Subscribe
Share
Share
Embed
"What if we Could?" A podcast exploring this question that drive us. We explore the practical application of artificial intelligence, product design, blockchain & AR/VR, and tech alpha in the service of humans.
Calvin (00:01.868)
All right. Hello, hello. Welcome to the, what if we could show I'm your host Kelvin today. Um, we are discussing as always the exciting and burning questions around artificial intelligence and tech alpha. And today I've got with me all of the four technical co-founders at mesh mesh. So we've got Bob, David and Kevin with me. And so really excited to kick today's one off. I think we've got a couple of exciting things on the list. We have. How the executive hesitates amidst the hype.
around AI at Davos, how AI's potential to create billionaires. We're also talking about GPT newspaper, an autonomous agent for personalizing news content. And we're looking at a couple of other things such as how lawyers are being outcompeted by a new model or a new LLM study called Better Call GPT. So let's dive in. Kevin, you were kind of fascinated by that AI at Davos article. Do you want to tell us a little bit about it?
Kevin Nuest (01:01.054)
Yeah. So the headline coming out of this, uh, from, from Boston consulting group, right? So no, no small organization. They're talking with a lot of large companies and 90% of execs. Cautiously approaching AI. So the takeaway from this, from my perspective is, uh, it echoes the end of last year and, uh, the reports coming out of last year of
teams getting together and starting to figure out AI, what are they gonna do? It's not gonna be a chat GPT because they don't wanna give them their data, but all the other reports and surveys said that they hadn't really got a whole lot meaningful done. I'm talking in the large collective of, call it enterprise organizations, mid, upper mid-size organizations. And here we are at the beginning of this year, kicking off the year with these same execs saying, hey,
I'm going to sit on the sidelines until the sorts itself out. And I thought twofold. One is this is a, um, a massive opportunity for the companies that get off the sidelines to be one of the 10%. You can go make some big gains in the next 12 months and 24 months. They're going to compound, uh, and, and do some big gains against your competition. And then the second is that.
It really shows, I think from, from my perspective, these execs are thinking about this as a zero to one moment of I don't have AI in my business and I'm going to implement the right AI and now I have AI in my business and we solve the AI problem and in 2020 at the end of 2024, 2025, and we move on to whatever the next problem is, they're not realizing this is, this is the state of business moving forward is perpetual iteration powered by AI.
It's a lot like more like software development in that, that continuous iteration, then it is solving very discreet stage problems in a business. So those are my, my opening thoughts here, why I thought it was interesting and why a lot of teams, I think are going to, going to, going to be caught blindsided, flat-footed, uh, by mid 2024 and be scrambling to figure out what to do this year.
David DeVore (03:22.376)
I mean, the other thing I think that was interesting about that is like, it's actually a certain class of CEO who goes to Davos, right? And, you know, there are, you know, so think of like the Fortune 500, the Fortune 50, it takes a lot to, you know, kill those businesses, right? And so in a lot of ways, they don't need to change.
Right. They don't have the, they don't quite have at this point, the pressure of competition. Um, they're not getting, you know, eaten alive by, uh, getting into live by whatever the next thing is. And so, you know, the CEOs are feeling, you know, comfortable and able to fly to Davos. Right. So I think, I think this, that's this. Yeah. And so you, you wonder, okay, it says 80% are, you know, backburning it.
Kevin Nuest (04:09.302)
That's, that's true.
David DeVore (04:17.08)
I think as you go downstream into mid-marketing and into long tail, like, those are the businesses that are fighting to do more with less. They don't have as many dollars to make big mistakes. They don't have the big audiences. They don't have the big sort of embedded products and so forth. And so I think that like all good innovation, it's going to come from the edges.
and you know, and probably not going to start at the top.
Calvin (04:53.064)
Yeah, that's an interesting question. Also, if you look at, I mean, I think what you say is right, but it's also a very different group of people that are potentially affected. I think what we've seen a lot with AI is, it's neither the people who are the sort of quote unquote highest value at a company, but you have all of those middle layer that are easily to automate. A lot of the more repetitive tasks, all of the mid-level.
David DeVore (05:17.541)
Yeah.
Calvin (05:22.496)
worker group, right? That's a lot easier to automate, especially in sort of the quote unquote white collar sector. Right? So like what you're saying is totally spot on. It's just not affecting the CEOs right now, but it will affect everybody who's like sitting somewhere in between that or further in the company. So how do you think this is going to hit them? Right? Because you're right, like currently they can sit on the assets and do asset preservation for a longer period of time. But it is going to sink on an individual role.
David DeVore (05:49.532)
I mean, I think we're seeing it. Well, it's gonna, but it's gonna creep in, right? I mean, so we see Microsoft embedding, as Google, Adobe, so existing tools and existing UIs and existing things that run a business today, it'll, AI is gonna creep into the embedding of existing stuff as opposed to be a.
net new thing inside of large organizations. You know, so I think you're right. But it's, you know, it's not like the CEO strategically is thinking, oh, well, hey, we're gonna do a press release, because now we're using AI, right? Like, it's going to be it's going to be, you know, a slow, internal change that happens with existing tooling like Microsoft
as opposed to a large business strategy.
Kevin Nuest (06:53.834)
So that throw a comment here and then Bob want to hear your thought on this, that sounds right, Dave, that these enterprise organizations would keep looking to their, um, long time trusted partners that are already integrated into their business, the Microsoft's, the Amazon's of the world for, um, guidance on how to navigate this and where to get started and where to start implementing AI in their business and entrusting them that they will continue to
Uh, those, you know, let's keep using Microsoft will continue to help the business keep pace with the AI change and that, and they're leaning really hard on those partners. So Bob, do you think that model will hold or is this different? Will, will relying on your large, you know, sweet partners like Microsoft and, and Amazon, uh, get you there. Or.
David DeVore (07:29.036)
Mm-hmm.
Kevin Nuest (07:50.318)
Is it different in the competition? Isn't the competition today for these enterprises? It's the competition for tomorrow. The, the, the mid markets that can all of a sudden become a large cap and start punching above their weight limit because they are compounding that AI in three and six month increments instead of 12 and 24. What do you think?
Bob Ullery (08:09.564)
is probably in my mind, all of the above, right? I think enterprises that stick with their big scale partners, i.e. Microsoft, AWS, whomever, are gonna get most of the way there, right? I mean, assuming they're gonna integrate it into the known existing processes. So they're gonna gain efficiency for sure if they stick with the big vendors. I don't see that not happening.
Um, I think there's a different way to look at this though, instead of like just improving processes and Dave touched on it, like making existing value more valuable, making existing products better. Um, that's a, that's a layering system, right? So layering on new capabilities of the base and exposing new things, but that would not include
net new propositions of value. There are going to be an infinite quantity of new takes on how to do old things. You're going to gain a bunch of value by optimizing the old way with the new tech. You're going to gain a multiple more devising a new process with the new tech that unlocks really the right way that everybody sort of wants that process to work, but never were able to on the CEO side to just a comment there, like.
I think if you're a battle-hardened CEO, been in the mix for 30-ish years, you've seen a lot of cycles come and go. We always look at the hype cycle curve as the framework of an easy model to think about. Initial hype curve goes up, then you'd go down to the trough of disillusionment, and then you finally find the slope of enlightenment to the plateau of productivity.
And I think, you know, if you've done that for decades, you are, you know, you've seen this dance before. We'll, we'll wait and see once we hit the trough where the real value is, and we will jump on the train as it starts to track up the slope, right? But history doesn't always repeat, but it rhymes, right?
Bob Ullery (10:24.456)
And so you think back to like late 90s, here's the internet. I'll wait and see. Here's websites. I'll wait and see. Here's web apps. I'll wait and see. Here's a mobile app. I'll wait and see. Here's social. I'll wait and see because of that trough. And I'm convinced a million percent. I don't know if anybody could talk me down off this ledge. I don't see a slope. I don't see a hype cycle here on AI. I see a laddering effect. We're going to have a, we are in the slope of hype.
And I think when we touch that, it's going to plateau and then continue to ladder up on a really, really compressed cycle because we have not yet understood or even identified like where the disillusionment could even happen. We know that this carries value efficiency, new gains, new takes on old methods for value. So I think they're going to be surprised on this one. The ones that are sitting on the sidelines, it's a long way to say that.
Kevin Nuest (11:22.066)
Yeah. And it'll be in to that's the slope of disillusionment is more nuanced in AI than, than in any other cycle because AI is just going to cover and span. All industries, verticals functions. And so now you're, you really would have to say, okay, what got overhyped as a function, as a niche, and that could happen or that can plateau, but yeah, betting, betting against AI.
I'm going to go air quotes here, AI compounding. That's just now we're just saying the new internet technology and you're going to take a bet against that. So I think that they're on the wrong side of the mentality on that and I have to take a couple hits to the face to learn that probably. But I'm not, Dave, to your point, I'm not concerned about the enterprises dying or having a little more runway to get there. They will figure it out. They'll figure something out.
They'll figure out at least steady state. I probably what happens is the, the CEOs have a rip core to pull, which is more layoffs and pulling more people out of the organization to hit their number while they're trying to do that, which is unfortunate, uh, but they're also probably trying to automate some of their work first looking at growth opportunities too, so fortunately it's probably net same strategy, show two slides to beside one another in the exact deck. And that's probably what it looks like.
But this.
Bob Ullery (12:47.96)
they're gonna have to make the SWOT graphic bigger, right? Especially on the threat quadrant. I think is probably when they'll start to realize that box is infinitely getting bigger and bigger and bigger where it needs a couple slides every time, because it sort of encapsulates the unknowns. The biggest threat to our current business process is that it's obsolete. And here comes the one person startup that has a better mousetrap. And...
Calvin (12:54.136)
Thanks for watching!
David DeVore (12:57.192)
Thank you.
Bob Ullery (13:17.373)
It's perfect and it's totally automated.
Kevin Nuest (13:20.59)
Great transition to this, the story that Dave you threw in here, man, talk about people, you know, using, using leverage to do big things. What's, what's this one about?
David DeVore (13:31.212)
Yeah, so this was this was Sam Altman. And basically, he you know, predicted that, you know, we're going to have the first, you know, individual, like single person billion dollar business soon, right? For billion dollar startup, right? Which, you know, in startup world, like unicorn is a thing, right? And there's never been a unicorn that is
created with one person, right? And so it's really interesting because, you know, okay, well, how does one person create a business that is a billion dollar business? And, you know, the only answer is they figure out how to replicate all the things that people would potentially do, right? And
And we've been doing that, right? So it's, you know, and you sort of start to figure out, you know, quickly. Gosh, here's all these buttons, I have to push to like get to this end result. Right? Here's all here's all the you know, and a lot of it is button pushing, like, you know, I would say like the you know, it's, you know, and we it's not like, we haven't been working on automation for a while, right? I mean, we've been all sort of been inside of marketing automation for 20 years or whatnot.
So the concept isn't really new, but I think the thing that is really interesting is also the, how quickly we can automate something and it'd be good, right? Good, you know, and, um, it's not just, uh, you know, it's not just a variable where we put first name in an email, it is a 100% personalized email for the person on the other end. Um, it is a.
100% personalized video for the person on the other end. And that allows for one person to really do the 10x right? So if you know, in it, you know, or more, right? All of a sudden, and I think it's really interesting. I mean, I've heard you say it before, Calvin. I mean, you are, you know, you have you've integrated AI into how much of your day? Do you think?
Calvin (15:52.659)
Certainly 60 to 70 percent at this point.
David DeVore (15:55.916)
60 to 70 percent. And, you know, and what do you think in terms of productivity? What do you think? Where do you think you're landing? Increased productivity.
Calvin (16:08.385)
I think 100% 10x. I think the big one, and I sort of outlined this a little bit in a post I talked about yesterday is I think in the terms of time, it has saved me to go from like zero to 60% and then go the rest of the way myself. And I can do that in 30 seconds now, whereas it has taken me four hours before. I think it's a lot of these like what might seem as minute improvements, but they make all the difference.
And I think that's also interesting when you talk about sort of going to that billion mark or right scaling up a business, right? A lot of it is then tried and true and repeatability. And now you start repeating. So, you know what the process looks like and the difference with what we can do with AI now in terms of automations. And you talked about it before. Yes, we did have automations for a long, long time. Right. The difference is now that now we don't have to account for every edge case, but we can get all those small nuances that we current that we needed people for or
David DeVore (16:35.215)
Right.
David DeVore (16:46.214)
Right.
Calvin (17:05.276)
thousand if and else statements. Now we don't. Now I just put something, and I've seen this when I build automations, I just put something in between, which is my LLM transformer. And I'm in there and I say, go fix this for me. Make this cleaner. And now instead of a thousand if else statements, I have one node in there that goes and fixes this for me. And so now my automation scale, they repeat, they work, and they can account for a lot of edge cases.
So now we are like in that automation land that we always wanted to have, but was sort of infeasible because it cost just as much to get there as it would have been to just employ people to do it. And that has fundamentally flipped.
David DeVore (17:39.836)
Mm-hmm.
Bob Ullery (17:48.812)
Not, I guess.
Kevin Nuest (17:48.818)
And I want to, I want to give Calvin his flowers here. Right. Calvin is a 10 X engineer already. They, they met him. So when he says I've 10 X that's a hundred X. And when he, when Calvin, when you say, Hey, it took me, you know, it used to take me four hours to go zero to 60%. Let's do a quick multiple of 10 on that. Okay. So an average or a mid engineer or early entry engineer, that's 40 hour week. Any company adding in meetings, that's two weeks.
That's two weeks of work we're talking about to get to like, hey, here's the 60% thing. What do we do with it now? And let's feel it out. And Calvin's saying, I can get there in 30 to 60 seconds. That's compounding power. That's crazy.
David DeVore (18:28.592)
Mm-hmm.
Calvin (18:30.711)
And if the, go ahead.
Calvin (18:35.648)
The fascinating thing is, I think, and we've touched on it before, why I said sort of that middle layer or the beginner layer gets eroded a lot. Like what I worry about in that scenario, and I've seen it myself now working at this is it was already really, really hard as an entry level software engineer to get a foothold, right? It has been for the last couple of years, right? The bar has continued to go up and up and up and up and up. So if you've just gone out of college.
And you're like, you know, nice, I've got a CS degree. I'm going to jump into the industry. Right. And they kind of, you know, and there's a lot of memes on TikTok where they say, you know, they want you to already have hacked it and be the awesome geoguesser and get all of that. Right. And you ideally written the Python script, you know, that can scrape a billion websites and get recommendations and you build your first app at 18. Right. So if that was the bar, and now we're at a point where
David DeVore (19:05.361)
Mm-hmm.
David DeVore (19:19.584)
Thank you.
Calvin (19:29.608)
When I work, I see myself struggling to work with beginner engineers because it's just like GPT is at that level. Right. So that that's kind of, I think getting to a better engineering level is really, really hard now because like the bar just skips you. Um, and that's what I'm a little bit worried about in terms of that development. If you were like somebody coming newly into the scene, right.
David DeVore (19:37.176)
Mm.
David DeVore (19:45.84)
Mm-hmm.
Kevin Nuest (19:51.116)
Yeah, but you can.
David DeVore (19:54.444)
Yeah.
Bob Ullery (19:55.316)
I don't think it'll be a good.
Kevin Nuest (19:55.458)
You can pair program with GPT four for, for think about that, right? The level that GPT four is at and, and for all the grief people give it, it's not smart enough Calvin's like, yeah, this is a great peer programmer to, to go at like, man, what is 4.5 or five going to unlock for like, oh wow, we got like a senior pure programmers sitting here beside me. That'd be, that'd be great.
Calvin (20:17.952)
Nah, that'd be awesome.
Bob Ullery (20:18.732)
I'll sprinkle on, like unless you're building, you know, mesh like companies that are going to iterate through problems autonomously, like my perspective here is AI is not going to 10x you except where your expertise lies. Right? So like Calvin goes from hours to seconds because he...
already has a formulated question in his mind that he's trying to answer faster, right? Without that expertise, what questions do you ask? He's still, it's like a student in school, right? You're gonna have to go through the process of learning and build up that expertise. And therefore, I think the first billionaire, solopreneur is not gonna be lucky. That's not gonna just randomly happen, right? I think that person is gonna be an expert in a particular field.
David DeVore (20:48.206)
Right.
David DeVore (21:12.56)
Mm-hmm.
Bob Ullery (21:13.028)
and have a really sharp perspective on where it's broken. Right? And if I could have 500 sets of hands, these are all the keys I would punch simultaneously right now. Let's automate that. And I would go as far as say, I don't, I mean, the moment we see the first one hit, we're gonna see the first 100 billionaires, like that week, right? This notion of like the.
David DeVore (21:25.174)
Mm-hmm.
Calvin (21:28.481)
Yeah.
David DeVore (21:28.616)
truth.
David DeVore (21:37.468)
Hmm.
Bob Ullery (21:38.7)
The playbook being released, like how I got to a billion dollars by myself. If you are not reading that article, the moment it drops, you are not going to be one of those people. Right. It will be a race, I think. And then it'll be a flywheel expansion out to every nook and cranny you could think of around expertise. That's exciting. I think it's exciting.
David DeVore (21:43.772)
Hmm.
Calvin (22:01.376)
Yeah, it is. It is definitely exciting. I want to, I'm going to jump one here just because I think it fits very well in the sense as we're as we're tapping into like, you know, getting human potential and getting that to a greater scale. I think it was actually really interesting to read that World Economic Forum article where, you know, it was an interesting thought piece that was released about and a very positive one, actually. And I think that was a great stride in terms of getting more push into.
sort of the practice in AI, they talked a lot about, you know, ethical innovation and trustworthy practices and all of that, but they were kind of coming at it from the standpoint, you know, how can AI usher into an economy that puts people first, right? So going back to the billion, first billionaire, right? How does AI enablement look like in terms of putting people first? What does it look like to you guys? Like Dave, what's your perspective in terms of that?
David DeVore (22:56.824)
And I really love and even this is I was I was paying attention to Balaji Shivasan and he this was this was pre really me even getting deep and probably pre props. And you know, he has this vision of that where this is all going is and where and really sort of crypto
transactions fit is in this concept of around this concept of bounties and tasks, right and the future of work of worldwide work, probably is not a job. That it is more akin to having a specialty and being given a bounty to complete that task. And then coming back with
proof of that task being completed, and then getting paid in very, very specific units of work. And so it's interesting as we've sort of been through this whole process of automating things around mesh mesh, we find these nodes, these things that need to be done by a person and can only be done by a person.
And at that point, when there's something that needs to be done by a person, the LLM sort of, or, you know, the, the AI runs out of room, right? It need, there has to be a person that, that does the next step. Sometimes those next steps can be as simple as uploading a video, but sometimes those next steps, depending on the use case, it can be complex, like they need to go someplace and take, so I need a picture from a location that, you know, or I need, um,
Or I need a conversation or I need a, you know, I need a piece of paperwork, especially, you know, in areas that don't have, that don't have, you know, access to the internet the way that we do. If there's a physical signature that's needed or whatnot. So I really love this idea. And I mean, I think that I really love, I really love, I really love the potential of artificial intelligence to enhance.
David DeVore (25:17.184)
human life in all the aspects. Like I get excited when I hear, I get excited when I hear about climate change opportunities. I get excited when with Easton, I get excited with, you know, anytime I hear about ways to make, make a larger middle class. And I hope that that, you know, on one level, there's sort of like, hey, here's layoffs from coming.
On another level, it's like, yeah, but there's also this, this whole new economy that is forming that, you know, it used to be, you know, before the world of jobs that we had over the last century, and everybody needs to have a job. Like people just made a living. This had a trade and they did that trade over and over and they're a master of that trade. And I really think that like part of where this is going is
a little bit more in that direction where it's a middle class trade class, more of a middle class trade economy as opposed to a top down jobs economy. So I think that's really what excited me about that article.
Kevin Nuest (26:34.594)
That's, that's so great. I, the trades approach is a really, I think, really good analogy. And it's, it's out of our, it's out of our collective historic time window for the most part, you know, longer than 50 years ago that like, that was really the thing for, for a lot, a lot of people. I learned entrepreneurship from my grandfather and not that he, not because he was an entrepreneur, but because he had a trade, he did construction. He did masonry. He, he grew up like.
doing an apprenticeship in masonry and construction work and then just did his trade, which meant running a business, a very small business. And I went and worked for him during the summers and I learned a lot about work ethic and the trades itself. But he never said I'm an entrepreneur. He's just like, yeah, I do a living. I make money for my family. I take care of them and that's what he did.
David DeVore (27:21.862)
Later.
Kevin Nuest (27:28.462)
And I agree that there's going to be a lot more opportunities like that. It's going to be like, where, where is your competitive edge? Your, your aperture could look a little bit wider than like I do masonry really well, but it's still like you were talking about earlier, Bob, where's my competitive edge and what do I really understand as an expert? And it may be pulling multiple topics together to then make a synthesized edge that is an advantage. But even with the power of AI, you can't know everything. You can't be better.
than average at everything.
Bob Ullery (28:01.124)
It was interesting how you were talking about the gig economy. There's sort of unlocked a weird analogy, right? Whereas like in terms of our lives, right? Like we sort of all were playing like NPC characters in a story before, right? Like if you were the merchant, you had a shop, people came into your shop and say, how you doing, what do you need? And you'd sell it to them, right? Over and over and over and over again. Or construction or whatever it is. And the shift that.
sort of clicked when you were talking there is like, now we're kind of shifting all of us to the main character in our own RPG game, right? Where you're just going on quest after quest after quest because you're trying to earn the thing at the end of the quest and your loadout kit is your trade, right? You're the carpenter, you're the graphic designer, you're the whatever it is, and you're just going from quest to quest to quest, still in the same lane of trade and expertise, but
David DeVore (28:58.152)
Mm-hmm. Yeah.
Bob Ullery (29:00.236)
You don't know who necessarily the next employer is going to be. It's fine to have a quest being the merchant too, or the role at the big company too. But I still see those as quests in the same way. And I think the employer might too.
David DeVore (29:17.056)
And the future then is not LinkedIn. The future is some system that doesn't exist right now, that where your accomplishments are probably logged on a blockchain, right? And, and you, you have, you have badges and completion and, and all of that. Uh, in the same way that you would, you know, um, all of that ends up being, uh, your identity for who you are and what you've done.
that is then accessible from anywhere for somebody to, you know, accept your, accept your, accept your task and, and pay you for it all, all in the same place. And so, um, yeah, it's super fun.
Calvin (30:02.024)
Yeah, we had that, remember when we had that idea on talking about sort of the, the Oracle, right? Like we, we've kind of followed that train quite a bit. And I think it's, it's interesting how we're on our end on Mezmer's side, slowly getting to the point where AI is enforcing so much of our decisions that you could, you could almost call it a little bit of an Oracle inside of the organization. And I have to say, I'm not against that vision at all. And I think if you, if you put that out, right. And artificial intelligence in its most utopian scenario.
Kevin Nuest (30:02.75)
Can I get?
David DeVore (30:22.844)
Mm-hmm.
Calvin (30:31.528)
right, is, would be a very benevolent routing system to ensure that it sort of balances out needs, you know, across humanity. So if you look at that, you know, that would be going back to your trade example. You are sitting here, you have your trade, you have you the favorite thing that you do, right, and you do it all day, right? You're just very good at it. Right. And so the AI says, oh, well, you know, there's a job to be done over here and just automate your routes, the task to you to do it. And so that kind of becomes that balancing router.
And sometimes the AI might make the choice and say like, you know, this guy over here, he hasn't made enough money in the last time so he hasn't had gotten a chance. So let's route it over here, right? Let's balance the system up. So if that's what I think would be the most utopian. I think it's actually to an extent an achievable scenario, but I'm saying the most positive utopian scenario that we can think of, of how this might work.
Bob Ullery (31:19.84)
It's not net new either. It's, it's scaled on an old idea, which is meritocracy, right? And a lot of businesses caught fire in like early 2000s, 2010s, where they were all going after meritocracy in an effort really to squeeze the most juice out of the skill sets they had in their resource pool, right? Their employees and let them do what they're good at. This is probably that just with much tighter, we're tight rails, but also
Kevin Nuest (31:20.118)
Yeah.
David DeVore (31:38.728)
Hmm.
Bob Ullery (31:49.748)
an infinite quantity of pathways that the rail can take, right? That's cool. If you've got skills to offer and value, this is good for you, right? I think if you're struggling to find your place in the world, it might be scary.
Kevin Nuest (32:05.122)
It's funny because Calvin and I were literally working on an automation last night and, uh, it was to automate some task work that I have that's recurring. And I just, I said something along the lines of, I just need the system to tell me what to do when I wake up in the morning so I can do the most important thing and it puts it on my calendar and I do that thing and move it forward. Like I literally want it to make that decision for me. I'm giving it the intelligence now. That's what the automation was to give it the intelligence to then tell me.
what to do and when to do it so that I can get it done because I already decided it was the most important thing. So that literally in practice and in a micro scale, I really quick one to, uh, I've pulled a couple of stats here before we move on from the billion dollar one person company, uh, that I thought were relevant and interesting. So y'all remember. When Instagram was purchased by Facebook for a billion dollars back in 2012, right? That was one of the
Calvin (32:43.509)
Yeah.
David DeVore (33:01.404)
Mm-hmm.
Kevin Nuest (33:02.134)
first big acquisitions or everybody's like, wait, a billion dollars for what? 13 people at Instagram at that time, billion dollar company, right? 20, not long after 2014, Facebook bought WhatsApp. Another one would take, what is, a lot of traffic, bought it for a lot of users for sure. $19 billion, right? Just a couple of years later, you thought they're off the rocker with the one billion two years before for some rich kids of Instagram taking pictures?
David DeVore (33:08.584)
Thank you.
Kevin Nuest (33:31.422)
$19 billion or some chat app, 55 employees at that time. And those, those are flirting with some, you know, it's close to one person, billion dollar companies as you can get. Uh, and that was, you know, a decade ago here at this point. So it's, it's not unreasonable or what this prediction, right?
David DeVore (33:47.769)
Yeah.
David DeVore (33:52.668)
Not at all.
Calvin (33:54.904)
Absolutely not. The next thing to get us a little bit, I think we've talked a lot about the high level vision of where this can go. We have another one here, which is sort of the GPT newspaper story. Do you want to kick us off with Dave about it a little bit?
David DeVore (34:12.988)
Yeah, so this is this is a GitHub repo I found and it caught my attention because we're doing something similar. I didn't think that it was I think ours looks better not to trash it or anything, but I thought that it was just a really interesting concept then and as we start as generally, which we've been chasing, which is like, how do you automate? How do you automate news? How do you automate?
content and content with a capital C, right? Meaning like, yeah, it's a newspaper. It's a podcast. It's a it's a banner ad, like, content is a big is a big. It's a book, right? I mean, it content encompasses a lot of stuff. So I thought it was interesting. Just to sort of see this in play and throw it out there for other people who want to use it. And then, and then hopefully, we're going to have
have some stuff for people to use as well and share here shortly in terms of our newsletter automation and content pipeline. And so yeah.
Calvin (35:22.496)
Yeah. And it's great that you sort of brought it back and I found it fascinating, right? The balance between sort of this high level vision of where I can go, but how do I make it actually tangible within the now and here? Right. And we see a lot of people struggle with that, right? It's like, yeah, I can totally get where AI is going to get to, but like, what do I do today? And I think it's great that we're picking, you know, that we're finding a lot of these examples that might seem like smaller solutions, but they're very nice.
automated solutions that you can put in your day-to-day workflow right now. Like who wouldn't want to have, you know, a fully automated newsletter being sent to the inbox every day. Right. Like imagine you just, you know, you're just getting the things that are most relevant to you being pushed to you every morning, right. Fully automated and tailored to you. Right. That's a great experience for you. And this is going back to sort of the Cambrian knowledge explosion here right there. Right. It's like, if you go to JetGPT, you don't know what you're searching for. But when you get your email, right. You know, just being
David DeVore (36:11.856)
Yeah, it's interesting. I mean, when I...
Calvin (36:22.124)
Push to you and you're being provided with knowledge every day. Lovely.
David DeVore (36:26.324)
A hundred percent. I mean, and it's really interesting, like, you know, like is content. And I talked about content with a capital C being really, really broad. Um, but content, even a bigger definition includes pretty much anything that we say at any given time. Right. And, you know, we've worked on pipelines, uh, where we were like, oh, okay. Uh, here's a call from fireflies.
We want to automate that into content for a follow up email. Um, and by the way, here's another, you know, a bit from that conversation that's going to spin off into a piece of content for the newsletter or a piece of content for, uh, that goes inside of the podcast brief and whatnot. And so part of, I think what's interesting is sort of tying it back to the expert. Like sort of, um, that the expert and a given point.
is creating content that has a mission, right? And then it's just, the content actually is at the point that it comes off the fingertips or comes out of somebody's mouth. And then off it goes to any conceivable channel is really exciting stuff, right? We're really getting to a place where content is, as close as possible goes from conception to creation.
you know, in like, you know, three minutes of gen time, right?
Calvin (37:59.201)
Yeah.
Kevin Nuest (37:59.338)
Yeah. And fireflies, the software we use to, uh, for note taking both internal meetings and external meetings. We have multiple, multiple pipelines and workflows coming off of those conversations, depending on what, what needs to happen next. So it's super, super high leverage to take the words that you're saying in a conversation and make actions happen out of it.
Bob Ullery (38:22.988)
at some point in the future, we will run this entire business just based off a daily standup together, right? Ideate and talk about what's going on. Yeah. And it'll just do its thing. We won't be one person billionaire, but like, what about the first four person startup to hit a billion?
David DeVore (38:28.132)
Yeah, once a day make decisions it goes.
David DeVore (38:34.682)
That's exciting.
Kevin Nuest (38:41.008)
I'd gladly split it with you, Bob. I'd gladly split a billion dollars with you.
Calvin (38:47.176)
I think it's fascinating that you mentioned it with the Fireflies example. I can totally see that. And it's actually a lovely, lovely idea, right? Like we've had that a lot. You're in meetings and we know how meetings sometimes go, you know, you have them very structured, you have a very structured intro, but still like, you know, a lot of thoughts come up and, you know, you go left and right, you go in tangents, you have ideas and you just have a natural conversation, right? And I think one of the marvelous things that we've seen with sort of the automations that we played is how it then helps us kind of put that back into focus.
Bob Ullery (38:47.83)
All in.
Calvin (39:17.08)
turn it into the next clear action items, right? Taking into account all of the things that we wanted in that meeting and pushing it back to us as tasks, right? Like basically just, you know, here Oracle system, take that regurgitated, compress it for us and give it back to me to compress.
Bob Ullery (39:34.124)
Yeah, I think part of that secret here is, where's the waste in time when you have meetings? You typically have a meeting to talk really about what you're going to get into in the next meeting. The first meeting is like, hey, intros, let's talk about what's on our mind, and then we get to brass tacks next time. Part of that delay is IKTs. You've got to bring a lot of people along, they need to have the knowledge on the thing that you guys are coming together to solve.
And something really interesting about this new world is like this notion of like not having to do IKTs anymore. Why? Well, instead of like trying to build up individuals knowledge in the old world I need to bring Kevin up to speed on this thing. Kevin is now able to carry forward the collective knowledge of the hive as he goes without even ever having to have an IKT. If I put something into our collective brain as an organization
Kevin has instantaneous access to it. So that I think that alone reduces, uh, meeting waste for sure. You can go from the first meeting directly to the fourth meeting. Right. Or at scale, there might not be a second meeting, you know, like meeting is kickoff. It's getting everybody's thoughts on paper in terms of priority and okay, ours or whatever tied to this thing we're about to do.
Everybody good with those things we just talked about? Yes? All right, Fireflies, execute.
Calvin (41:06.288)
Yeah, that's a fair point. That would be a good next step, that's for sure. I think Firefly is doing great to start with, but there's definitely a lot that can get from there. There's a lot where this can grow into. Might be...
Bob Ullery (41:06.445)
That's awesome.
Kevin Nuest (41:11.042)
That's a billion dollar company meeting efficiency is a billion dollar company.
Bob Ullery (41:24.952)
Challenge, just less thought on this, right?
Calvin (41:28.076)
Good.
Bob Ullery (41:30.684)
The challenge to us is not to build the first solo-preneur billion-dollar business, but to create the first zero-preneur billion-dollar business, or not even billion, just profitable business. I think that the most exciting thing about our thesis is this notion that we could theoretically deploy fully autonomous organizations.
to do things and that would be the nirvana finish line I think is when we achieve that.
Calvin (42:02.528)
That is a very, that would be a lovely finish line to get to 100%. I think that speaking of, you know, cause this might scare a couple of people, right? They're sitting, sitting here in their chairs and they're like, who help is all of that getting automated? Like the one thing that, that I found really interesting, we talked about last week, we talked about sort of the health sector, right? And a lot of the, the facts that, um, LLMs can accurately diagnose better.
than most doctors in 80% of the cases. Now there was another study coming out, which is called, and I actually love the title, which is called Better Call GPT. And it's basically, it was a study that was comparing large language models against lawyers. And lo and behold, it revealed that the large language model beat lawyers in contract reviews on accuracy, speed and costs. Right? So basically across the board, right? And this paper...
David DeVore (42:55.644)
Yeah.
Calvin (42:58.272)
It's like really an in-depth comparison, right? So they looked at, you know, everything, um, when they looked at, you know, completing reviews in seconds, right? Just outpacing the hours that were required by humans. So we're talking seconds versus hours and we all know how a lawyer charges by, by the minute. And so that's just another one of the pot here.
David DeVore (43:17.484)
I actually did this. I did this, I actually did. So my friend is going through an ugly divorce and I loaded all of the agreements up in there. I also uploaded, so all the agreements, I uploaded all the documents in there as well as sentiment analysis on messages back and forth and whatnot to basically like.
produce a GPT lawyer specific to her case for you know, for $20 a month. It's like, you know, it, yeah. And so it's, it's sort of a no brainer, right? Then it works. That's, that's the other thing that's crazy about it. You know, I mean, there's some things that are a little off, but, you know, and you want to double check, but at least for like, fact finding, and like,
Bob Ullery (43:55.853)
Yeah, beat it.
David DeVore (44:15.88)
asking it like, what, you know, what's the what's the, you know, what's the bounds of this liability? Or what's the bounds of that thing or whatnot? I mean, it? Yeah. When when
Calvin (44:26.428)
Yeah. And like finding loopholes in contracts, for example, right? Like finding issues in contracts, reviewing it, or just like a very, I think, I think one of the basic use cases that this legal discussion has started with was for laymen who have no idea, who are being put a contract in front of them, right? Help me understand, right? Asking questions about that contract, just getting reasoning from it, right? Just understanding it. Like that translation is just something that is unbeat by LLMs, right?
David DeVore (44:30.364)
100%
Calvin (44:54.716)
And I think we've had that notion for a long time and it's been discussed for a long time. The fact that now there is substantial scaled studies coming out that prove that, I think it's going to just have a wide impact. What do you guys think? Where is this going to go from this point? What's going to happen in that industry now?
Bob Ullery (45:17.684)
In law specifically.
Calvin (45:19.848)
Yeah, like in the legal sector as well.
Bob Ullery (45:23.204)
I think it gives a lot of optionality to the consumer of law services. So I think one flag I'd plant in the ground is your cost for legal services is going to go way, way down. Partly because of the time savings of the law firm you're contracted with. Two is kind of like a lot of people...
take on contractor work when they build their own homes too. You know, like, okay, I'm not gonna pay the contractor or the extra whatever to manage all these people. Like I'll take care of the landscaper, I'll take care of the whatever, just in an effort to save money. And we never really were able to do that in the legal field, right? That's a really big risk calculus to take on the legal process yourself without that expertise. And, you know, at worst, I think there's gonna be a bunch of time clawed back for...
uh, folks being represented by legal firms to do some of that work themselves, or maybe better identify the right lawyer based on gaps.
Kevin Nuest (46:29.27)
I think we're in a privileged position too, right? That we can get a lawyer if we need to. I feel confident that any one of us could go find a lawyer if we needed to get retained services, especially if it's mission critical and make it happen. There are so many underserved people that just, that arguably need it more in a lot of scenarios and can't get any access and they're locked out of it. So to bring that to the air quotes long tape,
long tail of the need for law services. And in reality, it's the majority that aren't able to leverage what they need. It's gonna be really, really healthy, I think, for society to be positive.
Calvin (47:17.268)
Yeah, so it's basically democratizing legal access, right?
Bob Ullery (47:17.62)
Law specifically, I think it's the one that's always poked out to me, right? Because you really think about, like, there's a lot of intrinsic value of a lawyer, right, with the expertise in motion, lots of trials and so on. But ultimately, what lawyers do is they look up reference material for precedent against their cases. And so, historically, they spend a lot of time in the library, right?
looking at microfiche, looking at law books of settled court cases and so on, and like to do that literally instantaneously now, kind of interesting. So the question, I guess to everybody would be like, all right, well, what's the new role of the existing lawyer in this new world? Like, what are they gonna do to evolve and continue to provide value, if not attach the liability to their organization for representation?
Kevin Nuest (48:10.478)
trust, right? That trust factor is something we've talked about of there's still this bridge between the power of AI and humans and. Not we're on the edge of this. I think was it Calvin, maybe you found this GPT. Um, and, and so we're out here. We're in, we're in. GitHub repos and referencing, uh, white papers and publish articles that have just been written. And so how is, how is go to the hundredth million.
person down the line in society, going to go find the right one. Is this the right GPT? Is there going to steal my data? Is it going to steal my identity? All those questions, right? So there's still a great role to be played for trust, but then the question still remains from a functional standpoint as well. Great. So where does law level up? We talk about that in other professions of now you don't have to do that work. Now you can level up and come over here. It might be my ignorance of.
what every lawyer is saying like, great, I wish I could go spend all my time on this other thing, but from an outsider, I definitely have that same question of where, where do most lawyers then add value beyond trust it with the rest of their 90% of their time.
Calvin (49:22.72)
And I think you asked the right question too. Like if you think about it, I think, you know, humans just don't like change that much, right? We just, technically we don't like to change. So we need to look at, you know, what is the risk of us not changing and have that be very present in our faces. And I think the lawyers just haven't needed to change as much in the past. There wasn't a hell lot of change in terms of adapting.
David DeVore (49:23.004)
Well, I mean it.
Calvin (49:50.7)
I mean, yes, you adapt to the legal situation to surroundings, but the profession itself hasn't needed to adapt. So now giving this additional pressure, I think is a great opportunity to have great change in there. For example, like what we said before, like can an A-class lawyer suddenly service the long tail because they can offer the services cheaper because 90% of it is now automated and comes at a fraction of the cost. I'm not saying this is what's going to happen.
But like there's some of these scenarios that might come out of this that are now feasible and like just pushing the need to change and giving a higher risk of not changing to a profession, I think could do a lot in advancing it.
Bob Ullery (50:30.852)
That's a good point, right? It's like the way the lawyer persists in this new world is through scale, maybe, right? So instead of doing 10 cases a year, you can do 100 cases a year. That drives price down for those new 100 cases, yet your net revenue probably stays consistent. But that reinforces the fact.
or I guess thesis around first mover advantage and laggards in this new world, right? So if you can do a hundred cases now and all lawyers can do a hundred cases, there is not a 10 X quantity of demand for cases. So this is gonna create vast competition and therefore those who adopt AI to achieve those efficiencies are the lawyers that will remain when the dust settles.
Kevin Nuest (51:23.202)
You're assuming it doesn't create a hundred X new cases though. There that is, that is a concern by some that litigious United States opens itself up to, okay, now let's have a thousand X, the number of court cases and filings and lawsuits because the threshold to do it and manage it and, uh, from a filing standpoint, it goes way down and now we have a court system that's jacked and not ready to handle that because it didn't come along with the automation early enough. Uh,
Bob Ullery (51:27.926)
Yeah, that's true.
Kevin Nuest (51:52.254)
because they're not incentivized to.
Bob Ullery (51:54.144)
Yeah, and then we introduce AI courts, right? The AI judge will hear your case now. It's decided. The three seconds, it's now decided and you won. Congratulations.
Kevin Nuest (52:01.098)
Yeah.
Kevin Nuest (52:06.67)
Futurama stuff right there. So we got.
Calvin (52:08.212)
Yeah, definitely a few drama stuff.
David DeVore (52:09.32)
I mean, I don't know. I mean, a lot of the paperwork is actually already templatized, you know? And it's not like an AI can walk into a courtroom. And you guys have heard the story where the AI was coming up with its own case law, right? And so I still think there's a lot of room for lawyers. I don't know, man. I think it's like morticians and lawyers are gonna be the last ones on earth.
Calvin (52:39.932)
Yeah, but this one thing I think it's just a great addition of pressure to evolve potentially, right? Maybe that's the best thing that we could get out of it. Maybe you're right, but the profession is not going to go anywhere. I think...
David DeVore (52:47.718)
Yeah.
Bob Ullery (52:50.072)
Sounds like the common theme too is like, you better pay attention and at least put some cycles by trying to figure out how it impacts your life, lawyer, doctor, whatever you are. Yeah.
Calvin (52:55.868)
Right, right, at least try and get better. Yeah.
David DeVore (52:57.208)
It'd be it'd be interesting to have we should we should have we should find I would love to find a lawyer who's actually like in it, you know, going hard after it and you know, and get them on here or her on here and it'll be fun. Yeah, I want to know what they're doing.
Calvin (53:11.976)
Yeah, let's get one on the podcast. Yeah, 100%. As we're coming up on time here, I think there's one last story that's maybe kind of fun to share. Bob, you want to want to get into the toothbrushes? Just a bit.
Bob Ullery (53:26.633)
Yeah, I do. I'll start with a retraction because when you read the article itself, apparently, the original article was in German and it was mistranslated. The German author used the toothbrushes as an illustrative point of what kind of bad things could happen with connected devices. In the article, and they just printed a retraction at the top like yesterday or something.
David DeVore (53:46.952)
Get outta here. The translation messed it up. Oh.
Bob Ullery (53:53.932)
But it's like, this was a poor translation. It like, the word could occur or something. Yeah, it translated could to did, like could occur, did occur, and people ran with it. But it was still interesting.
Calvin (53:57.003)
I'm sorry.
David DeVore (53:58.309)
AI screwed us again!
David DeVore (54:07.519)
Ah!
Calvin (54:10.064)
So maybe just for the listeners out here.
David DeVore (54:11.421)
That's the story right there. Like we just need to stop. That's the story. I mean, I mistranslated the whole article from one language to another. And so what was a possibility was called fact and now has been changed back to a possibility. That kills me.
Calvin (54:32.876)
Yeah. And so just for the listeners here, the original story was that 3 million malware-infected smart toothbrushes were used in a Swiss DDoS attack to bring down the website. So it's kind of a fun fact that this was a mistranslation then.
David DeVore (54:49.488)
There was no toothbrushes used to bring down the website. No.
Bob Ullery (54:53.756)
No, I blame the German dialect for that, but you know, so Calvin, but it opens up.
David DeVore (54:56.722)
But...
Calvin (54:59.629)
Yeah, yeah, not gonna lie, there is some confusion in that language.
Bob Ullery (55:04.128)
You know, even in false stories, there's something to pull out of it. Right. And I, you know, my mind went, when I originally started reading, I was like, all right, how does this connect AI? It replaced the word toothbrush with agent. And, uh, you know, it's something to consider. Like what happens if, uh, your, your mesh, your, your agent network is infiltrated. That's, that's bad, but.
The scarier part is like what happens when your agents decide a DDoS attack is the best course of action. Right? Kind of like another article we had seen around AI playing war games and it always ended in a nuclear strike. This was from like last week. Because quote the AI said, I just want peace in the world.
You know, so toothbrush, AI, whatever, important to lock your stuff down.
Calvin (55:57.964)
Yep, pretty good.
Kevin Nuest (55:58.558)
This reminds me of the story back in like 2000 where it was rumored you could take a bunch of PlayStation twos and make them cluster together to make a supercomputer that could launch missiles and stuff and they couldn't be sold in Iraq or that Iraq had a bunch of them or something like that. And it was this whole thing. And I'm just thinking that was, you know, have my PlayStation two sitting there. Like that thing's not launching any missiles like look how look how few polygons I'm getting out of this thing. It's not launching any missiles.
Calvin (56:28.2)
Yeah, it's like how much compute do you need in a DDoS per device? Not that much, really.
Kevin Nuest (56:28.278)
No supercomputer happening here.
Bob Ullery (56:33.004)
Why is Kevin in prison? Oh, he sold his PlayStation to some guy in the Middle East. It just...
Kevin Nuest (56:36.588)
Ha ha
Bob Ullery (56:38.84)
The wrong guy.
Calvin (56:41.008)
Well, on that note, I think as we're approaching the hour here, I think a lot of fascinating stories this week. Thank you very much for tuning in. And as always, if you want to find more news stories like this and get your daily dose of it, head over to meshmesh.io and subscribe to our very AI automated newsletter. And see you next time. Thanks, everyone. See you.
Bob Ullery (57:05.304)
See ya.
David DeVore (57:05.352)
See ya.
Kevin Nuest (57:05.998)
Thanks.