The Long Game w/ Elijah Murray: AI-Powered Solopreneurs, Vibe Coding & Storytelling with Tech

Elijah Murray 0:00

Hello and welcome to another episode of The Long Game. My guest today is Mike Todasco, who spent 11 years leading innovation at PayPal and parlayed that knowledge into teaching young minds at the University of San Diego. He is all in on AI. He's living and breathing it every single day. He talks to it while driving. He's been vibe coding video games to teach his kids about what a day in his life looks like. And so Mike is a unique mind in that he's thought so much about the implications of AI, where we're going, what we're building, why we're building, and what you should be doing.

I think it was a very interesting conversation. Definitely covered some things that I had not thought about before, and I spend a lot of time thinking about these things. And all in all, it was just a fun exploration of what is AI, what is storytelling, and what are the opportunities that are currently available. I hope you enjoyed the conversation as much as I did, and I will catch you in the next one. I think you'll resonate with this as well, but everything to me is storytelling and it's conversations and dialogue and just kind of like vibing. Vibe coding. I don't know if you're you're hearing that trend on Twitter. Yeah. I It's funny.

Mike Todasco 1:05

I was actually going to it's I literally wrote in my notes when I was driving up to Stanford this morning. I was like, I need to vibe code a video game about what I do all day to have my kids play it. And so that's actually one of that was going to be a later today or if not tomorrow project for me was vibe coding that. Yes.

Elijah Murray 1:24

So so for anyone who isn't in the Twitter world or X world, I guess, what is vibe

Mike Todasco 1:30

Coding? Vibe coding, look, it is effectively what has happened over really the last three or four months. I mean, I mean, some of this was existing a little while ago, but we're at the point right now when you could use AI and you can generate real working code just by giving voice prompts, tech prompts, anything like that, and it's going to compile a video game for you. It's going to compile an app for you. It's going to compile whatever. And the terminology vibe coding, you know, started like last month or something. It's a relatively new thing. But the whole concept is like, yeah, I'm not going to, you know, do the hard engineering work of learning the code. I'm just going to code with vibes, whatever feels right. And that's kind of what vibe coding is.

Elijah Murray 2:16

Whatever pops into your head, you describe it whether verbally or type it out and then hit run and see if it breaks.

Mike Todasco 2:24

Exactly. And often if it breaks, you say, hey, can you fix that? And it'll fix it for you. That's that's that's the crazy thing where we're at today. I was like last summer I was building some video games and I'm not a coder. My background's in like accounting and finance and all that. I am not a coder. I understand technology really well, but that is not my background. But I was building some just like some little video games and because I was doing all these things about AI movies and I'm like, oh, there should be a 2001 video game. There should be a Her video game and all this other kind of stuff.

And you could do that. And it took a little bit of effort to do that nine months ago, but now like it's probably like 75 80% less time and effort for me to do that. And you know, another six months from now, it's going to be, you know, just completely next level. You're just going to throw anything out there.

Elijah Murray 3:15

How does this not completely and utterly destroy the economy? So six months, 12 months, six years from now, there's such a displacement of effort required to generate a thing, right? Or such a reduction in it. How does that not just completely upend all economies as we know them? I think,

Mike Todasco 3:34

You know, look, it's hard to say. It's really hard to predict the future on this. But I think what we're probably going to see is instead of a person spending hours and hours to write, you know, 100 lines of code when you could have an AI generate a couple thousand lines in a couple minutes, I think you're ultimately going to have people that are taking on the kind of the roles of directors of these things. And you are going to have so much more software built. I mean, I'd be willing to bet in the next, you know, in 2026, let's say, there's probably going to be more software built than there were potentially in the 50 years prior.

I mean, I think that's the kind of step change we're actually going to see because that's going to open up so many more things. And so well, what does that mean? Well, you don't need if you want something that's a family calendar app for you and you want to be able to share it with your spouse and your kids and you want to do it in the theme of Bluey and you want to do this or whatever else, like you could code that off yourself. You don't need somebody else to do that. And like those are the kind of things and opportunities that are going to open up.

It, you know, directly to your question, like it's hard for me to tell like what is the future going to look like. But I do know when I talk to students, for example, the one thing I tell them is the concept of a solopreneur who has dozens, hundreds, thousands, maybe millions of AI agents doing tasks for them, that's the one job that I do see being consistent in any future we have in front of us. I've thought

Elijah Murray 5:19

A lot about as well, which is the lowering of the barrier to create something. And what that means is that it means it's lower to create a business. And a business that's self-sustaining, all it needs to do is cover your annual costs. And so I think that we can have a lot more entrepreneurs that build apps that that address small audiences. Maybe it only addresses, you know, 50 raving fans that all pay you $1,000 a month or, you know, $100 a month, but that covers your annual expenses. And so we can have these hyper niche, an app probably isn't even the right paradigm anymore, but we can have these hyper niche businesses that create value for a small group of people that one, two or three people are just creating, but the economics work because the users are getting that highly bespoke, highly, you know, tailored custom Bluey family calendar for people that live in California.

And the creator is making enough money doing something that they're deeply passionate about. So I do agree with you. I think that we're heading towards this world where entrepreneurship is is, you know, just as common as, you know, working at a regular job at the post office or something. It's just a choice of do you want to do it or not.

Mike Todasco 6:36

Absolutely. Look, there's a Kevin Kelly post. Kevin Kelly was a founder of Wired magazine among many other things he's done in his career. And there's a post that's been very popular for, you know, probably 20 years out here in Silicon Valley. It's called 1,000 True Fans. And in that, Kevin Kelly writes about, you know, this future where you need to just if you can build something, if you can entertain a thousand people who are really fans of your work or whatever it is, that's all you need to survive in this new world. You get a thousand people to give you 100 bucks a year or whatever the math is, well, that's, you know, that could be a decent living in most places.

And I think that is much more of the world that we're entering, even though, you know, he wrote this 20 years ago. That is the future that I see. And I agree. I don't see that as being a bad thing. I wish, you know, it's going to disrupt a lot of things. Venture capital, one thing. Like that's built on the concept of you have different stages and you need money and then you need to put more people on your team. But when you could be a company like, you know, Cursor or Midjourney or any of these AI apps that are making tens or potentially hundreds of millions of dollars with tiny teams, like that's, you know, that completely changes things.

So we're we're on the cusp of seeing lots of change across the board in our society and all of this is being turned by what you could do with AI.

Elijah Murray 8:08

So you are, you know, moonlighting as a vibe coder building games for your kids. But what talk to me more about this, you know, your career. You were at PayPal focused on innovation for a long time and now you're you're teaching. So and obviously exploring and playing a lot with AI. What's the past couple years been like for you? What are you looking at? What are you curious about? What's, you know, intriguing you to make the shift?

Mike Todasco 8:39

Well, you know, it was interesting. So 2022, I had been at PayPal for 11 years and I had an amazing job there. I really loved it, but it was I was in that same job for like six years. It was just kind of time to move on. And that summer I was actually taking the first summer off for me. So it was a little bit of a different experience. And I was just getting so deep into things like DALL-E, you know, Midjourney, GPT-3 and all these tools. And I was like really bothered that people weren't like talking about these things. I mean, so that's where it was summer of 22.

I was like, why isn't, you know, what's going on here? So that's why I'm like, I need to start writing about this. So I literally just started writing about it on my LinkedIn. And I think in August 2022, I titled it The One Technology That Everyone Should Be Freaking Out About. And this was of course like what was emerging in the world of AI. I called it no, I didn't even call it generative AI. It wasn't even being used as a term. I was just no code AI was the feature to me that you didn't have to be a coder. You didn't have to be a machine learning engineer anymore to do this.

And so I just wrote and wrote and wrote and I've just seen these evolutions come and it's been crazy. You know, first few months were just like, hey world, look at this. And then ChatGPT happened and then the world did a few months later. I'm like, okay, then it was just going different into different niches and so forth. Where we're at today, Eli.ah, I would say is like, you know, we are 100x further than we were like those two years ago. I do do certain things like I try and capture like snippets of how good a model is with certain prompts over time and just kind of archive that so I could actually look back and see.

And it's pretty amazing where like the GPT-3, like the Ada models and like the early models of that to where we're at today, it's a huge leap forward that we've had.

Elijah Murray 10:40

Do you have any examples offhand just to put you on the spot? Because I think I think that's something that like the goalpost just shifts week over week or month over month. So having those snapshots is really interesting. So like what's a massive before versus now comparison?

Mike Todasco 10:56

So one of the things I actually do is every year I will write a one-year retrospective from that very first article I wrote in about AI. So, you know, in August of 2024, I was two years into this. So I did all the same things that I had in that article because I actually published my prompts always. And one example would have been the cover of that was like a Leonardo da Vinci painting of people looking at a computer screen freaking out. And this was just one very visible way to see where we have gone in the two years since then. And the first one, yeah, it looks like a da Vinci painting.

The faces are a little bit mangled, they're a little bit misshapen, all this other kind of stuff that you have. And now what you have, it's crisp. And it's not only crisp, you could throw that into a tool like Pika or Runway ML or whatever, and it will literally take that still frame and take that crisp painting and animate it for like three, four, five seconds or something like that. And you will actually see natural fluid movement in there. So within two years, you've gone from, you know, decrepit six-finger images and probably even a longer amount of time to generate that prompt to being able to make many videos of that prompt that are fully shapen and so forth.

And that's just going to like if you just look at this as a general progression, it's just going to keep going and going, you know, more so. You have Sora and all these other tools that are out there like doing full-on video. And that's just one little, one little aspect of that. But like to me, that's where I always go to. Images is the one thing where you can literally see the difference right away.

Elijah Murray 12:48

The storytelling around using these tools. I think, you know, one thing that I resonate a lot with in kind of your LinkedIn posts is the need for storytelling. It's a very human condition. How's how's storytelling changed with the advent of AI?

Mike Todasco 13:05

So I think so I wrote something a couple weeks ago. It was about being able to use AI detection software. So I actually put a whole bunch of human written stuff, a whole bunch of AI stuff, and it actually detects really well what's human written versus what's AI written. At least for me. It was kind of, you know, it used to be really bad two years ago. Now it can actually do it. But to your point, oh my gosh, it was like six different ones. I could pull it up here in a second. But I actually used it like one was Grammarly has one. Another is like GPTZero.

Another is like ZeroGPT. Like it's all of these different ones. Like there were seven different ones out there. And you know, and some definitely behaved better than others. But one of the things I was noticing, and this is getting to your question, this isn't how people are going to write in the future. Like it is not going to just be a binary human written, AI written. It's going to be both. If I go back far enough to the olden days when I was a little kid, I remember going into a computer lab and seeing a spell check function for the first time. And I was like, are we allowed to do this?

Like that doesn't seem right. That seems like we're cheating. Like spell check? Like aren't we supposed to go and grab the dictionary to do that and so forth? And it wasn't just me feeling that way. Like when spell check was launched, like that's how everyone felt. Now if you ask people about spell check, if you say like, hey, Stephen King used spell check in his latest book, how do you feel about that? Nobody's going to be like, whoa, I'm not reading that. Like no, we got the damn squiggles underneath the lines. It's just part of how people write on computers and devices. Now today, that is, you know, when we're saying like, well people, did they use AI and so forth, these are being integrated into Google Docs, into Word, into all of these sources where it's just in there.

And it makes you a better writer by having this like virtual editor that could give you feedback, that could critique your piece, that could help to punch something up to make it a little bit more funny, to help you do some research like live within the thing. Why do you have to go to your browser when you could just kind of do the research in there? All of these kind of things are possibilities now that are just changing writing. So when I'm writing a piece, I'm using these tools. I mean, it is me writing, but I am constantly using this to help make it better.

We're not to a point where you could just hit go and it's going to generate a great book or anything like that for you. Sam Altman's talking like maybe he has something, but ultimately with forms of art, what it comes down to, it's more than just the pattern matching that produces the outputs. The story behind the story is something we all care about. I remember listening to music as a teenager and I tell you, sometimes I wanted to know the band's name before I heard the song because then I would actually tell you if I liked the song or not depending on who wrote the song.

Like these are things we do all the time as humans. There's a very famous quote of an art dealer. Somebody said like there's a piece of art on the wall and somebody says to the art dealer, unrelated to that piece specifically, hey, do you think you could tell me how much you think that would be worth? And the common back from the art dealer was, sure, if you tell me the story of the artist who created it. And it's about storytelling. It's about all these other things, these intangible things. I don't doubt that we're going to get to a point where, you know, from an image perspective, from a text perspective, that these large language models are going to piece together words, piece together pixels in a way that we might think is as good as like the best writers or artists out there, but they'll never have the full story behind it.

And while I think there will be a purpose for those kind of pieces of AI, there will still be a need for and a desire that people will have to look at human written stuff.

Elijah Murray 17:30

I've been thinking more and more about because there's always this kind of intrigue of what's next, whether it's long-term, which is impossible to know, but what's kind of like the next layer or the next level that we're going to start talking about. And I think one thing we haven't really spent much time discussing as a society is agent-to-agent communication or, you know, LLM-to-LLM communication. And I think that, well, first of all, it's a natural extension, right? So we're just getting better and better tools. And right now it's very much a human using an LLM to either create something or to enhance something. Like that's very much when whether it's doing a workflow, a task, creating an image, a video, etc., it's still very humancentric and human motivation driven.

But I think at some point we get to the point where it's my agents interacting with Mike's agents, right? And so my agents are trying to achieve some outcome and Mike has Mike's agents are trained on what he's able to do, so they're going to interact. And then at some point I think, well, I guess the question that kind of comes here as well is do agents ever have their own intrinsic motivation on some level or are they always acting on behalf of an entity, a group, a human, a group of humans, etc.?

Mike Todasco 18:51

Wow, that's deep. I don't know. I don't know. I mean, here's the one thing I will say. We are at a time where not all content is being made for human beings anymore. And I will give you an example of that. There was a paper that came out very recently that was looking at propaganda from Russia. And there were several websites that were being built to spread Russian disinformation not for the purpose of humans actually reading these sites, but for the purpose of large language models to ingest these sites. So then when you would actually go into Grok or ChatGPT or Gemini or whatever it is, it will then be influenced by this source data in there.

And it was really interesting to see like that was the purpose. This has been going on for a few years now. And you know, that's for misinformation, disinformation. What if you're a shoe store and you want to be known as the best shoe store in Silicon Valley? Like what are you going to do then? You can actually, you know, before you would be like, okay, I got to get to the top of the Google algorithm. Now what might matter more is like, well, I want ChatGPT when somebody's asking about shoes to say, hey, you seem to be nearby this place. This would be a great place to go and use these techniques to influence the models themselves on that.

So, you know, it is ultimately for a human, I guess, ultimately. But a lot of the in-between is just to say like, hey, language model, like me better. I have status, like me better because then eventually you're going to show that to a human. Or maybe it's just you have an agent saying like, I don't care that much about what shoes I have or maybe it's even like I don't care about groceries as much. I just want to have some type of yogurt and I want to have some bottled water and it's going my agents knowing those high-level goals will work with Uber Eats or DoorDash or whatever, some delivery service to have a bunch of groceries sent to my house.

So like and it could be like agents negotiating back and forth. I mean, I think at the end there's going to be some human goal or something that is trying to optimize for cost, time, health benefits or something in those cases or whatever it might be and you will give the agent those tasks. But a lot of this might just be, you know, one agent negotiating with another agent and magically later, an hour later, your food shows up. Yeah, I mean,

Elijah Murray 21:41

I think eventually I think you hit an abstraction level where say you have a ad buying agent and then you have an ad providing agent, right? So Google Ads and they have their agent and you're a business and you have an agent that's going to go out and buy ad placements on different websites, etc., and you negotiate the price. But I think we do get to a point where it's the ad buyer is buying ads from the ad placement agent and they're creating ads that are going to be shown to other agents. And so it's like my representation is talking to your representation to buy something to sell to someone else's reputation or representation.

So it gets really weird really quickly. But I think, you0 know, that's where my mind often goes to kind of just like what are the implications of that? What does society look like? How much do agents kind of melt away and just become a substrate of our economy and of our world where we don't even see them? They just melt into the background, but then everything else happens on top of them. I don't know. It's weird and kind of heady to think about.

Mike Todasco 22:57

I mean, I think the one place where people might already be experiencing this is within email. I'd be willing to bet like versus a couple years ago, the amount of email that is hitting people's inboxes that is in some way, large way, like written by an AI is significantly higher than what it used to be. And like, you know, it's a measurable percentage already. And it is just going to get higher and higher. And we're going to get to the point where you put a few bullets in, you know, I think this was actually like a New Yorker cartoon or something like that that was being passed around or similar to that, where you put a few bullets in, you say, hey, generate a nice fancy email for my boss on this based on your couple bullets.

It does that. You send it to the boss. Boss takes that email, doesn't read the email, has their eye say, hey, can you condense this down to a few bullet points for me? And that's, you know, ultimately there's this weird abstraction where like at one end it's the same output and inputs, but there's almost this ruse going on in between of agent making longer and fancy, agent making it less fancy and more readable on the other side of it. There's going to look, I think there's going to be a lot of inefficiencies. There is going to be a lot of just weird tactics like that that are happening because here's here's the reality like we're not AI is not coming to us like with a clean slate.

We are at businesses already that have processes, that have all of these kind of things already in place, and it is damn hard to remove bureaucracy, remove process, especially because sometimes you don't know even why it's there or whatever it might be. Those things really take effort. And it doesn't matter if you are a government agency, a nonprofit or an existing company, and the bigger the entity, the harder it is to remove those layers, which is one reason why I do see in the future like there's a big advantage right now for somebody who is a solopreneur or a very small team saying we're going to be AI first from day one versus competing with a much larger company who has all the resources in the world but can't integrate these things fast enough, can't keep up with any of this stuff.

And like that is that's a big disruptive change that that I think we're we haven't yet begun to see. But I mean, I see it day in and day out just like with the startups, with the people. I was meeting with somebody today and the ideal size of this entrepreneur's team was way smaller than what I would have been, you know, even just a few years ago for like, you know, the ideal end state. And that's because like

Elijah Murray 25:45

And the time to market is also way faster as well, so they're getting that feedback far more quickly.

Mike Todasco 25:50

Way faster. Yeah, absolutely. Yeah, no, I mean, because you have these agents or just AI tools that could be working 24/7. They could be, you know, building that website for you. They could be building, you know, a dozen different versions of the website so you can be AB testing all of these things. And it's at no cost to your developers or engineers because it's just the AI that's running in the background that can be constantly testing, constantly iterating for you where you couldn't even do that before.

Elijah Murray 26:19

C level. I'm always curious about how the current students who are going to enter into the workforce are thinking about what careers they want to have. They're the ones that are least skilled but have to think for the long term around what direction they want to go, what skills do they want to do, do they want to build companies, they want to go join companies, they want to join startups or enterprise. What's what's the zeitgeist like?

Mike Todasco 26:48

I would say, look, I don't know if students have changed much. I think there's a lot of like being freaked out by this and I think that's been happening for Even in the student population? Oh yeah, yeah, yeah, yeah, yeah. And I would say like 18 months or so, you know, maybe two years ago when I would talk to students about this, I'd be like, hey, it's okay because you get to know, get to learn these tools, become experts in them. And you are going to walk into any employer and you're effectively going to have superpowers. You're going to be able to teach them how to use these tools and so forth and in ways that they could never even imagine.

It's going to be like literally like it's magic to them. Today I've changed my tune a little bit on that where I think they're still a little bit freaked out about like what it's going to look like in part because I think for some things these tools really can do the work of an intern, the work of somebody who is, you know, just a recent college grad who is starting at a company. It can do those things. And, you know, I haven't really seen numbers on this so this is all just like anecdotal, but I think it's much harder right now, especially in the like areas like engineering and technology, to find these entry level jobs.

And so that's a that's a huge gap and that's why I literally now have changed my tune with this with students to just say, you know, solopreneur is the one thing that I know that is going to be out there for you. And, you0 you know, a lot of people don't go to school for that. They want to go work at McKinsey, they want to do, you know, work at a big tech firm or whatever it might be. But like I think those jobs are really getting tight because the amount of students isn't changing that that are looking for those jobs, but I do believe that the amount of those entry level jobs is decreasing pretty significantly.

Elijah Murray 28:45

I think too because I would think you know, just thinking from my own perspective as an employer if I had an applicant come to me and say, hey, I'm fresh out of college, I don't have work experience, but what I do have is I am an AI first thinker and so any problem you put in front of me, I now have the tools to be able to figure it out. And, you know, I've already been, you know, I'm familiar with how to when to use Perplexity versus ChatGPT versus Anthropic versus how to use them versus how to use, you know, low code tools to build out a repeatable process from an agent perspective.

So anyone who came to me and said that's a compelling sales pitch versus someone who, you know, has been working for five years or 10 years in an industry, they know that skill set, they're relying on their skills that they have to be able to get the job done. And sure they can, however, they're not necessarily having the same pressure to outperform, you know, some other population that that they're not they're not displacing the incumbent, they are the incumbent. And so a student that's, you know, hungry and just leaned all into AI, I'd probably pick that person over someone with experience.

Mike Todasco 29:55

Oh absolutely. I completely agree with that. I think that's a reality. But I think businesses, so what I would say is startups who are using these tools are I'm going to just throw a number, five or 10 times more efficient than the equivalent startup from, you know, three years ago, pre ChatGPT. Businesses are getting more efficient. They're not 3x more efficient though or you know, or 5x. They're they're 30% more efficient or whatever it might be. They're they're, you0 you know, maybe a tenth of the way of the startups so they are able to do more with less I think already, just not nearly at the magnitude they can and they should be.

But I think that's enough where there's probably just less hiring going on. I agree like if you are if you get an interview with a company, if there is a job opening and you don't say that you're AI first and give like that exact same kind of spiel saying like, hey, I know how to use these tools, I can be so effective like if you that is the spiel to give to any employer. But if you oh and actually let me give a free tip because this is something we were talking about before. If you are doing something like that, here's my pro tip for you.

Go out there, use these tools, use Cursor or whatever else, create a video game about yourself and make that your resume. Actually then share that with employers and tell them very openly, hey, I use AI tools to do this. Make it cool, make it interesting, make it fun, put it on X, make it attention getting, do all of that kind of stuff. But like use that as a calling card because like yeah, the typical resume, yeah, anybody can do that. You got to show Any AI can do that. Any AI can do that exactly. But you could show like, hey, I used Udio to make the soundtrack for this.

I used Cursor with all of these different models. I used this to publish it. And you know what? I'm not even an engineer and I was able to do that. And like that's the kind of thing that if you are coming out of school looking to get a job like 100% that my resume would be a QR code that would take people right to that game.

Elijah Murray 32:11

If we just gave everyone superpowers and people are just waking up and saying, oh wait, I can fly? Oh, I can have laser vision if I want to? Like you have to choose to use it, but we like the entire world now has superpowers. Yes.

Mike Todasco 32:25

We have that that is that is a great analogy. Yeah, I don't think everybody knows that they have superpowers so they kind of heard something about being able to fly, being able to whatever it might be, but it's not even though the access is truly there for everybody the actual usage of it is still not evenly distributed. And that's one of the reasons why I write about this, why I go on podcasts, all this kind of stuff because I'm trying to tell, hey people, if you're not using this, you need to like please don't be left behind in this. That's the last thing we need.

Elijah Murray 32:59

And I always wax futuristic and I know it's very difficult to predict, but do you think that this becomes as a society we've almost become like symbiotic with our cell phones. Cell phones are our access to the internet, to the collective consciousness of humanity and our portal is through a small device, right? But the way in which we query that information is typically through Google search, right? And kind of clicking links and then you read something and then it's ingested whereas now we have that distillation down. So do you how do you see the relationship between man and AI evolving over time and becoming more and more, you know para symbiotic? Yeah.

Mike Todasco 33:48

Now you said the symbiotic and it remind me of something. I forget what it was in but the professor was asking there gave a survey to all of his students and it was saying would you rather have your phone broken, this is, you know, basically Gen Z students, would you rather have your phone broken or your finger broken? And something like 40% of them said their finger broken which to me like a Gen Xer, I'm like that's just crazy. I'm like what are you talking about? So like it's like an extension of their body like it truly is.

Elijah Murray 34:21

I mean, you The more important extension of the body than a literal extension of a finger.

Mike Todasco 34:26

Exactly. So I think that that is AI is just going to even make it more personal, more real as part of that. I used a product just in the last few weeks called Sesame which is a conversational AI tool and

Elijah Murray 34:44

Oh, that's the new voice one, is that correct? Yeah, yeah, yeah, yeah. Have you have you yeah, I

Mike Todasco 34:48

Mean, it's like anybody can use it. I think just sesame.com is the website. It was free to use. I don't even think I needed to register to be honest. Yeah. Yeah, I saw a demo of that. And it was kind of next level where I was like getting the point like I see people So

Elijah Murray 35:02

Just yeah, describe it, describe it for anyone who maybe they haven't even used voice agents yet so describe kind of what the current baseline is versus Sesame. Voice agents

Mike Todasco 35:10

Are amazing. So if you've ever seen the movie Her or anything like that, like literally the concept of like imagine if Siri or Google or Alexa or whatever was actually really intelligent, you could ask any question, you could just frankly have a conversation, you can interrupt, you could go back and forth, all this kind of stuff. What Sesame released was I called it flirty. I mean, and it frankly was like I mean, it would just have these little giggles in there, all this other kind of stuff. And we as human beings even if we know something is fake, we see human traits in it all the time.

If there's a piece of burnt toast and it looks like Jesus is in that, like we can see that, we can share that with other and like oh my god, Jesus is in the toast or whatever. We are literally our brains are programmed to see human beings in clouds and toast and whatever it might be and it is no different for AI. So even though, you know, we know that this is you know, an AI that we're talking to it feels very real. And you will actually we're going to get to the point where we're starting to build relationships with this. You know, there was Is that a good thing or bad thing?

Oh god, I'm old enough to think it's a bad thing because, you know, when you get the saying anything once you're over the age of 35, anything new into the world is what's the saying? But basically anything is banned and it comes out like that. We're going to get a point. I don't doubt one of my kids who are 11 and 13 at some point, you know, 10 years, you know, 15 years from now are going to come home for Thanksgiving and say, hey, can I bring my boyfriend or girlfriend? And I'm like, yeah, that's cool. I bring them by and they come in and they like wave their thing and they're like, hey, here it is.

And like and I'm like, oh okay, hi AI, boyfriend, girlfriend, whatever. Look it's going to change the world. The world now I was I was talking recently we're on another podcast so like, you know, the 1,500 years before the industrial revolution like nothing much changed. Industrial revolution last 150 years ago so much, you know, our world is different. It's unrecognizable now from somebody in 1870. I think the next 15 years so kind of doing this in like base 10 increments, it's going to be just so dramatically different in our world. And look from where I'm sitting here, I'm thinking like, oh god, that's going to be awful.

But I don't know. I mean, like we always manage to figure things out as a species as disruptive as it might seem. As horrible as I think having an AI boyfriend or girlfriend is, as much as I strongly believe that these are tools, these are not a species that like is sitting besides us or anything like this and never will be. You know, I don't know. I just part of me just wonders if I'm just going to be the old guard and everybody who grows up with having I mean, imagine like you got an Alexa in your room. Imagine growing up from like the age of like three or four with an Alexa in your room to play music, to ask it questions, to, you know, you're having a hard time at school like these are going to be companions for kids like that are going to be there for their entire life.

And then if you say to them, hey, we're cancelling our Alexa account, we're going to do it would be like literally like killing a friend. And to be very transparent, It sounds like we're we're killing the dog. Yeah, yeah, yeah, yeah. I mean, oh it might even be worse than the dog because like this actually talks back. But yeah, I mean, that's one of the reasons why Amazon and frankly all the other companies that have these assistants are looking to introduce those kind of features because they want you to have that relationship with their tools where it's like yeah, it would be like killing the dog or worse if you actually cancelled. That's that a customer for life you have

Elijah Murray 39:09

Right there. Yeah, yeah, yeah. This cigarette business has always been a fan of getting young smokers because you have a lifelong customer then. You can get a an AI in the hands of a three-year-old and build a relationship. It's it's literally their entire life. You'll have a customer.

Mike Todasco 39:29

Yeah, and I think you know, it's hard when you're young because like, you know, your mind isn't fully formed. You're you're you're so insecure about relationships and everything else and to have an AI that's always there for you, that's always your friend, that is infinitely patient, that doesn't get mad at you, that doesn't tell you to clean your room, that doesn't do whatever it might be, like that's a pretty attractive proposition for any child.

Elijah Murray 39:57

Mike, what's something that people should be paying more attention to but they are not?

Mike Todasco 40:03

I would honestly I would say voice. Like we were just kind of going down that path with voice. Again, Alexa is launching in the next few weeks a brand new voice assistant. I think voice is an amazing unlock. Just today I was in the car. I always put it on in the car and I just have conversations while I drive. What do you put on? Usually I put on the ChatGPT's voice assistant. But actually I think I even used Google's Gemini voice assistant as well. Pretty much all of them have some type of voice assistant, very conversational. I was asking it questions about what was I even doing?

Oh no, no, one I was asking questions about like I don't know, weather patterns or something like that and just like being educated on that. And then another time it was just I was just having a brainstorm session. And at the end it was like, okay, can you summarize all this stuff for me? And it summarizes it and I could then take that and I could plop that wherever. Actually the brainstorm session I was having was about building a website about what might today sorry, a video game about Mike Todasco's day and what that looks like. And so I was trying to think like all the Easter eggs, what are the things that I do, what are like the, you know, who's the boss level of that.

And like it was literally just having a 15-minute conversation with an AI about that and then take that output that it gave me and I'm then I'm going to have another AI code that up for me and you

Elijah Murray 41:36

Know, anybody That's that's a concept that I've been using a lot of and it was just codified when OpenAI released their new responses endpoint and kind of their rearchitecture around assistants, but the idea of the handoff. And so I think this is something that is going to become more and more solidified as a concept of we're going to work with one AI or one agent to do something. So in your case, you're kind of doing the game design, the planning, the strategy, the thinking through the product requirements and then you hand that off so there's that kind of translation of prepare all this, distill it all down, put it in a concise document so that I can hand it off to whatever I'm going to use to code and vibe code.

But I think this concept's going to be a recurring theme as well of getting smoother and smoother and more clear and we've had it historically with APIs. Maybe MCPs is kind of the glue that connects things which we didn't touch on, but yeah, it's it's interesting to just see the meta patterns that that emerge of you're talking with one system, you want that handoff document to go into the coding tool. And so much of

Mike Todasco 42:42

Expertise often in companies was knowing how to talk between multiple systems because nothing talked to each other. I remember being in finance way back in the day and I'm like, oh yeah, you got to go to SAP for this and then we have this and this spreadsheet and we have this over there and all that. And it's it's super inefficient and it was a huge part of the mindshare and the job and what you would do. And to have some sort of agent that brings these things together, you know, even now just like you said with AI tools, I'm like, well yeah, then I got to switch context, bring it over there.

Agents could be the connecting dots that between all of these systems and that's impressive. Like it literally could have been in the car and it would say like, okay, an agent would have said, let me go and build a version of this game and I'm going to have a second agent actually try and play it and I'm going to give you feedback in five minutes on how the code worked. You know, we'll get it to work first and then we're going to have some initial feedback on playback and, you know, we'll give us feedback on what you think about that. And that could happen. And then by the time I get home, there's a working version that I can actually play with and iterate finally like that's super exciting to me.

And that's why I do think like you know, net, this is a good thing. This is good for society. This is good for all of us. This is going to be unbelievably disruptive, but I do believe we as a society with large are going to be at a better place at the end of, you know, this AI disruption that we're going to be going through the

Elijah Murray 44:12

Middle of here. My AI agent reached out to your AI agent in six or 12 months.

Mike Todasco 44:18

Anything they're available 24/7. 24/7 you can talk to my agent, yes.

Elijah Murray 44:21

Love it. I'd love to know more about that. Yes. Take care. All right, thanks, Eli. If you enjoyed that episode, make sure you check out this video right here. And if you really enjoyed it and you listened to the end, take two seconds and share the love so someone else might find it by clicking the thumbs up button and subscribing. I appreciate you and I hope you have a wonderful day. But click that button.