interesting. I think it's going to work really well and I think consumers are going to be really happy. [00:38:48] Speaker A: I totally see it because I'm really tired of like the current app format, which is just designed to completely stop you from scrolling your social feed. I love the idea that I'm having a conversation, I'm getting value as if, like I'm talking to a salesperson. Right, exactly. [00:39:13] Speaker B: Interrupted Interruption as an ad format is just terrible. [00:39:18] Speaker A: That's right. Yeah. It's basically what, I mean, what Facebook gave us. They want to keep people inside of the stream and the only way for you to sell something to them is to interrupt their scroll. But hopefully this chat model completely changes it. So. [00:39:35] Speaker B: All right, I got another topic Are you ready for this one? The new X algo. [00:39:40] Speaker A: The new X algo. [00:39:42] Speaker B: Can you handle this? [00:39:43] Speaker A: Oh man. There is propane DAO into 1313 commandments. Duly open source to X algorithm. [00:40:02] Speaker B: So there's the. Right. So just to get up to speed, so this is the post from X Engineering directly to the GitHub repository where they posted the algorithm. [00:40:14] Speaker A: That's right. They open source the latest and greatest [00:40:17] Speaker B: X algorithm and so nerds have gone through on explaining to everybody exactly how the algorithm works, right? [00:40:26] Speaker A: Yeah, that's right. So like, I mean there's. There's things in there that's kind of obvious, right? Like out of network retrieval is the only way real way for you to grow. Of course it means that you have to write content that is appeal to people that's not directly following you. Makes total sense. Replies Boost your content. Blocks and views will hit your content and your reach really, really hard. Power users replying to your post or reposting your content will boost your signals. Will boost your content. Of course. Right? Like all of these kind of make sense. But the one thing that kind of stood out to me is the concept of embedding which locks you into kind of like this niche. The way that I interpret it is that X algorithm rewards your account talking about the same thing versus talking about multiple things. The way that he described it is your discovery will die for weeks if you post off niche topics. And also that you need to be one dimensional social. How do you feel about this? [00:41:43] Speaker B: Dude, I'm really clear. I do not like it at all. I think it's terrible. I think the starter pack video is cool. Like it looks cool. But I think this idea that like. Okay, so let's start from the top. So the idea that like you need to like post in a niche is like maybe the worst thing about this. So I think what made Twitter great was like people would post like, I got an opinion on this coffee and then I'm doing this vibe coding thing. And then I had this for lunch. Like it was this weird variance of content that like represented the real person. And I thought that's what was like really fun. [00:42:18] Speaker A: Yeah. [00:42:18] Speaker B: About it and all the dumb good morning tweets and funny memes and like it didn't have to be all connected. So the idea that you have to just be now an influencer who just talks about like B2B SaaS sales or you know, how to, how to do marketing lead generation, like just sounds horrible. Like I think it's horrible. So I don't. I Don't like that part at all. I also think that like, that like followers don't matter is just bonkers. Like, why even have them? Like, why even. Like the whole idea behind social media was like you were able to collect followers who wanted to hear from you and then you could interact directly with them. And that the more you had. Yeah. The more, the more reach you had. Like, it's super obvious way to do it. And like the idea that, that like, from what I can tell, has no impact. Like it's just a random choice, like random selection. Like it just. What do they do? Just throw the tweets out there. Like, I don't understand how that works. [00:43:27] Speaker A: I mean, it kind of makes sense, right? Like if you go through your follower, maybe like 20, 30% of them are still active. The rest are probably. [00:43:37] Speaker B: It makes no sense, Paul. Does not make sense. It does not make sense. Everyone has dormant, dormant followers. That's correct. That part's true. But like the idea that like your followers don't matter, I'm not okay with that. [00:43:57] Speaker A: Like, what does these numbers mean? I'm just growing a number for no reason. [00:44:02] Speaker B: Right. So you still have. That's the other part that like hurts my head is like they still exist and then those are the only people I can like send direct message to. So they kind of matter. It's just weird. I don't. So I don't like that part at all. I don't. So I don't like the niche part. I don't like followers don't count part. And then I don't believe some of these claims. So if you're a billionaire tech guy, you can post whoever you want and you get plenty of reach. Doesn't matter like what niche you're in. Right. So like it's not clear that this is applied to, to everybody. [00:44:38] Speaker A: Right. There's definitely levels, right? Like if you have a hundred thousand followers, then it's different from if you have 10,000. Different from a thousand. [00:44:48] Speaker B: See, followers do count. [00:44:50] Speaker A: So follow. But they're not going to say that to you. Really? They're not gonna say that. [00:44:59] Speaker B: It's funny. I, I did like, so, so, okay, so enough complaining. Things I do, like is like, I do think the starter pack idea is a good idea and it's good for New Year we. And they. I think that's what he said to Nikita. And they had this a long time ago. Like in the early days, you sign up for Twitter. Yeah, it kind of had like, help me follow people and I had like these starter packs and stuff because I. I've been working with some founders and creating new accounts for them on X and like, dude, starting as a new account, I forgot how hard it was. It's very confusing. It's like just really hard to, like, make an impact and navigate the experience and find people to follow and just kind of. You have to kind of like, choose your own adventure on X in a way that. Yeah, is really difficult. And so, like, I think. I think anything they can do to make the new user experience more palatable is good because, like, they need new users, they need more people to join the platform. I feel like I've known so many people have left. I don't know many new people who want to join. So at a high level, at least. I like the. I like the aim and I like the. I like the concept. [00:46:09] Speaker A: Yeah, the starter pack is like a low effort. You know, this. I feel like this has been tried [00:46:13] Speaker B: and tested, but, dude, this is the best. This is the best starter pack that they've. [00:46:21] Speaker A: That's right. We had. [00:46:25] Speaker B: This one's good. The Elon Baby mama starter pack. [00:46:36] Speaker A: Hold on. I think. I think Ricky's wrong. This. Pull this up again. This is. This is the best out of this week. Oh, man. Where. Where are we? Give me a sec. Yeah, this is. This is probably the hilarious. The most hilarious one this week. And we're. Yeah, well, we're gonna leave it as that and it can't. This is like super entertaining and it's basically going to be like, exactly like this. [00:47:25] Speaker B: There we are back. [00:47:25] Speaker A: Yep, you're back. [00:47:27] Speaker B: I was laughing too hard about the baby mama, so. [00:47:30] Speaker A: Yes, this is probably the best thing I've seen on X for a while. I saw this as like, oh, my God, this is the funniest thing. And also, like, the response is hilarious too. [00:47:49] Speaker B: What did they say? [00:47:50] Speaker A: Yeah, Tiffany responded to this. That's cool. [00:47:54] Speaker B: Yeah, no, she's did. She is. She's really funny. It's funny. I've had people who, like, have no idea who she is at all. Like, I spend way too much time on X. Like, I think about the. [00:48:06] Speaker A: Way too much. [00:48:06] Speaker B: Yeah, yeah, yeah. I. I think of all these people on there as if they're like, they're like my friends or something. Like, oh, yeah, like so and so and blah, blah. Like. And I do realize that regular people have just no idea. [00:48:19] Speaker A: Clue because we're. We're constantly online. That's. That's what. Yeah. [00:48:23] Speaker B: Constantly on. On Twitter. Okay, so we got through the. Do you have anything else you want to Say on the algorithm? [00:48:29] Speaker A: No. Let's see where it goes. Yeah. [00:48:32] Speaker B: Do you have anything you want to say on that? So the Brex thing. We have the Brex thing. Like, do you know anything about this Brex stuff? Like, what do you, did you put this in here? [00:48:40] Speaker A: Oh, yeah. Okay, so let's just, we should talk about this, this, because we've talked about either things going IPO or money getting returned to investors, right? So, like, there's liquidity events and it's, it's, it's good to talk about from what is that the link? Is that the best link? [00:49:05] Speaker B: I've seen a few other ones. I think this one's the same as, like, that I've seen. [00:49:11] Speaker A: Interesting. So looks like the breakdown from this $5 million exit, which by all means, it's a great exit. The founders are going to make a couple hundred million dollars each. The employees that were, you know, onboarded through, let's say in the last three years will probably get paid. But somebody did a analysis on the return of the investors. Besides the first two rounds of investors, everybody got to the 3X, you know, which is like not amazing returns, right? Like 2 to 3x4 VC fund that investing Series B is, it's good, but not amazing. [00:50:03] Speaker B: So just go back to the investment numbers. Oh, is this the same table? Yeah, the same stuff. [00:50:08] Speaker A: Yeah. [00:50:08] Speaker B: So, like, so, so Series A, it only returns 64%, correct? Is that what that's saying? [00:50:16] Speaker A: That's exactly what it said, yeah. Yeah. [00:50:19] Speaker B: So like on a hundred million, it didn't even double it? [00:50:26] Speaker A: No. Correct. [00:50:29] Speaker B: Wow. [00:50:29] Speaker A: Yeah. [00:50:31] Speaker B: So like a Series A, you would expect like a pretty good big return, right? I mean, like a venture firm wants like three, six, ten times, right. [00:50:41] Speaker A: Looking for ten times that Series A, that's for sure. Right. So, so it does basically what, it means that the price that you're coming in at on every single round matters a lot. Right. Brexit, the highest point, were valued at 12 billion. That's what they raised the, the last rank on. [00:51:06] Speaker B: Yeah, I, I, so I've seen a couple of debates online about this, and from what I could read, there were some people, there was a lot. There was, there were some debates, there was arguments over the numbers, but my conclusion was like, the numbers are basically correct. There might be some nuances here and there, but yeah, they're, they're correct. So there's people talking about like, liquidation preferences, there's all kinds of other stuff. Right. But they're like, basically these are, these are probably fairly accurate. [00:51:36] Speaker A: Well, they're they're going to get. [00:51:37] Speaker B: Yeah. [00:51:38] Speaker A: Like there's liquidation preferences. Like on series D investors they have liquidation preferences. Right. [00:51:45] Speaker B: D has some preferences, which makes sense because like DEED typically doesn't have like a great return profile. Right. Like they're not looking to double or triple that money. If you're like on a series D investment, I don't think they're aiming for zero either. That's not a great. That's funny. Well, we got it all back. I think that. Okay, I would say this one view or one analysis is that this is us working through the extreme valuation of zerp. [00:52:24] Speaker A: That's right. [00:52:24] Speaker B: So there's zero interest rate environment, inflated assets and you had the incredible frenzy during COVID where not only did you have 0% rates, but you also just had just liquidity from both the Fed and the fiscal stimulus from the government. There was just way too much money floating around that. Inflated assets all over the place. [00:52:50] Speaker A: Right, exactly, exactly. [00:52:52] Speaker B: And so I still think we're working some of that out of the system and that what I think it means going forward is that. And it's hard to say if investors are being more disciplined now in Silicon Valley, but there's big deals people are chasing and those deals are expected to be really big. But I don't know if the days of a company like Brex being valued at $12 billion are coming back. They're not some, they provide like a commodity level service to some degree. Right. There's some innovation there. But like, compared to like what OpenAI is doing, it's not anywhere near as like new. And I even think OpenAI is probably overvalued at this point. Like I don't know if they're going to be able to blow off their [00:53:43] Speaker A: trillion dollar or half a billion trillion dollar valuation. [00:53:47] Speaker B: Right. But I would argue that it is an entirely new technology that has like world changing implications. And I would say that like Brex [00:54:00] Speaker A: does not represent it. [00:54:02] Speaker B: So like not, not, not on that level. Right. So there's nothing wrong with that. It's a great company but they're not like a world changing. Yeah. Technology 100. [00:54:11] Speaker A: So another, if we try to like project it even further, it could be a conversation that I'm facing to like SaaS companies. So like these are stock terms. Right. Figma IPO was late last year, mid last year. Stock is down 76% since. [00:54:31] Speaker B: Yeah, it's. I was looking at this today. What is the valuation? It's still valued at like 17 billion or something. [00:54:38] Speaker A: That's right, yeah. Market cap. [00:54:40] Speaker B: Right. [00:54:40] Speaker A: 17 so 14 and a half billion dollars, which, you know, I'm still got. [00:54:46] Speaker B: So glad you finished. I'm there. [00:54:48] Speaker A: Oh, it's, it's. This is still kind of like, you know, insane validation for a company that's not. There's no bolt around the company anymore. Right. AI is literally eating all of its lunch. And the fact that somebody, some bank gave it this valuation at the beginning is crazy. Crazy. [00:55:12] Speaker B: Well, if, if Brex is worth 5 billion in the market today. [00:55:17] Speaker A: Yeah. [00:55:18] Speaker B: Is it? Wouldn't FIGMA be worth something closer to that? That's why I brought up the number. Like 17 billion. Still sounds high for figma. [00:55:26] Speaker A: Right, right, right. [00:55:28] Speaker B: Like, I think, like, I think, like, I think revenue wise, size wise, it's comparable and like innovation wise. Right. Like, this is an interesting company. They have really cool technology, but it's not a game changer, like the way anthropic and OpenAI are. [00:55:44] Speaker A: Right. [00:55:45] Speaker B: And so I still think this company's like way overvalued actually, if Brexit is the new mark. Right. If you try, like, hey, SaaS companies leading unicorns that have been around for 10 to 12 years, 15 years, this is what they're going for. This thing's still too high. I still think it could be a good buy, it could be a great company, but it's just expensive. [00:56:08] Speaker A: I don't think. So. [00:56:09] Speaker B: You're not a believer? [00:56:10] Speaker A: I'm not a believer anymore because, like, really, this is the disruptive force of AI. It's just, it's put a, put an idea in people's mind that these SaaS companies has nowhere to go. [00:56:24] Speaker B: Yeah, a lot of people keep saying that. I, okay, I'll take that aside. [00:56:29] Speaker A: The stock market believes this, right? [00:56:31] Speaker B: Yeah. [00:56:31] Speaker A: Yeah. [00:56:31] Speaker B: Oh, my God. I, I, I, I even, I, I Even bought some SaaS stocks yesterday because I looked at it and said I think it's bothering me out. So obviously do not do what I'm doing. Do your own research. Do not follow any of my advice at all. And I make money and I lose plenty of money in the stock market. So I'm not a stock genius. Just preference it all with that. I looked at it and I said, I think the market's totally wrong. I think that these big public SaaS companies matter a lot. They have big brands, they have big sales teams, they have relationships, they have security and privacy and audits and all of these other things that still matter in SaaS. And maybe vibe coding is going to eat up engineering, which would only be good for them. It would lower so here's the optimistic it would lower their engineering costs. And I don't think that like some vibe coded tool is going to come along and dethrone all of These guys, like ServiceNow, Salesforce, Workday, you know, just some like some intern vibe coding something at some bank is not going to get them to not use any of these like software. Like it's not, it's not, it doesn't pass compliance, it doesn't have like third party auditing, it's not secure. Like there's a million reasons why. So I think the market is totally wrong. This idea that like AI is going to just destroy Adobe and destroy Figma and destroy every single public SaaS company out there, I just think that that's, and I think that like it's, it, it, it's peaking right now and I think that people will actually quickly figure this out, that there will be areas that get disrupted. Vibe coding is, is real, people will spin up their own apps. But like it gets limited in terms of impact to these large public SaaS companies. And I think that that's maybe a bit of a contrarian view at the moment, but that's what I think in some sense. [00:58:31] Speaker A: Yeah, I agree that vive coding is not disrupting SaaS as most people think there. [00:58:40] Speaker B: So on that scale, yeah, on that scale it could be good for them. Like I said, it could lower engineering costs. It might actually be great for the industry. [00:58:48] Speaker A: So one thing that I actually have heard people say is the fact that companies like AWS and Google and Microsoft and Oracle, the infrastructure providers, because of AI, they rather invest money into building AI infrastructure rather than building more servers that traditional SaaS companies need. And that could actually affect the fundamental bottom line of SaaS companies. [00:59:22] Speaker B: How would that impact them? [00:59:23] Speaker A: You can't provision more servers. [00:59:27] Speaker B: They're going to limit, limit them from provisioning it. [00:59:30] Speaker A: You literally. So for example, hard drives, Hard drives is literally a commodity, a rare commodity right now. So if you're a company like Adobe who stores petabytes of video data in the cloud, you are actually competing with AI companies on data storage. And if AI companies are paying, let's say twice as much per gigabytes of storage, your bottom line goes up. [01:00:00] Speaker B: Yeah, but that's just a cost. It's just, it just means like they might have higher cost and storage or something. It doesn't mean that like their business goes away. [01:00:06] Speaker A: Of course. Of course not. Right, but the stock market puts that into perspective. Right? [01:00:12] Speaker B: I think the market is wildly, I mean, Mr. Market is always wildly emotional. I'll give you another, again, another good example. And I, I keep losing money on Uber, but I think Uber is, is like yeah, I, I, I think that everyone's wrong. So I had a big, was I talking about this? I had a big debate with a, a investor. I know this venture capitalist. We were riding Bike Scout. It was a while ago and I told him like, I think, I think Uber has a lot of potential be big companies. Like no, it's going to get completely disrupted by Tesla Robo Taxi. And yeah, I'm like, but dude, that was like two years ago. Like the Tesla Robo Taxi isn't anywhere close to disrupting anything, right? And so it's an existential threat and it's a possibility. But I think the reality is that you're going to need a company like Uber to do the fleet management. I also think that just like the marketing, they've got all these apps installed out there and they can actually drive a lot of demand. We'll see if that matters for Waymo. I think in a market like in San Francisco, yeah, sure, everyone's going to download the Waymo app and use it and maybe LA and maybe even, even Seattle. They're supposed to be testing it in Seattle. Hey, if you want me to test it, I'm available. But I think that like once they try to scale to like non tech addled people that spend all of their time, effort and energy like pumping up their Twitter following, that demand is going to matter. That like being able to tap into Uber to like get to get people to buy rides will matter. That like there'll be a limit to like how many people just do word of mouth and PR that they can get to download the Waymo app. And so I think that like Uber has a couple of advantages over some of these other players. But the stock doesn't trade like that. Like we showed like a couple weeks ago, Tesla's at an all time high and Uber is kind of like, you know, kind of trucking along. Right? So the market for the last five years, it's okay. [01:02:10] Speaker A: It's okay. Yeah, it's okay. [01:02:12] Speaker B: It can be doing better. Like I said, I've lost plenty of money on that stock. I can't ever seem to make money. The world size we're trying to sell it and I get back in, I saw it again. I'm like, I still think that like I'm gonna figure this one out, but I seem to never, I seem never, I seem to never figure it out. It's keep taking the pain but, but I won't give up. I still believe in my thesis that like Uber will actually matter significantly in terms of like rideshare. I still think that the market for them is big, that they're expanding. They're growing. There's a lot more demand that they can capture that I don't think they've tapped out the market any anywhere. So I'm big, I'm a big fan and big believer in that. And they, they do have a partnership Waymo in some markets already. And so you know, I saw the CEO Waymo year, year ago, year and a half ago talk about this. They were actually very explicit about having partnership with Uber and they talked really positively about it. So there you go. All right. Anything else left in our doc or what are we going to do? We're good for the day. All right. Awesome chat with you today. [01:03:25] Speaker A: Likewise. Okay. [01:03:27] Speaker B: All right, take care.