Episode Transcript
[00:00:00] Speaker A: Yes. I'm curious to see what it looks like.
[00:00:02] Speaker B: Okay, so technically, I'm live.
[00:00:04] Speaker A: Okay.
[00:00:07] Speaker B: It should post under my account.
[00:00:10] Speaker A: Okay, I'll go to your account.
[00:00:12] Speaker B: So apparently, I could also let you broadcast this.
[00:00:21] Speaker C: Yes.
[00:00:23] Speaker A: Oh, yeah, I see it.
Cool.
And how nice does that work? So chat works now. We can see what's going on.
Do I need to. How do I broadcast? Do I share a link, too?
[00:00:48] Speaker B: Yeah, you can share that link next time. Oh, here we go.
[00:00:52] Speaker A: Where's the link?
[00:01:00] Speaker B: I'll share with you in the private chat here, so you can share that link.
Go directly there.
Cool.
[00:01:11] Speaker A: Will this link be the same when we go live again, or is it different each time?
[00:01:14] Speaker B: It might be different each time, but. Okay, so the.
What's really cool is I can pair you up if I add you.
[00:01:31] Speaker A: All right. Posting it.
Cool.
All right. I posted on my.
On my account.
So I can just put this Link, like, on LinkedIn and Twitter everywhere, and then people can just use that to watch that way.
[00:01:51] Speaker B: Yeah, so there's also a LinkedIn. LinkedIn one.
[00:01:55] Speaker A: Okay, cool.
[00:01:57] Speaker B: Here, put that here.
[00:02:01] Speaker A: So that's the LinkedIn one.
[00:02:03] Speaker B: Okay.
[00:02:05] Speaker A: I'll do that later. So will the links be the same when we go live again?
[00:02:11] Speaker B: I guess. Yeah, we'll just stay on.
[00:02:14] Speaker A: Yeah, we could just, like, I could just go. Turn the camera.
I could just go mute. And then.
And then when we're ready to go live, let's turn on.
[00:05:34] Speaker B: Sa.
Sam. Sa.
[00:07:30] Speaker A: Hello. Hello.
[00:07:31] Speaker B: Hello.
[00:07:33] Speaker C: Hello. Hello. How are you?
[00:07:36] Speaker B: We're live.
[00:07:38] Speaker A: We're live. But we need to post a link in order for people to see it. Right.
Because, like, actually, even Arden could post a link on his LinkedIn and stuff too, and it would.
[00:07:53] Speaker B: Exactly. Do you want.
[00:07:57] Speaker A: Yeah, those are the links I get. Let me do it, too.
[00:08:01] Speaker C: Cool.
[00:08:03] Speaker B: As you can see, we're still figuring this out.
[00:08:05] Speaker C: Yeah, all good.
[00:08:08] Speaker A: I think the backstage part is fun. A lot of podcasts do that now. Like, you know, kind of banter in the beginning, and then you kind of like, in 10 minutes.
[00:08:18] Speaker B: Yeah, exactly. Do a little bit of intro. Talk about cool stuff.
[00:08:24] Speaker A: I even wrote Arjun. Nice formal intro.
[00:08:28] Speaker B: I gotta thank you.
[00:08:29] Speaker C: This will be fun.
[00:08:34] Speaker A: I'm telling everyone we're going live at noon, so we can kind of do a little backstage and then.
[00:08:39] Speaker B: Oh, interesting.
[00:08:40] Speaker C: Okay.
[00:08:41] Speaker A: Right.
[00:08:45] Speaker C: Yep. Sounds good.
[00:08:51] Speaker B: Yeah, I'm still trying to figure this out. See if we can have, like, a pre show.
[00:08:57] Speaker A: Yeah, exactly.
[00:08:58] Speaker C: Chat.
[00:08:58] Speaker A: Pre show for. For 10 minutes. So, yeah, we started this as, like, an X base, and I was really into trying to make the X base work, but really tried technology just like, it would. It would like stop working. And then I was the host, so I couldn't log back in and like restart the space off. And then Paul's like, where are you?
[00:09:18] Speaker B: I'm like, everything stopped.
[00:09:21] Speaker C: Yeah, exactly.
[00:09:23] Speaker A: That was quite funny, actually. I was like, oh my God. I can't restart it because normally I go on someone else's face and that always would happen. Right?
[00:09:30] Speaker C: Right.
[00:09:31] Speaker A: Oh my God. I'll put my headphones on.
And then. And then we had like, I guess we. When we started doing the live stream like this, I felt like the conversations are better because you can see the other person and there's just a lot of information.
You can see when they're ready to talks. You don't talk over each other. Like when it's audio only. That might be what holds back a lot of these audio only platforms like Air Chat and all that stuff.
The like visual context. Yeah, yeah, it's. It makes it.
It's not as. It's not as good of a conversation.
[00:10:13] Speaker C: Agreed.
[00:10:16] Speaker A: All right.
This looks good. This is different than what we used last time, Paul.
[00:10:26] Speaker B: Yeah, last time we just use the zoom. We just recasted zoom.
[00:10:31] Speaker C: Yeah.
[00:10:32] Speaker A: Platform. I see. Oh, my God. Dude, this is gonna be so professional.
Almost.
[00:10:40] Speaker C: Yeah, that's great.
[00:10:41] Speaker A: So much fun.
[00:10:43] Speaker C: Margin.
[00:10:43] Speaker A: Are you at home today?
[00:10:45] Speaker C: Yeah.
[00:10:48] Speaker A: I. Yeah, my home setup is much more comfortable. I broadcast last time from 90 down in Syra, down in downtown Seattle.
Yeah, like, nice. My home set up like just the office and the lighting and I got my monitor and books I never read. And you know.
[00:11:07] Speaker C: Yeah, all the good things.
[00:11:12] Speaker A: I swear there's someone who gave me this book and I read most of it and I totally forgot about it. And then I post something like, I haven't read a book in years. And they literally called me out on Twitter and were like, you said you read my book. I'm like, I actually did read it. I totally forgot.
[00:11:26] Speaker C: Got called out. Yeah, exactly.
[00:11:29] Speaker A: I felt bad because I was like, I. I actually did read it too. It wasn't like bull. Like, I did read the book. And then I'm like, oh, dude, sorry.
[00:11:36] Speaker C: That's funny. I have in general, so many unread books. No, no, I just have so many unread books right now. I feel like there's. There's a term for it in some language where you collect. I think it's in Japanese. They have a specific term for folks who buy books and they are left unread. So I'm gonna go Find that later.
[00:11:54] Speaker A: Someone said, that's the new signing up for your sub stack, dude. I subscribe. I never read any of it, but I love that you do it. I have them all in my inbox, ready to go someday.
[00:12:07] Speaker C: Sundoku is the Japanese term for habit of buying more books than one can read and letting them pile up.
[00:12:14] Speaker B: Nice.
[00:12:15] Speaker C: Yeah, nice. That's a good one. I'm definitely in that category.
[00:12:19] Speaker B: Yep.
[00:12:20] Speaker A: I always felt bad because I'm the kind of nerd that would, like, really read it and it would happen a lot, or someone would give me a book and then I would read it and then it would come back with, like, a lot of opinions, and I could tell that, like, maybe they hadn't read it or something. And then I'm like, oh, yeah. Then on page 27 and then on 52, and I disagree with this and this and that. And then. Then I felt bad. I was like, oh, like, I really. You were just giving me that as a gift. I see.
Yeah.
I'd see that work. Like, people give you, like, books at work all the time. And then I would, like, read them. Yeah.
[00:12:54] Speaker C: When I was young, actually read them.
[00:12:56] Speaker A: The management would give a great thing to do the whole thing and be like, dude, you guys did. You got.
[00:13:01] Speaker C: They're like, what? You actually read the book?
[00:13:05] Speaker A: Oh, God help me.
[00:13:06] Speaker C: No, it's great. That's a great life lesson in there. If you're a young and ambitious professional, read the books that they give you. It's actually good for you.
[00:13:17] Speaker A: If I give you it, I. I've read it. Like, I wouldn't give or I tell you, like, I did not read this, but a lot of people have said it's a.
That's actually a good idea. I want to work on a reading list. I did have a few people request that. And then I realized I hadn't read a lot of, like, new books. All things I was talking about were quite old, so I gotta find a few.
A few new ones.
[00:13:38] Speaker B: New ones.
Sahel's book is really good.
[00:13:42] Speaker A: Oh, yeah.
[00:13:43] Speaker B: Really, really good.
Which one is this from Sahil.
[00:13:48] Speaker C: Oh, yeah. Yeah. Huh. Yeah, he's great.
[00:13:50] Speaker B: So good.
[00:13:51] Speaker A: I got this. But it's not really a book.
It's more like a journal.
[00:13:56] Speaker C: What did you get get done this week?
That's good. Cool.
[00:14:03] Speaker A: Yeah, it's got some quotes and stuff.
[00:14:05] Speaker B: Right.
[00:14:05] Speaker A: God, I was like, I love that whole, like, what'd you get done this week?
[00:14:10] Speaker B: But it's technically not a slightly of a meme on Twitter now.
[00:14:16] Speaker A: It's got Quotes and stuff. It's kind of like reading Twitter. It's got like little things which is, it's kind of fun. It's good for me. I like it.
But yeah, I need to find some new interesting books. So. Sahil's book is good, really good.
[00:14:33] Speaker C: Yeah.
[00:14:34] Speaker B: I usually don't.
[00:14:36] Speaker A: Yeah.
[00:14:37] Speaker B: What else?
[00:14:40] Speaker A: Yeah. You have other recommendations?
[00:14:44] Speaker B: I don't know. Buy back your time. I read that for the first time this year. That was really good by Dan Martel.
[00:14:50] Speaker C: Oh yeah, Dan. Dan's an old buddy. He's great.
[00:14:54] Speaker B: It's awesome.
[00:14:56] Speaker A: I still like old books. There's some like classics that I recommend to everybody.
I think the number one, like classic, let's say biz book I'd recommend is Andy Grove, like from intel is Only the Paranoid Survive.
That's another book where like I feel like it's, it's really famous, but I feel like a lot of people have never read it or something because like all the advice in there is really counterintuitive to a lot of things that people say that are popular.
[00:15:22] Speaker C: Right.
[00:15:22] Speaker A: Yeah. I've written a lot about like he believes in one on one meeting. Should be at least an hour, huh?
[00:15:30] Speaker C: Yeah, a lot of stuff.
Yeah. That's good.
[00:15:33] Speaker A: Yeah. A lot of people out there pushing the whole like, like everything is, is about efficiency.
[00:15:37] Speaker C: Yeah, yeah.
[00:15:38] Speaker A: Five minute meeting, like all this stuff and.
[00:15:42] Speaker C: Yeah, yeah, right.
[00:15:44] Speaker A: Like, and he actually, he goes into detail about why he thinks a short one on one is, is not good.
[00:15:50] Speaker C: It's not good. Yeah. I think that if you're doing one on ones all the time, then you have to keep them short. But if you are doing a true one on one where you're actually connecting about someone's career and longer term, you know, just strategy and what they want to do with their life, then you can't do that in 30 minutes, frankly. That just doesn't work.
[00:16:08] Speaker A: Yeah. He also is saying that it, it forces people to like only discuss things that can be discussed.
And so there are arms that are much larger and they never get addressed because you're.
[00:16:22] Speaker C: Yeah. You're forcing everything it did a short period of time. Yeah, that's right.
[00:16:27] Speaker A: Yeah.
There's also some good stories about like the early days of intel, like they've probably been around a while and how they were in like the memory chip space before they got into CPUs and how that transition happened.
And like it's so old. Some of the stories are great. You know, like he's talking to engineers that like run different like factories like this Stuff, right?
[00:16:50] Speaker C: Yeah.
[00:16:50] Speaker A: Not like SaaS relations.
[00:16:51] Speaker C: Yeah, yeah.
[00:16:53] Speaker A: And like how much time it took and convincing them to like, you know, they would have to change over the plant to build something new. It wasn't like easy to get people to be convinced of like, you know, changing what they were manufacturing and stuff like that.
High stakes.
I seek stuff.
Paul, what's the weather like in Toronto today?
[00:17:20] Speaker B: 55, 60. Around 60 Fahrenheit.
[00:17:24] Speaker C: It's reasonable. Yeah.
[00:17:25] Speaker A: Got warm.
[00:17:26] Speaker C: We got in here 62 in Marin county right now.
[00:17:32] Speaker B: Oh, nice.
Seattle, California.
What is Murray near Seattle? That's.
[00:17:43] Speaker C: Marin is just north of San Francisco. It's the county north of the Golden Gate Bridge.
[00:17:49] Speaker A: Closer than San Francisco.
[00:17:50] Speaker C: Yeah, yeah, very close. So from here to the Golden Gate bridge is like 15 minutes.
[00:17:59] Speaker A: Dude, Seattle summer is already over. We had like a week.
[00:18:03] Speaker C: You're done now.
[00:18:04] Speaker A: 80 degree weather. It was beautiful. And then today I like wore like a winter coat to go to the gym morning. Like, I went outside, I was like, I'm not overdressed.
[00:18:13] Speaker C: Like, this is actually, this is actually the weather.
[00:18:16] Speaker A: It's actually cold. Yeah, it's a little gray today. I hope it'll be nicer this weekend. Like when it. We had some nice weather and then it, you know, it'll get temperamental and rain, but it should be a lot nicer in a week or so.
[00:18:32] Speaker C: Nice.
[00:18:33] Speaker A: You do get a couple of months where it's. It is really. It's like nicer than San Francisco. San Francisco is always cold and foggy too.
[00:18:39] Speaker C: Yeah, it was yesterday. It was freezing.
[00:18:42] Speaker A: Like it is here. Yeah, yeah, exactly. But in Marin, where. So Paul, where he's in. In Marin.
[00:18:48] Speaker B: Yeah.
[00:18:48] Speaker A: Weather is much. It's like north across the Golden Gate Bridge and that place is like. The weather is really nice.
[00:18:56] Speaker B: Interesting.
[00:18:59] Speaker A: Yeah, the. The microclimates of like the Bay Area are fascinating. Like, it's really dramatic, the differences.
[00:19:07] Speaker C: Yeah. Oh, yeah, totally. Even within the city, it's crazy.
[00:19:10] Speaker A: Yeah, like, like 20, 30 degrees. So where we lived in Walnut Creek was really hot in the summer. Like 110.
[00:19:18] Speaker C: Yeah, surprisingly hot. It's kind of eerily cut.
[00:19:24] Speaker A: It's only like 20 miles or something from like, it's not far, but the climate is just like night and day. Different.
[00:19:34] Speaker C: Yeah.
[00:19:37] Speaker A: All right, we got one minute. Now we're gonna go live.
[00:19:39] Speaker C: All right, we'll get. Jump in. Yeah. Or.
[00:19:40] Speaker A: Oh my God, this is so fun.
Okay.
You saw we put in the.
[00:19:46] Speaker C: Yes.
[00:19:49] Speaker A: Yeah. Let's kind of go through some of the. In news items and stuff.
A few Reddit threads. In there?
[00:19:55] Speaker C: Yeah. Oh, yeah, check those ones out.
[00:19:57] Speaker B: A lot of good news this week.
[00:19:59] Speaker A: Yeah, there's a lot of good stuff to talk about.
Ivan wrote us official intro now.
[00:20:14] Speaker C: Nice.
[00:20:15] Speaker A: Kind of kick off the space. It's kind of blabber on. I was like, okay, we're gonna take this seriously.
Intro. I'm gonna write Arjun a beautiful bio.
[00:20:24] Speaker C: That was very nice. Thank you.
That was great.
[00:20:29] Speaker A: All right, we're live, right?
[00:20:31] Speaker B: We're live.
[00:20:32] Speaker A: Excellent. Well, welcome to the Gregor and Paul show where we break down the Latest in startups, SaaS, AI and whatever the Internet is debating this week.
We aim for smart takes. Many times we end up with dumb ones.
We'll hit on a few scorchers and of course we love memes way too much. We stream is live on X and LinkedIn. We upload it to YouTube. Someday we'll get around to putting this on Apple podcast so people can listen to it at the gym. God forbid they would want to do that.
Today we have our first guest, Arjun Dev Arora, who I have known for many years. So I wrote this nice bio for you. Arjun, are you ready?
[00:21:13] Speaker C: Thank you. Appreciate it.
Let's do it.
[00:21:16] Speaker A: Early stage tech investor and managing partner at Format1 in San Francisco. He built and sold Pioneer Retargeter, served as a partner at 500 Startups, who is entrepreneur and residents at Expo.
I met you at Retargeter.
[00:21:36] Speaker C: Yes, many, many years ago.
[00:21:39] Speaker A: Let's see, you've been as an angel investor. Early checks have gone into Flexport, Notion Virtual Health Branch, Firefly Aerospace, Comm, Tonal Pipe Grove Collaborative. And then you've had exits with Teachable, True Bill and Starship.
Very impressive. And then of course you also back funds including Streamlined Ventures, the house fund, Suse Ventures, Caffeinated Capital. I have to meet these guys. Like I'm shocked. I don't know them.
And what is the last one we got here? Rocket ship VC Beyond Venture. You've chaired $100 million Healthcare foundation and now sit on the board of the SF Jazz center, which I love. Huge fan, been there many times and you've been honored by the White House and the UN and we are so glad for you to be the first guest on our show.
[00:22:37] Speaker C: Yeah, thank you both excited to be here and yeah, I'm looking forward to diving in on all these fun topics. So thank you both.
[00:22:43] Speaker A: Oh my God. Yeah, we got a lot of stuff to discuss this week.
So Paul, should we kick it off with some of the nerdiest news out there?
Arjun to weigh in.
[00:22:55] Speaker C: Sounds good.
[00:22:56] Speaker B: Absolutely, yeah. Big news in AI this week.
The first thing was a release of a paper called the Illusion of Thinking by Apple.
Essentially, this paper outlines from Apple's perspective on whether if LLMs are conscious or not or are they able to go about and reason.
[00:23:24] Speaker A: Your dogs bark in there, buddy.
[00:23:28] Speaker B: Sorry, guys.
Essentially, they published a paper talking about if a.
I'm sorry, guys.
All right, you want to. You want to take over this one?
[00:23:43] Speaker A: Yeah. So, Arjun, did you see this? It was this paper that the Apple team put out, which. There's two ways to talk about this. One is, it's quite funny that Apple, which appears to be behind in the AI space, is putting out papers debunking AI. Right. My tweet was like, look, is this what the Siri team has been working on instead of making Siri better? Because it doesn't really work for. For me at all. Right.
Which is. Which is one interesting kind of way.
[00:24:16] Speaker B: To look at this.
[00:24:17] Speaker A: On the other hand, I do think they made some great points that like, you know, perhaps AI can't, or we're not in the realm of achieving something like AGI.
Arjun, what are your thoughts on this?
[00:24:32] Speaker C: Yeah, it sounds like with the paper, they were. They're basically saying that, you know, accuracy collapses under complexity, that, you know, as problems become more complex, the model uses fewer reasoning tokens. So it's basically saying that it's quitting and not actually, you know, seeing things all the way through. And then, you know, there's also saying that it failed to kind of apply, you know, things that actually worked or understanding kind of the algorithm.
You know, even when they knew what the algorithm was to work on the problem, they chose not to use it, or they meaning the LLM in this case. So these are kind of the core, you know, findings from.
From the paper. I, you know, my sense is that ultimately, you know, it is clearly working. People are using it. You know, they're finding real value from it. So I think, sure, there are structural reasons why, you know, the algorithm LLMs could be improved. But ultimately, and maybe this is my bias as being an operator and investor, I'm looking for ways in which folks are using it to improve their lives or solve problems. And obviously, and clearly LLMs are doing that. So this feels like a bit more of a technical, nuanced, you know, well, what about ISM sort of thing. And while, again, I appreciate the research that went into it, it feels like they're kind of missing the force through the trees on this one.
You know, people are using LLMs, they're loving them, they're solving real problems, they're helping humans, they're, you know, doing the right stuff. So I think not acknowledging that in this paper, you know, kind of makes it feel a little bit tone deaf. But I do appreciate the technical depth and rigor of the paper and why it was necessary to, you know, share some of these nuance points around how exactly, you know, the LLM algorithms are working. Sure. So if that helps us improve things going forward, great. I don't think anyone's walking around saying these things are conscious by any means today. So you know, consciousness is a whole other rabbit hole of a topic, but we're nowhere near that.
So that, that alone is way, way out there. But anyway, you know, they've gotta, they gotta start try and stay relevant. So kudos for their attempt there.
[00:26:43] Speaker A: Paul, what do you think?
[00:26:44] Speaker B: Yeah, 100 right. Well like benefits Apple to.
I'm not gonna take a dig at Apple, but they're, they're a little bit behind on the curve in terms all of the LLM research 100% agree with the take on consciousness.
I think that's a completely other separate discussion.
I don't think we're there yet.
The interesting test that they threw LLM or the interesting test the paper through the LLM behind is that so they gave it a really long running task, let's say like a game with a rule that has 10 plus steps.
The problem with LLM right now is that with every single step it's kind of like a dice roll. That's how I look at it with like inherent problem chance that it just doesn't work around let's say 70 or 80% chance.
So after 10 steps the LLM could be as far as off 95% chance that it's completely off the course and then it gives up.
I think that's really interesting property of the LLMs right now, which means that you'll always have to have some kind of intervention. Maybe it could be another agent LLM that's always correcting its course or human intervention.
So kind of like from an engineering perspective, I'm wondering what the best way it is to change and use these in a long running task that's not just a one off chat message or prompt.
That's my biggest takeaway from this paper here.
[00:28:37] Speaker A: Yeah, go ahead Arjun.
[00:28:40] Speaker C: No, I think to that point I think there are other ways of, of training that may, you know, help solve that in longer run as we look at different models or methods for training. But yeah, I think it's still a very complex problem to say, okay, as we're going down this thread and the probabilities have decused over time, how do we one know that and understand it and then course correct without, without giving up and continuing down that kind of probability tree in the right way. So these are, yeah, those are incredibly complex problems that OpenAI engineers are working on. But I think there are some new AI technologies are starting to bubble up around, you know, training at scale and kind of having evolutionary methodologies for training that are, that are really interesting that may start to solve some unique, some cases of some of these problems, but won't be generally applicable.
So just I think reinforces the interest again maybe more from an investor perspective. But there are vertical AI, you know, companies that are building their own models or training their own models for specific subsegments where they can potentially, you know, reason more thoughtfully within more, you know, confines of, with certain boundaries. Right, right. Not where they're limited into kind of a general intelligence perspective, but are limited by balance of, of the vertical they're playing in. And within that context there may be ways to, you know, speed this up even faster and maybe in some universe they, we start to stitch those vertical things together in kind of almost, you know, meta agent driven process that would search each of those vertical foundation models and then use that to find the solution. But these are all for now just thought exercises and the market has to mature enough for, for that to be able to be possible.
[00:30:22] Speaker A: What was Apple's motivation to put out a report like this? I think that's where a lot of people were confused. Not that I disagree. They said I, I actually agree 100 with what their ultimate, ultimate conclusion was that and I read it as like the paper trying to debunk AGI that like, which I believe like we're not close to creating AGI. And I even wrote some notes where, and I've heard this before, but it's not popular to say the, there's people way smarter than me who made the point that computation is not the same as consciousness. And there's a belief that AI is some ultimately just advanced form of computation and that the debate is like whether lots and lots of computation does consciousness emerge from it. And I think this paper says what I, what I agree with is that it doesn't that computation and consciousness are two separate things. We don't actually understand what conscious is at all. Which is a very like philosophical conversation. Right.
And very academic.
I liked when you started. Arjun was like hey look, people are getting lots of value. These tools work. They're great. AI is awesome. There's vertical AI. There's all kinds of stuff happening. So why would Apple put something like this out? Is it just like an academic paper that perhaps they uploaded and caught and at least just caught the attention of people on Twitter or what do you think?
[00:31:48] Speaker C: Yeah, that last bit sounds like it's probably the most reality. I can't imagine the Apple PR team is putting this out with the intent of saying, hey, this is our formal stand. So it's, it's likely more of a paper that caught, you know, caught fire and, and then, you know, turns into this conversation. So that would be my sense. But hey, who knows?
[00:32:07] Speaker A: Yeah, I don't think it'll get beyond the nerd world in this podcast, but it's still really interesting. Like, I loved, I love that they put it out. Actually, I thought that was quite cool and I was reading some of it. I just was like, this is wild. Like, it's really interesting to see that they, that they did that.
[00:32:25] Speaker B: So I think this is actually a pretty good transition to the next one. And to answer your question, I have a hinge take, but maybe we can dig into Skill AI acquisition first and tie it back together with Apple take.
Originally, we weren't going to include this, but I think it's a pretty big announcement that Meta announced a 49% acquisition of Scale AI.
It's actually Scale AI is a.
They've been in the AI game for a really long time, since I would say 2019 or 2017, around that time scale, which is super interesting. I think this acquisition, because Meta, since the launch of their Yama Model 4, has been kind of falling behind on the foundational model curve.
So with this acquisition, I think they are trying to recover internally on a new direction of their research.
Yeah, I will. I would love to see like, where.
[00:33:42] Speaker A: How, how does this like, help them get, get ahead? I agree that they're falling behind. You know, they took the open source route. But how, how does this acquisition, you think, help them?
[00:33:50] Speaker B: So personally, I think they are, they are falling behind on attracting new talent.
It's, it's all, it's a complete talent pool.
Companies are like, OpenAI are attracting talent under the guise of we are almost reaching super intelligence. And META has for the very long time kind of refused or danced around the notion that superintelligence AGI is around the corner.
[00:34:27] Speaker C: Yeah, I think one of the other things that's important here is that the Scale AI understands data. Right. So the data moat that gets Created when Facebook has access to all the, you know, proprietary or semi proprietary data that SCAL has been able to aggregate and understand for AI use cases also creates, I think, another point of defensibility over time. And you know, if in, in any, you know, you're only as good as your training data in, in, in any AI universe, and so this I think gives me a competitive advantage in that regard as well, then certainly agree on the talent side, that definitely makes a lot of sense and also just a continued commitment to, you know, the vertical. So I think sharing, you know, with the street and other folks that we are, you know, committed to AI in a meaningful way and here, you know, we're taking active bets to help us accelerate in that direction. Particularly in light of, you know, other kind of publicly traded companies doing similar things or, you know, setting themselves up to do similar things. It's important that they make sure the street knows that they're not only talking about it, but they're committed to it. Obviously they've already invested a tremendous amount, but this just continues to show that, that they're, they're there.
[00:35:39] Speaker A: Interesting. Paul, is there any specific technology that scale has that Meta needs or you think it's just talent?
[00:35:46] Speaker B: I, I think it's purely talent.
I, I think they, I think Meta has a problem of attracting talent.
That's, that's my complete.
Some crazy numbers throwing around are they're throwing like $75 million to $100 million at top tier AI talent. That's that like, that's crazy.
[00:36:12] Speaker A: Yeah.
[00:36:15] Speaker C: Super intelligence team. They're throwing that money at those folks.
[00:36:18] Speaker A: Yeah, yeah, I guess, I guess if you take that into account.
Right. So that coupled with this shows that like Meta is clearly willing to spend and to try to get their.
Whatever they need in terms of resources and that. And that for some reason they don't necessarily have the firepower that maybe OpenAI or some of the other players.
[00:36:43] Speaker B: Correct. And I think for the longest time they kind of danced around the notion that even the entire concept of LLM was a thing because if they're last head of AI or you know, so like, I think that's in terms of marketing. I mean, this is a marketing podcast. That's a, that's a very tough stance to stand behind in terms of tech, to be one of the large companies to say that LLM is not the future of AI.
[00:37:15] Speaker A: I mean if you're just like, like at the highest level way I would break it down and the way I see it right now is like OpenAI is clearly the leader and everyone I talk to, like. Like, I know customers right now that say they're getting significant amounts of traffic from ChatGPT, which none of the other platforms are doing. That was like a eye opener, maybe. Like, I only started hearing that six weeks ago. And is the new number a billion users? Is that correct?
[00:37:42] Speaker C: That wouldn't be. Yeah. I don't know the exact number, but that doesn't shock me.
[00:37:45] Speaker A: Like, holy moly. Like, that's, like, enormous. And, like, they're. Their tools are fantastic. Right. So you're looking at, like, Claude, which has been able to build this very cool niche.
[00:37:56] Speaker C: 400 million weekly actives.
[00:37:58] Speaker B: Sorry, just.
[00:37:59] Speaker A: Who's that? Who has 400.
[00:38:01] Speaker C: 400 million weekly actives for Chat GPT.
[00:38:03] Speaker B: That's a lot.
[00:38:04] Speaker C: Yeah, I just looked at.
[00:38:05] Speaker A: Yeah, so like, it's a billion accounts or something. I mean, the numbers are just astonishing, right?
[00:38:09] Speaker C: Yeah.
[00:38:10] Speaker A: And so, yeah, you've got Claude in, like, I'm gonna say second place.
I'm not convinced that Google's in third and second place with, like, AI mode and all that. It's kind of convoluted. I mean, you could probably spin it some way, but I think closet, when it's like, built a business and people are using that regularly.
Gemini, and those things are interesting, but they've got a bunch of conflicts in terms of, like, how they're executing to their user base. And then, like, you know, Facebook, they put that AI bar on all their apps. But, like, I've never used it actually. I've never ever used any of those apps to do any kind of, like, AI stuff. So I can see how from the, like, Facebook perspective, they're like, we're behind. Like, people are not using our tools in the same way that they're using other tools. Maybe they have a whole different way to go about using AI, which is kind of what I would argue. But they don't want to give up, right? They don't want to, like, be left out of the professional tool space.
If I was them, I'd be just acquiring everyone. Right. Just, like, whoever I get my hands on. So scale AI, Any of these vibe coding tools, like that meta would have a suite of developer tools. Seems really logical. I think a lot of developers would use it.
[00:39:22] Speaker B: Interesting.
Well, Zuckerberg actually came out and said that he wants to make AI best friends for everyone.
[00:39:31] Speaker A: As friends.
[00:39:33] Speaker B: AI, like, best friends for everyone.
That's his vision of AI that also.
[00:39:40] Speaker C: Stitches everything together in some way. He's bringing the metaverse coupled with bringing AI friends for everyone, coupled with AI investments so that, you know, there's a cohesion in that narrative. Now, I don't know that I necessarily want to have all my best friends be AI, but, you know, that's a big thing. I get the intent. I get. I get where he's going with that.
[00:39:58] Speaker A: Yeah, it's. Everyone says that, like, I think maybe our age and stuff, like, we have this reaction. We're like, God, I don't really want an AI friend.
But what I, for me, what's an unknown is do, like, non Twitter nerds care. Right. If you go on Instagram and it's a bunch of, like, AI stuff and you're 20 years old, does it even matter to you? And I also think that, like, the average person who's using Instagram, let's just assume that they're so desperate for, like, interaction and engagement that I don't think they care if it's AI. And I think that that's what's motivating. And. And, like, I think that's what, like, Zuckerberg knows, that they've got this giant platform of people.
They're not AI experts, they're not developers, not people like creating tools, they're not business users. Right. They're just like regular people. They go on there, they post their family, and there's a bunch of AI stuff happening. Like, they might even like it. It might even be something that is enjoyable to them. So. And I just feel like I'm not a good person to even judge this.
[00:41:02] Speaker B: Yeah, yeah.
[00:41:05] Speaker C: Well, one thing we know is he does have the data, so I'm sure he knows and understands how people are engaging with their real friends and then what the implications are for how one might engage with a AI friend. So if you've got all the data around how people engage, how friendships start grow, change, how they take photos together. I mean, the. The amount of information that Facebook has and Instagram has on relationships writ large via proxy through WhatsApp, Instagram and Facebook, I mean, they could craft the perfect friendship over a period of time, knowing that data just for you. So it gets. Gets real interesting real fast.
[00:41:42] Speaker A: Has anyone tried any of those, like, friend AI? There are those devices and apps like I have no interest in, so I haven't tried it. Like, my wife won't even let us get an Alexa. She's like, we're not doing that.
So, like, I already, like, I've made my bed and I'm lying in it, but I'm curious about how you guys have tried that or no, One has used any of those things.
There are like friendship apps. Right.
[00:42:03] Speaker C: The closest thing I've used is Tony AI, which is a conversational, like I can call it and talk through a particular idea and it'll synthesize it into basically a post. But I think that's something that you maybe not necessarily building a relationship with, but it does get to know you over a period of time. So something like that is, is, you know, certainly interesting, but it's not necessarily friendship. And I guess, you know, being the business nerd that I am, I'm talking to AI about LinkedIn posts. But you know, there, there might be other things to, to, you know, talk about, but yeah, there's there. They're starting to show up in all kinds of different places.
[00:42:44] Speaker A: I would get excited by the use case or like we keep talking about agents as employees and like on your team and that they maybe they have a Persona and you talk to them and you can interact with them multiple ways. They are on, they're on Slack on stuff like that, that, that I can actually like imagine an engagement and a like universe where that's not actually creepy. Like, all right, those are my AI bots and they're on Slack and they're kind of like real people and maybe they even have even like I would give them names and stuff. It would be kind of fun. Like I could see how this could become something that I would fully embrace. Right. So I think there's a future that this stuff all, all plays out. Like, I think it's inevitable it all plays out like this at some point. I don't know when.
[00:43:27] Speaker B: Interesting.
Shall we move on to?
[00:43:33] Speaker A: Okay, yeah. What do we got next?
[00:43:37] Speaker B: Every season YC puts out a Request for Startups list.
This one is the Summer 2025 edition.
I, I thought it's super interesting. If you go down the list of 14 ideas or call for proposals or requests for startups, however you want to frame it, almost all of them are AI startups ideas except maybe one which is software tools for robots.
[00:44:09] Speaker A: Do you see this Arjun?
[00:44:11] Speaker C: Yeah. And I'm just going through this list. I've, you know, I've probably seen, you know, handful of companies working on each of these spaces. So, you know, I would say voice AI. We just talked about that kind of interesting idea, the AI personal assistant kind of tied, tied to that. I think that one, you know, feels like something that Apple could, could knock out at some point in time in some capacity. But they've been pretty particular about getting access to, you know, people's Data and have, you know, had a hard line around getting or, you know, engaging with that access to emails or text messages. So I think that makes it harder, but you've got to find a way to be able to get access to those things to make that truly helpful. Healthcare AI, I think happening, you know, everywhere, particularly on the data system side. That, that piece to the introduction, you know, was involved with the St. Francis foundation and saw some of the inside administrative realities of hospitals. And yeah, there's a lot of work to be done just to help, you know, smooth some of that over.
And that's, that's kind of outside of the actual technologies like RAD AI, which is looking at radiology and actually looking at images and using AI to understand what might be going on. This, I think point is specifically about, you know, automating the healthcare data system so all the workflows that live inside healthcare. I'm on the board, you know, Zentist, which is doing all right payments, but please.
[00:45:30] Speaker A: Yeah, I was gonna say I went to one medical this morning for like a checkup and I had to agree verbally to be recorded by AI.
[00:45:37] Speaker B: Yeah.
[00:45:37] Speaker A: And that they had a healthcare scribe assistant.
I guess it's just a microphone in the room. Like, I didn't have to do anything.
But before you go in, they're like, do you agree to be recorded by AI? And I'm like, yeah, sure. Cool. I'm all for it. Right?
So, yeah, that was super interesting, man. The stuff's already, already out there in the marketplace, right?
[00:45:56] Speaker C: Totally. Yeah. That just helps doctors.
All that time that they need to spend transcribing things on the back end, just immediately done and they can move forward. So that. That's pretty cool. The. The other one that jumped out, you know, AI voice assistance for email. I thought that was pretty cool. Just to be able to talk through my mailbox, you know, while I'm driving. That would be awesome. Get way too many emails.
So it would be nice to be able to process email, you know, in that way. And then on the personal finance side, you know, there's some great tools like obviously Copilot Money and you know, all the stuff from the rocket Internet folks, the former True Bill, which you mentioned as well earlier. I feel like these incumbents have an opportunity to integrate AI into the personal finance stack on already. So if you've built the connectivity and gone through all the hassle of being able to help people plug in all their accounts, the next obvious step feels like they can just integrate AI in, you know, into those pre existing platforms. So that feels like a spot where it's probably better suited for incumbents than. Than startups. But if someone builds a truly kind of AI native experience from. From day one that you can almost talk to about your financial life and still integrates everything via plat or host, however else they do it and makes that pretty seamless, that could be. That could be compelling. It's, you know, hey, I just bought a house, or I'm just, you know, thinking about buying a home and what. How should I think about planning? Or I'd like to save for a vacation. Can you keep me on track and, you know, pay me if I'm spending too much on Apple devices or whatever it might be? You know, that way you kind of get to get the pushback you need to hit your goals.
So that feels like something. Maybe there's an AI native application for that could be interesting. I think those sorts of companies, just every generation we have a really interesting new personal finance business, starting with Mint way back in the day, if you remember that one. But I feel like we've had another interesting one every few years.
[00:47:55] Speaker A: You're definitely spending too much on Apple adapters and cables, I can tell you that.
[00:47:59] Speaker C: You can see it already. I get all the thanks.
[00:48:01] Speaker A: I have like a pile of them now. There's like. And then I got a new laptop and they actually, they went back to some things and like, I was like, oh, my God. Yeah, so we know the answer. I don't need AI to figure that one out.
[00:48:13] Speaker C: That one's easy.
[00:48:14] Speaker A: I thought this was. I had to make fun of this. Like this, this post. I get, like, that they're serious about like all of these categories, but I just thought they were quite funny. Like, it was literally like startup for AI. Fill in the blank. Like, it was just basically everything, right? Like finance, healthcare, full stack AI. Like just everything with AI on it, which is quite funny.
So I couldn't help but like, you.
[00:48:40] Speaker B: Know, it's like we're going back to the Uber 4X days.
[00:48:44] Speaker A: Yeah, exactly. Poke some fun at it. Although I thought, like, design founders is interesting.
Uh, maybe. And I fall into that category. I'm finding, like, a lot of utility with these tools. Like all the graphics I create for this podcast, right, that's all done with AI. And it emerged from, like, me experimenting. And I was like, oh, if I take our LinkedIn photos, can I get AI to like, just put them together? And I was like, what? And then I realized, like, if I added Arjun, it could just put them into the into the, into the design. I was, I couldn't believe it. Like it's amazing and it can do all the text and everything. So there's lots of cool stuff happening out there and I like that they're pushing for that. Like people are utilizing these tools and trying to take it to the next level and produce interesting things. I think a lot will emerge from that. And then on the agent side there is so much cool stuff happening. I've been working with some companies that are building MCP server protocols. How to call another guy that had a startup that they had a custom agent builder. And what they found was like a little early, there wasn't a huge demand. So they built a bunch of like pre built agents and they have thousands of people using these things.
[00:49:57] Speaker C: Yeah, very cool. What are some of those use cases? Out of curiosity, dude, what are some of the agents that are getting the.
[00:50:03] Speaker A: Traction it sounded to me from the short conversation we had was that a lot of them were some form of a chat bot. But like my, my understanding of it is that if you're like a small community, like you have a Discord server or something like that, it's not easy to get a chatbot to monitor and manage that for you. And I guess what he told me was like people are using offshore labor to do that right now. So you've got like a big discord community of 50,000 people.
They're hiring like offshore labor to kind of be the moderator and that they have a pre built bot that can just moderate for you and if it struggles anything it flags it to like a real person. Like it's pretty basic actually. We think about like what it needs to do and I don't know a place where I could go get one of those easily right now. Like if I needed like an AI moderator, how would I do that? So that was pretty cool Use case. Also it was interesting that like custom ones were like not something people were ready for.
[00:50:59] Speaker C: Right, right.
[00:51:01] Speaker A: Yeah. That there's just too much overhead like what I do with it. My guess too is that I think a lot of. So if they go like, I don't know if this is SMB, but they go like small market. I think pre built will be really popular if there's all kinds of just maybe this even goes back to our conversation about AI co workers.
Like it'll be normal for everyone to have like an AI chat bot. You could have an internal, right? Like hey, we got 100 person company, ask the bot. The bot would just know everything rather than having like Assistance and stuff. Right. Is someone scheduled a meeting, why this get canceled? What's happening here? Like there's all kinds of basic stuff.
So the idea of like these AI co workers I think is coming really soon.
[00:51:44] Speaker B: For developers. That would be really helpful. We have a term called rubber ducking which is like you talk to a rubber duck on your desk to debug things.
[00:51:56] Speaker C: Yeah. So like this is a way to talk it out. Yeah, that makes sense.
[00:52:00] Speaker B: Exactly.
[00:52:00] Speaker C: Right, right.
[00:52:01] Speaker B: Exactly. So that would be, that would be really good.
[00:52:04] Speaker C: Yeah.
[00:52:05] Speaker A: Was there anyone here that was surprised? I mean they all seem strict. AI, healthcare software tools, robots, future of education, security. Like they all seem pretty obvious.
Nothing was like, yeah, we talked about.
[00:52:20] Speaker C: The personal finance, residential security was kind of interesting. Is that if you have, you know, the effective AI kind of you have multiple cameras in your house and then it's tracking you versus your family versus random intruders and then can alert the right people. So, so it has an understanding. Or if the neighbor's dog walks by, it's not going to freak out, you know, that, that sort of thing versus other things might, you know, take different action and maybe integrates with your calendar so it knows if people are delivering things or you know, email. So it doesn't, you know, take issue. Or if one of your friends is staying over, you can just talk to it and say, hey, buddy's coming by late at night to deliver something. Don't, don't call the police.
You know, I don't know, there could be something interesting there.
[00:53:04] Speaker A: Huh. That is interesting. I hadn't thought. I, I didn't interpret that way. I like it like, hey, it doesn't turn the lights on when your kids are sneaking home late.
[00:53:11] Speaker C: Or yeah.
[00:53:13] Speaker A: Yeah, you're on vacation, your neighbor comes and it doesn't set off the alarm. Actually there's a ton of use cases.
[00:53:19] Speaker C: Yeah, there's a ton of those use cases. Right. Where you could, like not that I.
[00:53:21] Speaker A: Ever stuck home late and came into the window.
I never did that.
Uh, yeah, I'd be. I, that's like dad mode. Like please don't modify mom that our son homely and yeah, I would, I would totally program that in.
[00:53:36] Speaker C: That's great. See, these are real AI use cases.
[00:53:40] Speaker A: I, I bring commercial grade AI security. I thought they were talking about. Or maybe there's another way to interpret. This is like digital.
I thought it was like family networks, digital security.
And maybe, maybe it's both. Like I actually like the physical real world security too. Like, and like yeah, people need that. But home security like my brother has his own home network and he has a bunch of like stuff with his kids and everything. And so I think there's more of that and there's all kinds of weird stuff that could happen. Like really worried about like hacking and people doing weird stuff.
[00:54:16] Speaker C: Just. Yeah, that's a great, yeah, great point.
I mean ideally something you just plug right into your modem or in something that you know, can watch all the packets coming in and out, just understand weird and then make sense of it. I mean just all the corporates obviously have these solutions endlessly, but there isn't kind of an easy obvious home solution for that. So. That's a good point. Yeah, right.
[00:54:38] Speaker A: Like you could just put it on and so like grandma or whatever, like don't worry Mom, I got this thing so that you're like your phone, everything is all protected and, and like you're not going to get scam calls like that There'd be huge market for, for that. Like it just was like a blanket system made it really simple. It just was like, like a shield around your house so grandma doesn't get scammed.
[00:54:57] Speaker C: There you go. The marketing rights itself.
[00:55:02] Speaker A: Maybe I, I'm getting old enough. Maybe I even need that. Or like.
[00:55:05] Speaker C: Oh, all right.
[00:55:08] Speaker A: That was fun. What, what's next?
[00:55:11] Speaker B: Apple? WWDC 25?
I. I mean it was this week.
I think it ends today. Today's the last day.
Love to get your take on it, if there is any.
[00:55:27] Speaker C: Yeah, it sounds like the big, big reaction was to this kind of, you know, UI that was see through and that's what folks were going. Yeah, crazy about the glass. Yeah. So I, I think there was a lot of hilarity that ensued on, on X around this particular topic. I, you know, don't necessarily have a strong reaction in either direction. Just feels like Apple continues to, you know, innovate and push boundaries on, on UI and ux. And I don't know that it's going to change my Apple experience dramatically, but I can certainly see how it makes for good Twitter fodder. So that was fun.
[00:56:00] Speaker B: Yeah. Yeah.
[00:56:01] Speaker A: Did they launch a new device?
Did they launch a new os?
[00:56:07] Speaker B: Just a new os, New design system, Transparent thing?
[00:56:10] Speaker A: Yeah, like, I don't know. I guess so. I remember when WDC was like the place to be.
[00:56:17] Speaker B: Maybe.
[00:56:19] Speaker A: Was it 15 years ago now? 10 years ago, used to be really big. We used to do events around it and everything and then it kind of faded and then gtc like Nvidia GTC has become that where everybody wants to be all the startups. So it's lost its luster, which isn't a huge surprise. Right. Like they've been kind of milking all the inventions that Steve Jobs is created for years now. And kudos to them. Like they've done a lot of internal things in terms of like making that company so efficient.
It's really impressive.
It's not as impressive from like an innovation or like product perspective or consumer level or like, you know, products that you and I would experience.
So I think that's why it's like perhaps not as interesting to people.
I did, I did personally like watch some of it and I kind of look at Apple and now it's almost like a huge fan of it too for like since, since I was a kid I guess. Right. Like I've been following Steve Jobs since, since I was little, like literally. And they seem like a caricature of what they once were. Like I feel like it's like, it's like if Apple fanboys went and started a company, it would be Apple today. I was watching some of these videos and stuff. It just seems like they've, they've got this very like over polished professional approach that I feel like perhaps we've moved beyond. And the point that I always make with this one in particular is that I remember the day when I saw the first video I've ever seen of Larry Ellison talking and he filmed it on his phone. And I knew at that moment like, oh my God, everything has changed. Like it's not unprofessional to just do like a TikTok style video or just do a live broadcast like what we're doing.
It's totally accepted now in the business world where I think prior to that, prior to Covid, everything, you know, aimed to be television quality, professional production value. Like if you didn't do that, you weren't serious. And I feel like Apple's kind of been left behind. Like they still kind of approach it that way, but I don't think that's how the market perceives things. And so they, they, for me they seem a bit behind the times, which is so surprising for a company that was interesting, so innovative.
[00:58:31] Speaker B: Right.
[00:58:31] Speaker A: And had really interesting like marketing for a very long time.
[00:58:37] Speaker C: Yeah, that's really interesting. I think the, the marketing world to that point has, you know, shifted dramatically towards more of this kind of authentic, you know, connection and with, with human beings versus necessarily with a faceless brand. And we're still early in that kind of transition and there's certain pockets of it where it's taking off more than others, but that it feels like that's not, you know, not at all a game they're playing in any capacity in terms of their, how their marketing is going.
And they've stayed kind of tried and true to their tops down, big announcement style, which again feels very Apple. I think it would be a big shock if they went hardcore influencer only right out the gate. So it didn't have to be a transition over time to integrate more of that. But it feels like to your point, Greg, they've stayed like really sharp on that, that side of the spectrum today.
[00:59:24] Speaker A: Yeah. I mean they've got animators dilemma. Right. I don't, I don't know what they, what I would do if I was them, like, would just go all casual. It would seem super weird. Like Tim Cook shows up with a skateboard is like, hello, fellow kids. Right. It just would be like a meme.
[00:59:36] Speaker C: It just would make any.
[00:59:38] Speaker A: Right. It wouldn't make any sense at all. Like that'd be even worse. I'm like, yo, Tim, just like stay. Just do that. Just keep doing what you gotta do.
[00:59:44] Speaker B: I mean, if, if Zuckerberg could, could have made the transformation that he did.
[00:59:49] Speaker C: Yeah, that's fair. But he did it over like three years. Right. And I think it was very well crafted.
[00:59:54] Speaker B: Did.
[00:59:54] Speaker C: Of course he started, he started a little small and then kind of, you know, made his way to, you know, surfing in a tux and Lake Tahoe on back of a surfboard or whatever.
[01:00:03] Speaker B: Right.
[01:00:03] Speaker C: So, yeah, it takes, takes a while to get there.
[01:00:05] Speaker B: So.
[01:00:06] Speaker A: Fair point, I think, I guess, I guess like, okay, if I, like in a, if I was running Apple or what would be my recommendation for them to like overcome this? They remind me of like classic brands from New York when I was younger and I worked on like cosmetics and beauty and what they would do or what some of them did, just acquire younger brands and just make it opaque. Like no one knows. So I was Apple. I would just try to maybe, maybe experiment, like starting a new brand, find a new thing. Like try to get into like what the kids are up to, but just do it at arm's length. Right. So that it doesn't poison what they're doing. Because they're so huge. They're like an excellent company, the best company ever. Right. And so don't like, just keep milking that, like keep doing that and then try to invest in other brands. I think that part would be at least I'll try it. Right. They've got nothing to lose. They can even do it by. In secret. They could just, like, buy parts of, like, cool, interesting companies that are doing, like, hardware, software stuff and the consumer space. Like, they've got a ton of money, you know, So I don't know, could they, like, maybe, you know, were they. Were they even bidding for TikTok or something like that?
[01:01:16] Speaker B: Yeah, I mean, I think they tried to do that one right. With the acquisition of Beats.
[01:01:23] Speaker C: Yeah, that was a great example. Yeah, but what happened?
[01:01:27] Speaker A: Yeah, they were criticized for it too. They just own it, right?
[01:01:30] Speaker C: No, but, yeah, they integrated it into their stores. You could buy Beats headphones and. Yeah, actually, that's a great. That's a great example. I think that one worked well for them. Yeah. Right.
[01:01:38] Speaker B: They just need to do more.
[01:01:39] Speaker C: Yeah, I think.
[01:01:40] Speaker A: Yeah, I love that. I'd love to see them. Like, I wish they would do more hardware stuff. Like, I know they wouldn't do this, but for some reason I watched this Tetris documentary about these kids who were, like, setting world records on Tetris, which was awesome. I don't know how I stumbled into it. It was really fun. And because I watched that on YouTube. So, like, putting all these, like, indie video game videos and stuff, and people are making, like, all these really interesting, like, devices out. There's all kinds of stuff happening in the hardware space I wasn't aware of. I don't spend all the time in games. I have a lot of free time. But it felt like something like, why couldn't Apple be exploring with that? Like, trying, like, devices that are, like, dedicated to gaming or something? Like, they might be able to carve out a space. The way Microsoft built Xbox.
They did a really good job. Right? Like, they put all this money up front. They hired all these great game developers. They built this cool brand. They did. Exactly. Guess what I'm suggesting. Like, Microsoft brand is, like, not cool.
It's a huge business. It's really successful. So they're like, how do we get hip and cool and do stuff for the kids? We'll create this whole new Xbox thing. And I would say from a high level, they succeeded.
[01:02:54] Speaker C: Xbox is cool.
[01:02:55] Speaker B: It was a success.
[01:02:56] Speaker A: People love it. Right? Like, so Apple could do something similar from that perspective and even have a model like, with how, you know, Xbox did.
[01:03:05] Speaker B: Right? Yeah, that's a great recommendation.
Great. They. They've kind of alienated the gaming community a little bit through Epic Games. They. They just. This whole thing just settled on the fact that, you know, they take 30.
[01:03:21] Speaker C: Yeah, 30 in everything.
[01:03:24] Speaker B: So Epic Games just got the judgment that they're allowed to now charge off of the platform.
[01:03:32] Speaker C: Oh, okay. That happened today? Or okay, wow.
[01:03:35] Speaker B: Maybe a couple weeks ago.
[01:03:37] Speaker C: That was a huge news.
[01:03:39] Speaker A: All right. Okay, so here's my pitch. If I was going to speak Tim Cook, I'd have one slide and I would say, you guys need to stop being perceived as focusing on extraction and start being perceived as focusing on innovation. And you got to get back to being the Apple that you once were. And then everyone left. 101 slide. Thank you very much.
[01:04:01] Speaker B: Where do I send the invoice?
[01:04:03] Speaker A: Oh, tell Tim Cook to call me. I'm on Twitter all the time.
[01:04:08] Speaker B: Cool.
[01:04:08] Speaker A: All right, what do we got? We got one. Do, do we have one more?
[01:04:11] Speaker B: No, I think let's move on to Reddit threads. Okay, so, so what part of our segment is we just go on Reddit? Uh, we try to stick to SaaS or startup related subreddits to kind of see what people are asking about, what people are talking about.
Sometimes it, people ask really cool things.
Do you want to take the first one?
[01:04:37] Speaker A: Yeah, this was, this was mine. So this was a cool post on the SaaS community. So Reddit SaaS he posted or or they posted. It finally happened. Got my first paying user. I was seriously thinking of shutting down my product yesterday after a week of marketing and receiving mixed feedback, I started to feel like it wasn't going to work out. Which I want to point out, like you only did marketing for a week and wanted to give up. So if you're out there trying to market your indie SaaS thing like keep going like it's going to take more than one week. But there's a great, there's a great moral of the story. So this morning I woke up to notification someone purchased a the premium version. Man, what an overwhelming and incredible feeling to start the day. I'm feeling more voted than ever to keep going and genuinely grateful for this little win. Huge thanks everyone who shared valuable feedback. So I love that this story. Like the guy was like working on this thing and I see this with all the time with these SaaS projects and couldn't get any sales and spent an entire week doing marketing. Is really disappointed. And then he finally got got a sale. I just thought it was, it was excellent. It's just like a really simple Reddit post. Quite positive which isn't always the case with content on, on Reddit and, and he had like a lot of nice, a nice feedback. So I, I, I thought it was cool.
[01:05:59] Speaker C: Yeah, that's chasing the entrepreneurial dragon. It's the best feeling in the world when it starts to work. But. Yeah, well, I'll second the grit point. I think if you're going to be a founder, you know, it's hard. I don't know, you know, I don't want to be the old guy and be like, oh, you know, back in my day we used to, you know, have to eat glass for 10 weeks just to get one reply back, let alone a sale. But you know, I think there's.
Yeah, that, that's great. It's always fun to see people win and when you start to see that success and then the next part is when it starts to snowball. And so when one turns into 10, turns into 100 and things really start to roll, that's the. That's the next feeling that I think is worth the Reddit post. But. Yeah, but you know, it takes resolve, so you got to keep at it.
[01:06:40] Speaker B: That's good.
[01:06:41] Speaker A: A whole week of marketing and then the tool was even interesting.
Do we have the URL in there?
So it was a way to organize screen captures, which I felt was kind.
[01:06:53] Speaker C: Of a cool tool. Yeah.
[01:06:56] Speaker A: Yes. That was another reason I wanted.
[01:06:58] Speaker C: Yeah, yeah.
[01:06:59] Speaker A: I wanted to feature that. He's got a way. Yeah. Okay. Gary. Yeah, it's a cool tool how to organize screenshots, you know. Yeah. I take a lot of these. People take a lot of these. The idea that there's a way to kind of organize these I thought was like super.
[01:07:11] Speaker B: I thought it's.
I have 200 files on my desktop. That's all screenshots same, right?
[01:07:18] Speaker C: Yeah. This is amazing. I actually. This is great. Yeah.
[01:07:21] Speaker A: So you solving good problem like and like because Paul and I love the joke in the segment how like we're just tired of seeing like product hunt clones and directories of other SaaS. Like the. And the SaaS Reddit world people kind of hit on the same products all the time. So when I see something different, I like to highlight it and give them some.
[01:07:43] Speaker C: This is great. I would totally pay for this. I to your point? I have so many just random images and getting them all off my desktop into one place and then having hopefully the. In an ideal world the AI auto tags everything for me, I don't have to self. That would be really awesome.
[01:07:59] Speaker B: Yeah.
[01:08:00] Speaker C: I assume he just got started so he'll get there soon enough. But.
[01:08:03] Speaker A: Okay. Feature request.
[01:08:04] Speaker C: Feature request. Yeah, please program understand the content that I've uploaded auto, you know, put it into its respective folders and then I'm happy to pay to keep it safe and not have it clog up my desktop. So that would be great. Plus, it's like a living history of your, you know, of your time on your computer. It's almost like a little scrapbook of your life. It's kind of fun to track it from that perspective. So, you know, could have like a Google Photos like visual story with some music behind it. About all your screenshots over the years.
[01:08:36] Speaker A: Yeah, and they did a good job too. I like the way the site was put together. They made it really clear there's like a free trial and there's paid version. They got FAQ at the bottom. Like they did everything right.
[01:08:45] Speaker B: I think it's a great looking website.
[01:08:47] Speaker C: Yeah. Well done.
[01:08:49] Speaker A: Yeah. Getting it set up. So hopefully he'll get some more.
[01:08:52] Speaker C: Hopefully get a couple more.
[01:08:54] Speaker A: Yeah. All right, that's a nice one. And then this one is another I pulled. Right.
All right, because this was a little, this is a little different. This was good. So this, this was in SAS too, right? This guy wrote post this thing saying sending 15 emails every day changed my life completely. Every morning before I head to the office, I send 15 cold DMs. It's the single most important habit I've built as a student. Cold emailing let me build cancer simulations with PhDs while in high school, land a hundred thousand dollar GTM roles at startups, schedule four full time big tech interviews in under seven days. And it goes on like to all the success he's had. And there's some nice comments, but there's a lot of people that were kind of mean and doing very Reddit trolling on here saying like how foolish this was or just kind of hating on this guy. And I called out this one person who responded because there was just a lot of like negative feedback and replies on this, which I think was ridiculous. And so someone said huge respect for this. I feel sorry for anyone who sees this post and feels compelled to write something negative. Success is about simple daily habits that compound. Then he replied with a long post about like, you know, how people are being super critical. And so I, I see this all the time on Reddit that like generally when people are negative it's because they're jealous or envious. And people really don't for some reason.
Like a lot of the success stories that are on here, particular if they sound trite or they're like similar types of things that get recycled on the Internet. I mean there's a lot of things that people say about like habits and cold plunges that Everyone likes to make fun of. Right.
But I think, like, this is a great one and like a habit of like, sending 15 messages or DMS or whatever you do every morning is a great one and something I've had success with as well.
[01:10:51] Speaker B: Yeah.
[01:10:52] Speaker C: The only reason I'm here is because sent hundreds of thousands of emails, you know, out the gate in the early days of Retargeter. And that's, you know, we had to start cold and then eventually started to build a network and a series of relationships, and that then led to being able to send warm emails and now send hundreds of those, you know, every week. So, you know, reaching out into the world and connecting with folks and asking for what you want, it's the only way you actually get anything else. So you're going to get kind of what you want without reaching out. So, you know, this is the, the essence of the entrepreneurial ethos. And so I think that's awesome. And have nothing but respect for that, and that's what it takes to build. So.
Yeah. And you know, the Reddit kind of negativity is anytime you jump on the Internet, you know, people are gonna, somebody's gonna have an opinion or get triggered or be upset. And so you have to focus on what you want and execute. So, yeah.
[01:11:43] Speaker A: What tips you have for people who perhaps want to or will cold DM.
[01:11:48] Speaker C: Someone like you, it has to be highly personalized and feel like there's been a tremendous amount of thoughtfulness and research that's gone into that just to kind of cut through the noise because there's a lot of volume of stuff that comes through or, you know, the other thought is not make it cold, make it warm. So find a warm introduction and use that as the way to, to get in. But, you know, cold, cold emails that reference, you know, specific things that show that there's a real desire to connect for a reason that makes sense and is thoughtful and is highly personalized. You know, those can get through the, the filters, certainly. But, you know, you do that coupled with a warm introduction and now you're really talking. So that, I think, is key.
[01:12:30] Speaker A: Paul, do you respond to cold emails that you get?
[01:12:34] Speaker B: Very rarely, and I don't even get that many.
[01:12:37] Speaker C: Yeah, probably like one in a thousand is, I think, probably where I'm at. Yeah, yeah, yeah.
[01:12:44] Speaker A: I. If it's a. If it's like, if it's relevant a lot of time I'll write back like some kid was pitching. I don't know, I can't even remember he's pitching. But he had some podcasts, and so I did it, and it was super funny. Like, I got on there and he was totally shocked that I did it. His little. His little podcast, I was like, I was happy to do it. It wasn't, like, a lot of effort for me.
[01:13:02] Speaker C: Yeah.
[01:13:02] Speaker A: And then sometimes I will respond to people who are pitching me because I want to hear their pitch.
Like, I'm just always curious, like, what people are doing out in the world and getting, like, just some feedback on what people are doing when it comes to.
[01:13:17] Speaker B: Of sales.
[01:13:17] Speaker A: Like, are they doing things that are different than what I would, or am I going to learn anything from this?
And so sometimes I just like to hear people's sales pitch. I'm usually pretty, like, straightforward. Like, hey, I'm not really in the market for this. I don't want. I don't want to, like, yeah, you know, create the wrong expectation.
But I like to listen to what they have and understand how they do the proposal. And sometimes I learn things from it.
[01:13:42] Speaker B: What about cold dms on Twitter?
[01:13:46] Speaker A: Yeah, cold.
[01:13:47] Speaker C: No Twitter.
[01:13:49] Speaker A: I've responded to some cold DMS and had some.
[01:13:51] Speaker B: Wow.
[01:13:53] Speaker A: I don't know if it's cold. So I. I close a customer recently this way.
[01:13:58] Speaker B: Okay.
[01:13:59] Speaker A: I don't. I don't remember. I don't know if that's. If that's technically cold. I can't remember, like, how he found me or how he connected on. On X.
But I got a DM and it was not someone I typically DM. And he was like, asking something about Viber SaaS and what I do. And I wrote him back and said, hey, you're a candidate for this product that I have. And boom, I sent him the link. He's like, this looks good. And I was like, let's jump on a call. We jumped on a call and he bought it. So it's almost like, yeah, so I guess I've sold stuff over Twitter. DM's out.
[01:14:30] Speaker B: Gotcha.
Have you. Have you responded to people soliciting, have I bought?
[01:14:38] Speaker A: Have I bought.
[01:14:38] Speaker B: Have you bought anything? Correct.
[01:14:40] Speaker A: Yeah. Yeah. So these guys had. So I did go to a sales call once. These guys had a. They had an AI tool, like an. I was looking for some kind of, like an AI Twitter automation tool. Just trying to check them out.
[01:14:52] Speaker B: Yeah.
[01:14:52] Speaker A: And so I went to their sales call for it, and they made a tool that, like, did. All of the people hate this stuff, but the AI comments they see all over the place, they had a tool to do that.
It was kind of expensive because they had to pay for the token. So that was one thing. Like I was like, I'm not a candidate. Like things seemed really expensive. But what I want to let people know is that they had use cases that made sense. So while I'm not a big fan of these auto commenting things, they generally provide really low value comments. I feel like they maybe even have stopped people stop using it. They were able to show like use cases that made sense. So if it was like customer service.
[01:15:30] Speaker B: Or.
[01:15:32] Speaker A: Like the CEO of certain types of companies, they actually were appropriate. So like there's some issue somewhere.
I'm just using an example because they have tons of issues. Pg. Right. And so they could go out and actually comment on everything that says like PG&E sucks and everything's down and what's going on. And there could, a bot would say like it's a known issue, there's an outage, it's gonna be like, that would be useful. And they had, they had a couple of examples like that.
So I wasn't a fan of it for like personal use, but I actually think that there's, there's ways that those things can be, can be valuable.
[01:16:06] Speaker B: Wow. How about you?
[01:16:08] Speaker C: No, I think I opened my Twitter DMs for a while. I think I'm going to shut them back off.
It got a little chaotic in there, so I think I have to turn it back off. I got a lot of requests to Joe join Dallas. So that, that was fun. But you know, oh yeah, all kinds of other crazy stuff. So I think mine is unfortunate I have to turn it off after this.
[01:16:28] Speaker A: I guess. I never, I never look at the inbound DMS on LinkedIn. They have a different folder. Right. It's like it just goes to this black hole somehow on, on X I have like all these DMS where I talk with people and sometimes there's like another and I don't get that many. I, I don't, I don't feel like I get that many. And so I'm able, actually I do look at them but on LinkedIn now I never ever look at them ever.
[01:16:51] Speaker B: So I don't know. It's wild.
Reddit deals are actually pretty good.
[01:16:57] Speaker A: Oh yeah.
[01:17:01] Speaker B: Effort in the dms.
So I, I, I personally answer a lot of Reddit dms. Just people asking for help yesterday. They pitch you their products, but mostly it's just people generally asking for help. So I, I do respond to those a lot more than anywhere else.
[01:17:23] Speaker A: I've had at least like 10 or 15 customer calls from Reddit DMS yeah.
[01:17:28] Speaker B: Crazy.
[01:17:29] Speaker A: Yeah, they're pretty. They're pretty. I, I don't get a lot of erroneous or spam DMs. You have to accept people or something, staff the queue you and say, and then have to be approved. So generally, if the first one is not something I'm interested in, I, I don't, I don't respond or I just like, ignore them.
But I hardly get any that are like junk.
[01:17:51] Speaker B: Agreed.
Shall we move on to meme of the week?
[01:17:56] Speaker A: Yeah, yeah, yeah.
Yes. Let's.
[01:17:58] Speaker C: We'll.
[01:17:59] Speaker A: We'll wrap up and just do do me of the week. Are you going to share this?
[01:18:03] Speaker C: I will.
[01:18:05] Speaker A: Okay. This one takes a lot. Do, do do. This one. This one takes some explanation. It's funny though.
It's like trying to explain memes to my wife. Sometimes she's just like, yeah.
[01:18:18] Speaker B: Memes are. Memes are a thing that you have to. You kind of have to.
[01:18:24] Speaker A: Well, this one needs, this one needs to be explained.
So like, so like with the. I think it's hilarious. So with the scale Facebook, let's call it acquisition, someone made this with AI. So this one is like an actual rendering or an augmentation of the Johnny Ives and Sam Allman photo that they did for their collaboration. And someone did a version with, with Zuckerberg and what's his Alexander from Scale AI.
[01:18:54] Speaker B: Yeah, of course.
[01:18:55] Speaker A: Like, I had to give them the broccoli tick tock haircut and the earrings. So.
So I think it's great. I think for me, this was meme of the week for sure. But what do you think, Arjun?
[01:19:06] Speaker C: That's a good one.
I like it.
Yeah. The next generation got to keep generationally relevant here.
[01:19:13] Speaker A: So the best part was like, my wife actually got the broccoli haircut piece. She's like, oh my God, no. So it's broccoli haircuts that your cousins are into or whatever.
Oh, that was fun. Okay.
Yeah, yeah. I was gonna say, Arjun, did you enjoy be on the show today?
[01:19:33] Speaker C: This is great. It was a fun conversation. Good to, you know, wind our way through lots of interesting topics. And it was great that we covered everything from, you know, AI technical, you know, depth, all the way to marketing and insights and the whole nine yards. This is a blast. Thank you both for having me. Appreciate it. Of course.
Yeah. Super nerdy. I love it.
[01:19:55] Speaker A: Paul and I are both like, that's going to be our differentiator. Like, there's so many podcasts out there. How's yours?
[01:20:00] Speaker C: Different?
[01:20:00] Speaker A: Don't worry, ours is only for nerds. Like, it's really nerdy, right? We'll talk about the news, but we'll only do that nerdy parts.
[01:20:08] Speaker C: That's great. Well, thank you both so much. This is fun.
Okay, thank you very much. See ya.
[01:20:14] Speaker A: Take care.
[01:20:14] Speaker C: Thanks, everyone. Bye.
