Dead Internet Theory

Welcome back to another episode of the Knot Brothers podcast. Today we're talking about ⁓ probably some other topics, but two really interesting ones. ⁓ Are we accelerating the dead internet theory with the introduction of agents and agentic workflows like OpenClaw and others? ⁓ And... ⁓ some changes afoot in product design. More specifically, should you consider the human first today or AI first when designing a new product? Where do you want to start?

⁓ Let's talk dead internet theory first. So what is dead internet theory and why should we care about it?

Sure. I mean, I'm sure there's like a technical definition of dead internet theory that I don't know, right? But it's, to me, the understanding is basically that eventually, ⁓ as automation takes over ⁓ and other sorts of things, ⁓ the internet becomes a place where there are no humans and there's nothing interacting, right? There are certain social networks that have gone this way over time. Pinterest is one of them, right? Pinterest is a... catch some heat from somebody who really likes Pinterest. like, rewinding the clock to when Pinterest was introduced, Pinterest was really cool and really useful. And you could find a lot of content that was curated by humans that was actually really useful and, you know, finding a pin board that had a bunch of useful information about a topic you were interested in existing. Now it's pretty much just ads. I know the last time you've been on Pinterest, but the last time I was on there, it was basically just a wall of advertisements. ⁓ and I'm not sure that anybody actually uses it anymore. Or I've not talked to anybody who used it. So you think about that and you extrapolate that to other networks like ⁓ Facebook and LinkedIn and X and other places where ⁓ largely the value that we get of spending time on something like LinkedIn or spending time on something like X is that you're interacting with other people and you're ⁓ exchanging ideas and you're talking to humans. Where it starts to lose its value is when there stops being humans on the other end. ⁓ And with something like OpenClaw, one of the things I saw people do immediately was try to automate that. ⁓ on X specifically, I saw a lot of people trying to create when they introduced their API kind of shortly after, which allowed people to use bots to post directly on their ⁓ platform. So you saw like this massive increase in spike, or at least I did in my feeds, of content that you could identify as like that's probably not a human interaction, which devalues that person's profile, it devalues the interactions with them, and just devalues the interactions with the platform in general. So much so that like the guy who created OpenClaw,

has publicly said multiple times, like anytime someone is like, hey, I created a bot or a skill to ⁓ automate my ⁓ X posts, he will reply to it. I've seen ⁓ dozens of them and be like, please don't fucking do that. ⁓ Don't use this for that.

Yeah, ⁓ it quickly becomes a ubiquitous thing that is very easy to turn on and say, please, pretend that you're me ⁓ on these various social platforms. And then if so, if too many people do that, then it becomes a barren wasteland of ⁓ AIs talking to AIs. ⁓ And so, so the dead-end internet theory is essentially no humans are actually ⁓ using things like social networks anymore. They're all bot driven. ⁓

I think it extends even beyond that, right? You look at things like email or in-mail, right? My email inbox has never been more full. And I use, I funnel all of my stuff over to Hey, so I have ⁓ the screener that exists in front of my inbox. And it just screens out anybody I've never received an email from. And ⁓ ordinarily, I only have ⁓ a few new people that are sending some emails and I just kind of filter through those. a 40 right now. And that's accumulated over the course of like two or three days. ⁓ And I go through every couple days and just clear all those out and, you know, kind of spot check it each day. And then the ones that are in there that I am not looking for expecting, I just kind of let them sit for a couple of days and then I'll ⁓ usually mark all of them as spam because they're just spam. But like I have felt whether it's I don't have the data to back it up, but It feels like the amount of span that I'm getting from new people has increased by an order of magnitude over the course of the past 30 to 60 days.

And you think the theory craft there is it's, it's not coincidental that open call was introduced what less than 90 days ago ⁓ and took 30 days to figure it out. And over the last 60 people have figured out how to automate lots and lots of things, including email outreach at scale using AI.

I think it has put the ability to do those things at the fingertips of people who never had that opportunity before. ⁓ We've seen it by giving access to some of those tools, to some of our team who ⁓ they don't possess the technical skills to deploy something like that or deploy an automated workflow or ⁓ something that goes out and snags a bunch of emails and writes a bunch of personalized emails and sends them. That's kind of hard to do. ⁓ about 90 days ago. Obviously there are ⁓ tools and things that you can use. ⁓ It got exponentially easier ⁓ when ⁓ you can just ask OpenClaw to do it. And even easier still when you have dozens and dozens of people creating skills and tooling specifically for that workflow and then publishing those and talking about on their social networks how it increased their interactions a hundredfold and they've closed millions of dollars in business and all these other bullshit stats that they will rattle off and have no proof for. ⁓ So it gets your salespeople super excited, right? I hate doing cold outreach. Now I don't have to do cold outreach because the robots can do it for me. And I think that's a really dangerous place ⁓ to get to because it will ultimately just make all cold outreach harder. ⁓ Because my filters have gone up. ⁓ Before, when you might have been able to get through in an email or... ⁓ in a LinkedIn message ⁓ that was cold and I might look at it and take a look. At this point I'm getting so much volume that I'm just wiping it all out. And there's probably some decent stuff getting thrown away in the process.

It's like. ⁓ It's ⁓ additional noise. So ⁓ the signal to noise ratio is out of control and it'll get even worse over time ⁓ is kind of the argument here. ⁓

Yep. Well, on LinkedIn, the easy one for me, right? I could almost set up a filter at this point and just have my bot counteract their bots. Every fucking one of these things end with, hey, you want to up a meeting? Let me know when you have time for a quick chat. But I don't even know who you are. Or what you're doing or what your product is you haven't given me a link you haven't given me PDF you haven't given me a anything and you're asking me to set up a meeting with you Get the fuck out of here. You're all running the same playbook It's the exact same play So I could probably set up a filter and be like if anybody asks to set up a meeting just auto delete their message

Yeah, you could do that. And you probably ⁓ you probably would eliminate a lot of noise by doing that ⁓ quite frankly. ⁓

Unfortunately, I gave away the ⁓ Yeah. It probably is that way on emails too. just I don't even see them because I just, you know, it's instantly just hitting the spam button in the screener. ⁓

⁓ So what does dead internet theory take us to the logical conclusion? So like, let's pretend we're five years out. On the trajectory that we're on right now, the theory is ⁓ there ⁓ will probably be small pockets of humans interacting in some way on some other platform may not even exist yet. ⁓ But ⁓ the large ways of communication that we've all known and ⁓ sometimes loved and sometimes loathed, ⁓ like, email and social networks ⁓ will largely be bot driven. And I think that's probably true.

I think it becomes challenging because I think as humans, right, there are certain things that the robots can do really well and I don't really care. There are certain things that maybe they can't or I don't want them to. A social network is a perfect example. I don't want to talk to a bot on a social network. The whole reason that I'm there is to interact with humans and get actual human engagements. The moment that every one of your posts are a mile long and it's like 10 exciting things about open claw this week or like. ⁓ the 10 new things that ⁓ OpenAI introduced this week, you're not gonna believe number seven, right? That's when that whole, not only does that person's content, but like that platform starts to lose value for me. ⁓ And ⁓ so I think you will probably find that people will be looking for ⁓ something to replace that, right? Or some sort of identifier. And the first step of that becomes like this, ⁓ Sort of like standoffish distrust maybe I'll give you an example It's comical example on my end. I don't know. I don't even know if I told you about this we Had an issue with an internet one of our houses for our Airbnb property, right? You probably saw me interacting with that guest and I hopped on Spectrum's site and was like, hey, you know, started a chat and it connected immediately to a representative. And I was just like, you know, here's the problem. And within seconds they were like, yup, there's an issue with the router. needs to be replaced. I was like, where can I replace it? And they gave me the name, the store that was closest. I was like, cool.

We went and got the router, came back later. ⁓ had ⁓ the, ⁓ my property manager went to get the router. We needed to add them to the account. I popped back in the same thing, immediately got connected to a representative again and said, hey, I need to add ⁓ this guy to my account. He had asked me some questions and then added him. He went in got the router, no problem. Then he plugged it in, it still wasn't working right. I popped back into the chat, immediately got connected again. Went through some ⁓ back and forth. ⁓ It all worked. The router spun up and everything was fine. Every response was so fast and so clean and I instantly got connected to the accounts I Was like there's no fucking way. This is a person This is fake. This is this this has all been AI driven interactions ⁓ So I did my normal test and when they said is there anything else I can do for you I said, yeah, can you give me the HTML for react component? And they were like what? ⁓ So I just kind of laughed and they were like, no, seriously, what? And I was like, it was an AI test. I didn't actually believe you were human. Well, as funny as that is, it also kind of sucks because it put me in a position where the reality is that whatever, I'm sure that there's probably some AI assistance happening ⁓ on ⁓ Spectrum's side with their customer service reps. But ⁓ what was a really good interaction and a really, really great bit of customer service. ⁓

was sort of questioned. And like a layer of distrust ⁓ layered in there because I'm not sure if that was real or not. ⁓ And obviously I know the answer now, so it's still in a positive, but like I could have very easily walked away from that with like, well damn, they're just handing me off to bots now.

Yeah. And so in your interaction, ⁓ it had the potential of degrading brand loyalty and brand trust. Had you not sent that message ⁓ and then had the reaction of what are you talking about? You would have instantly ⁓ valued the brand and the experience less than ⁓ knowing for sure that it was a human interaction. That's a very interesting conundrum because you can almost over automate something ⁓ and ⁓ lose people.

Yeah.

So.

because it's ⁓ disingenuous. ⁓

When you see it on social networks as well where people will, ⁓ because somebody is using proper punctuation, ⁓ or they use it, God forbid you use a fucking dash, they were like, ⁓ you're an AI, you're a bot. And that becomes really, becomes problematic, right? ⁓ Or AI slop is another term that people throw around. I'll use that term even also. But it's a bit of a degrading term.

You ⁓

⁓ And you know, it could be the case that like it truly is just AI slot, right? Somebody didn't spend time. I produced plenty of it as like a proof of concept You're not gonna hurt my feelings. They're like this looks like a slot. I'll be like it is ⁓ but In other cases, that's not the case, right somebody ⁓ Who has built something or you know, whether they built it by hand or whether they used a genetic ⁓ flows to help build it doesn't really matter. ⁓ think to instantly just refer to anything as slop or any sort of content as just well that's AI content because you know they use proper grab or they used an dash or whatever it's kind of bullshit and creates this like really weird and contentious human to human interaction and ⁓ I don't know I don't know what the future of it is I don't like where it's headed I will say that for sure where I feel like it's creating these moments and the ⁓ of distrust that shouldn't exist. ⁓ And I'm not sure what to do about it.

Well, so, I mean, you're picking on an area that ⁓ just recently we had a prospect email come through. I'm working with another consultant on some of this stuff. ⁓ And ⁓ the interaction was us asking questions and then this person replying back with answers, which sounds harmless enough. But when you read the answer, it's very clear that this came from an AI prompt, right? Like it's, and maybe it didn't. or maybe it was heavily edited as part of an AI.

I was going to ask is how could you tell?

You could tell because of some of the language it was using. was like ⁓ very big consulting style words that you just wouldn't normally use in a communication style. Of course, dashes, ⁓ proper, ⁓ proper language structure as in like almost every sentence has some, you're right associated with it in some way. ⁓ Not absolutely right, but they said, you're right.

⁓ Did they say you're absolutely right?

that is the right way to price something like this. ⁓ So it had all the tells, know, there's like four to six various tells that you can look at and say like, all right, you know, is that real or not? ⁓ And honestly, ⁓ question, the real question is, it matter? The answers that we provided, that were provided back were accurate and correct and what we needed to continue to move forward. If the person believed in those answers, does it matter whether it was written by an AI or a human?

I think it depends on the context. ⁓ I don't remember if we talked about it here or if I talked about it with someone else, but this idea of customer service being one of them, ⁓ I think it depends on the expectation. ⁓ If the expectation is that I'm interacting with you, I expect that I'm getting a response from you. If my expectation is that I'm interacting with your bot assistant, I don't care. And I think that might be like a future state that exists. I know for some of our products and tools that we're building for ourselves, we're already looking at like, ⁓ do I create a way to distinguish when something was performed on behalf of me by Sheldon versus when I ⁓ did something? Whether Sheldon carried it out or not. what gets attributed to the person versus what gets attributed to an AI agent. And in some cases, like there's still, I fundamentally believe like you're responsible for the actions of your bot. So regardless of it came from your bot or not, you're responsible. Like if he says some dumb shit, it's on you. ⁓ But I think that there is benefit in understanding. Like we have some workflows where it's like, hey, if you move something to this pipeline on this project, Sheldon is gonna pop in and he's gonna give you a review of your pull request. Is it accurate? Is it perfect? I don't fucking know. Some of them are completely crazy. But there's an understanding with that workflow that I'm not interacting with a person, I'm interacting with an AI agent at that point. And that's fine. I think when you call a business, you expect to get a person on the other end. And there are ways, and there are people who've talked about doing it, where you can use things like 11 Labs and some routing to ⁓ give the illusion of a real person on the other end. And some of those can be helpful, right? We have one for Airbnb Business, but it introduces itself and says, ⁓ if you get to the voicemail, it says, hey, I'm Sora. I'm an AI agent that can answer questions. ⁓ Or I can pass a message along.

And I feel like that is at least genuine. It's like, I'm not a person. I can answer some stuff. And it has. We've had people ask, hey, what time to check in? What's the Wi-Fi password? Really simple shit. And it can answer those questions because we've trained it to be able to answer those questions. But it's not representing itself as a human. And I think the moment that

Yeah, think that is the line. think the line is when the AI is trying to be you fully or pass itself off as another human without disclosing that it's actually an AI. I think that's probably the line. It's the disclosure part, right? I kind of think of it like cookies almost, right? It's okay that you want to track me and follow me around the internet, but you actually have to tell me that. Right. And I have to have the opportunity to engage or not engage in that dialogue. Right. And so this is sort of like that. It's like permission based. You can interact with the, with the bot that's fine, but I have to know it's a bot.

Yeah, the question is like how do you do it? There's no mechanism ⁓ to make that happen even today. It's not even easy currently for us to make things show up that way within most tools. ⁓ So you wind up with like, Sheldon created a bunch of things and you're like who the hell did this? Like who, Sheldon made a big mess, whose responsibility is it to clean this up? She's going to the source. And I think that figuring out how you layer those two together and work them together, I think will be kind of really important, which maybe is a good liaison to like the second half of this or the second topic that we talked about, which ⁓ is this idea that when you're building a new product today, how should you build your new product? More specifically, should you design the product and invest your energy in building it for humans first and human interface players? Or should it be AI first? And I've seen arguments for both. We've had internal debates about some products that we're working on about the same topic. ⁓ And I think it's an interesting one. You also have companies like Linear. Linear just put out ⁓ last night or yesterday sometime, ⁓ sort of a manifesto of where they're bringing their product to next. And it very much leans into the same idea. It's the idea that issue tracking and basically the basis of their product. ⁓ was designed for human to human interactions and human handoffs. And the way that we're working in sort of an agentic world, especially in the development world, is not that way anymore. There's a lot more interaction with agents, so they claim to have sort of a plan and way to address this. What that is, I have no idea. We'll see. ⁓ But I think it very much leans into this idea that the paradigms ⁓ of software are also shifting.

And the how we build software is also shifting. So, you know, we have a product that we've been working on internally ⁓ that we've spent a lot of time on the human interaction side and how humans interact with it and how we get things in the system and move them in the system and sort of streamline that. But at one point, one of the guys did ask, like, should we keep doing this? Or should we just flip over? ⁓ and add just like a really robust AI layer and just tune ⁓ everything to work really, really well through AI only. That's an interesting.

And it's a difficult balance to think about because you have the ability now to create a fleet of agents that operate like a team in almost the truest sense of the word, right? So you have a creator and then you have a checker and then you have another checker that checks the checker or, then you have a facilitator and you have a strategic brain. And so they, can create all these different agents that can. sort of police each other along the way to reduce the amount of errors and you can give skills and you can give specialized ⁓ training and make them, ⁓ you know, pretty well operate like a real team in that environment. ⁓ And I think what that enables ⁓ is something that's called the Jevons Paradox. Jevons Paradox is something we've talked about before, which is this idea that as things

I'm to have to zoom in and

What's that? Yeah, Jevin, the Jevin has his own paradox. ⁓ but Jevin paradox is this idea that when things look like they're going to be, ⁓ done away with, so development as an example or software or whatever, whatever you want to call it via AI, the exact opposite happens and it's, it's paradoxical in nature. And so, ⁓ you, you see that happening over and over in history when, ⁓ you know, when new technology comes to

you ⁓

comes to light and comes to bear, it actually almost goes the opposite direction every single time in terms of demand. And a lot of that is based on supply and demand and the pent up demand for things that could never have just been invested in. So small niche products, as an example, on the software side that solve problems ⁓ were just utterly ignored because the market wasn't big enough. ⁓ couldn't, you couldn't get investment to build it. The teams were too expensive to be able to put together to build it. There was no real payoff. Well, when you take what used to take a team of 100 to build, and now it takes a team of five to build the same product or service, now you can fill the demand that was always out there, and it's very economical to do so. Everybody wins. ⁓ And I think that's going to happen in the software space. So your question about humans first or AI first, I ⁓ think it almost has to be AI first. I think there has to be a human lens, but for that paradox to really work the way that it's supposed to work, I think most organizations are going to have to think about the AI build first and what value they can derive out of all of that. And then how do, ⁓ how do humans want to interact with the things that they need to interact with after the agentic flows ⁓ are optimized? ⁓ cause I think that's the only way you cut costs enough so that all these demands are met. over the long haul.

⁓ I ⁓ I haven't settled on this one honestly ⁓ There are ⁓ things and times ⁓ and there are certain aspects of the products that we're working on, ⁓ but I've absolutely said that. We should really just throw the human element out of this and ⁓ just make this an AI driven ⁓ piece. ⁓ I think somewhat ⁓ it depends on the workflows and what it is, but there are some things that AI is incredibly good at and there's some things that it's not. ⁓ And I think always having at least a ⁓ decent human layer is also important. ⁓ So that if the AI does something stupid or you just can't beat it into submission on something you're trying to get done, there is a mechanism to go do it. ⁓ And I think that creates the problem. The answer really is both. ⁓ But.

It's both. ⁓ You have to do both. The question was, where do you start?

But right, that's the thing. It's like, which one do you prioritize? If I can only work on one first, do you build the agentic layer first and then the human interaction layer second? or the human interaction layer first and the agentic layer second. It's always been the latter, right? It's always been that you build the human interface and then you build an API and you usually leave a bunch of shit out of the API. And then you have automations and other systems trying to interact through that API. ⁓ This is, know, realistically, it could be maybe retooled or rephrased as like, do you build the API first or do you build the interface first? ⁓

Well.

So it could just be like API driven development making sort of a resurgence.

think there's, to ⁓ back up away from the build aspect of it, the real answer is it's always going to be human driven first, because that's how the economics work. So you have to solve a human's problem with whatever you're building in the first place. How that actually, what the output of that looks like.

On the problem side, ⁓ So the other example, you mentioned it when we were kind of getting ready for this. It's a more concrete one. ⁓ But it's the CMS-less website. That's been one we've dubbed it AI CMS, ⁓ which is this idea that for years, ⁓ everybody has demanded a CMS. We've built content management systems. ⁓ content management backed websites for dozens of companies, hundreds of companies probably at this point. ⁓ And I can probably count on one hand, maybe two, ⁓ how many of them have ever logged in and made a change in their content management system. Most of them, ⁓ most of them unfortunately don't make any changes once they launch their new site. ⁓ Of the ones that do, usually those changes are running back through us anyways. And so for years, you've had to be the ⁓ sounding board for me. Like, why the fuck are we building this in a content management system when we have to make the change anyways? All it is is putting hurdles in front of us and making it more difficult for us to do our job. But it makes, it's a nice safety blanket. It makes people feel good. When we built... ⁓ sort of our new website, one of the things, ⁓ we ripped it out of the content management system it was in because it frankly pissed me off that day. ⁓ And ⁓ we had some updates to do and I was like, what if, is there a way to architect a system using, if I was building this today, is there a way to architect a system that allows me to just not have a content management system? ⁓ And so we did. Our website does not have a content management system. But people with zero knowledge of coding development can push updates to our website. And the way that they do that is they open a card in Fizzy and they describe, they say on this page, I need to change this to this and whatever it is. ⁓ That gets picked up by an AI agent. It pulls down the site, ⁓ creates a new branch, executes that work as described.

deploys that to a preview URL and provides that back to the user. And then they can look at it and say, yep, this looks like what I wanted, or no, this does not look like what I wanted. And then we can ultimately say, hey, merge this, and it gets deployed. And that has created an interesting workflow that's allowed us to. to honestly make more changes in less time with higher degrees of accuracy and better previews and approval processes than we've ever had in any continental system we've ever built in. And then, obviously, if you need enterprise-grade ⁓ approval systems or... ⁓ other ⁓ things, this doesn't work, right? I'm not talking about those scenarios. But in a scenario where you just need a basic content management system or a content-based website, something you normally would have put in like a WordPress, you might not need that shit anymore. And I think that's a good example of like... Ultimately, we're putting the human and you know, the end result is like how do I make human lives easier? Right logging into WordPress and making a change is not a positive experience either even once you pile on like we have we've Added pieces to it to make interactive website builders and all kinds of other stuff to make it as good as possible but To a non-technical person who only does this once a month or once every quarter It's still not a fun interaction no matter how ⁓ nice you make it.

One of the cool things, ⁓ just picking on that example, ⁓ is ⁓ it's not just making small changes to a website. ⁓ This is, in some cases, making completely new components that match the theme of the existing website ⁓ in some way, shape, form that would normally take a designer getting involved and a developer getting involved to code it, and then ⁓ somebody to actually plop it on the website and ⁓ user test it. And ⁓ the fact that a non-technical person can use agentic an agentic CMS like this to say, please make a module that does this, does this, this is the copy I need, da, da, da, da, da. And you can go through that own QA process and then have somebody that has a technical brain actually look at it before we actually merge it and deploy it. ⁓ Like that is an ⁓ incredibly different workflow than what normal content management looks like ⁓ on a website experience. And you take that and you apply that to virtually any other type of job. of how do I make the human's life easier? I think that's the right ⁓ way to think about it when you think about AI or human driven first and product design. ⁓ And I think it changes ⁓ with every function or discipline that you're looking at, but you take like, I don't know, let's take finance as an example. QuickBooks for the longest time has had this ability to match transactions, right? So you spent money at this place one time, you categorized it this way, I will henceforth categorize it this way forever more until you tell me otherwise. Simple little rule, but same idea. ⁓ The idea is to make the human's life easier. So how do you take that and then you now extrapolate it and you say, well, what if I didn't even have to make that interaction in the first place? If the AI can say, if it's from a place that looks like this, I'm going to auto categorize it this way. I'm making that decision on your behalf because you've trained me to do these different things. Now ⁓ I've removed that as another decision point. ⁓ And so really, think what we're talking about here is making the human's easier ⁓ is reducing the number of decisions that the human has to make ⁓ or enabling better decisions to be made with less effort.

Yeah, I mean, I think I could buy into that. And I think the question at the end of that becomes like, how do you do that? Normally with the answer for like our website, going back to our website example would have been, we'll make a more user friendly CMS. And in our case, we're like, well, what if we just throw the fucking thing away?

which was a risk, right? ⁓

And it worked. It was 100 % an experiment. And ⁓ still, think to this day, is an experiment to an extent. ⁓ But it was a different paradigm, right? We had to build. We had to build the site in a very component-driven way with ⁓ skills and guardrails to enforce brand standards and enforce component usage and add review workflows to ensure adherence to that. ⁓ Because as you described, anybody can come ask for anything now. So that becomes Pandora's box, right? And you can see it in many of the requests that come through where it'll just go reroll and create something from scratch that's off by one from something that already exists. Luckily, there's a review agent that exists that ⁓ prioritizes that and says, hey, we've already got a card component. You don't get to make a new card component. We're going to shove this in the existing card component. And if we need to make changes to it, it will be a global card component change, not a today I'm feeling like this should be slightly different. And that's how you maintain a cohesive feeling website and brand ⁓ is through these enforcement layers. But those enforcement layers were not necessarily things we would have to consider if this was in a content management system. You just wouldn't have that option.

Right.

So it does shift the logic by going this AI first, AI CMS route. It completely shifted the logic in where we spent time. ⁓ to try to make sure ⁓ that you can have good interactions. The results that you get are the results that you get, and that there's some sort of continuity and best practices always observed without having to tell it to do it. Because ⁓ largely, we want people to be able to interact with this that don't know about component-driven design. They don't know ⁓ what a component even is. They don't know how to identify if it used one or created a new one. And they don't really know what the brand best practices are. And you don't need to because it knows.

Well, in kind of taking that to the other examples that we've even talked about. So lots of little artifacts and tools that we've used to create things internally to make humans lives easier have all been based on those component oriented designs from the website. So had you not built that, we would have these artifacts that are looking like IBM and Disney and whatever else that, that ⁓ the AI decided to, pull as brand standards ⁓ in that timeframe. But because we built it. using components that AI can understand. And we've said, Hey, when we're, you're designing things on Oodle's behalf, this is a design language that you should be pulling from. ⁓ It, still does stuff that's off by one off by two kind of situation, but it's not a completely different look and feel. Yup. Yup. So it feels like us.

at least in the ballpark. And right, I think that's one of the things that we've done with like all of our interactions is every ⁓ AI system that we're working with has an understanding of our brand, our brand essence, brand voice. ⁓ color usage, know, a bunch of other shit that's baked in there. And that helps, it helps them, it helps to ensure that even if we say, hey, take a first pass at this or take this information that I have, I've written it all up. It just needs to be reformatted into something. Take this and put it in a format that what it spits out on the other side looks like us, feels like us and sounds like us consistently.

Yeah. ⁓ mean, the messaging that we continue to tout amongst our team, and I think we've said it on ⁓ this podcast at least five times so far that I'll kind of reiterate, ⁓ is everything that AI is doing should be in pursuit of making humans' easier. ⁓ And ⁓ in many cases, it should be about reducing the amount of busy work, mundane work, stuff that doesn't require thought ⁓ or... you know, is ⁓ over and over again type of activity, right? Like those are the things that you can eliminate first from ⁓ just, you know, the things that nobody likes to do anyway. ⁓ And if you can have ⁓ AI agents do all the things that nobody likes to do anyway, it enables humans to do ⁓ a lot more thought-provoking work ⁓ and do higher quality, higher impact work.

I think that's the...

That's sort of the trouble and like disconnect I think exists today, right? You have a lot of people, you have AI that's capable of doing a lot of things. You have a lot of people really fearful that AI is going to replace them. ⁓ And you have a lot of executives and people who don't understand this shit claiming they haven't figured out and making some sweeping changes ⁓ because of that, right? ⁓ the one that immediately pops to mind is like ⁓ Amazon has done ⁓ an enormous amount of layoffs claiming AI efficiencies and all these sorts of things. And then it has had the highest levels of problems to the extent that they had to have a mandatory developer meeting with their engineers to say like, hey, this AI stuff is breaking shit and costing us money. ⁓ So maybe you didn't have it quite figured out. And I think our path and how we've looked at it and how you and I have talked about it has never been in the lens of, know, okay, how can we use AI to replace people? It has always been how can we use AI to give our people superpowers effectively and remove all of this bullshit that nobody likes anyways. ⁓ Which is kind of funny because sometimes people cling to it, right? When you're like, hey, you don't need to do this anymore. There's a little bit of like an identity crisis moment that happens. But you you just have to go on that journey and be like, do you enjoy doing this monotonous activity that you've complained about for the past five years and I'm telling you, you don't need to do anymore? No? Okay. What could we spend that time on instead? Because now Sheldon's got that.

Well, it goes back to the principle that I ⁓ think will become more and more true. It has become more and more true since we even first introduced this idea, you know, ⁓ 90 days ago or first episode of the podcast, whenever that was, ⁓ which is people, people ⁓ won't be replaced by AI. They'll be replaced by people that know how to use AI, which means that if you are a person that is a creator and not a box checker, you are totally safe, but you have to be a thinker. was talking with one of our team members about this the other day. I actually think that ⁓ we have evolved as a society towards deep ⁓ specialization in a lot of cases. ⁓ And what AI in the way that we're creating these things, not to say that specialization is not important and that it's not going to be meaningful, but I think the role of the generalist, people like me that know a little bit about a lot, but I'm not highly specialized in any one particular thing, ⁓ is actually incredibly empowering. because it allows people like me to be able to do things that I would never have been able to do before and to conceptualize things that people like you can go implement because you do have the specialized skills. Whereas those things would never have happened before because they never would have gotten off the ground because the effort was too high. And so going back to Jevons paradox, it's this idea that people that ⁓ can be empowered to think about and create new things that would have never been thought of before. And when you can do that, and empower yourself to be a creator, it doesn't matter whether you can take it across the finish line. The idea is ⁓ arguably the more important part. Putting together what should the thing be, what problems does it solve, why does it solve those problems, and can you create a quick prototype just so that we can understand what the experience should be. And if the answer to those things are yes, then you can say, is this a high enough priority for us to invest our energy and resources into actually making it real? ⁓ Or, just like any other brief, This is just it's just an interactive brief, right? ⁓ You could just throw it away. It's not important enough. ⁓ But the fact that you can take it so much further so much faster and it's not just words on a page, it becomes something that you can say, huh, I can see the value in this or not very quickly.

Yeah, I think that's an interesting one to think about and one that we've talked about ⁓ also. I ⁓ love the fact that people who ⁓ are non-technical can now create technical concepts or artifacts that illustrate things as a working prototype or interactive wireframe. ⁓ And again, that's another area where people get a little weird, right? They're like, well, you know, why do we even need developers anymore? Why do we even need designers anymore? ⁓ If you've seen the results of any of those or you've tried to see them, it's really clear. ⁓ They ⁓ look kind of fucked up and they don't really work right. But the ideas are there and the concepts can be fleshed out. And the nice thing is, ⁓ usually this is stuff that you wouldn't figure out until you've ⁓ written the requirements, you've done the designs, you've implemented, you're eight months down the road and a million dollars spent. And then you're like, man, this concept just doesn't work. This interaction doesn't feel right. We need to retool this some. And now you're pulling designers back in, you're pulling developers back in, you're retooling stuff. And instead, it gives the ability for a consultant or ⁓ an account person, a project manager, a ⁓ designer, anybody. to sort of build out those interactive workflows and interact with them and see ⁓ how it works and how it feels. This is no different than like Figma back before Figma was the thing or InDesign, or not InDesign, what was that? The one we used before then. Doesn't matter. ⁓ But the idea of like, once upon a time we had wireframes, then you could click on the wireframes and go to different pages of wireframes. could kind of simulate.

these interaction workflows, ⁓ then they got smarter and then they got smarter and now we can just build whole ass prototypes. ⁓ And I don't think that puts anybody's, it doesn't really put long-term anybody's job at Jeopardy. I think there will always be this short-term contraction and this short-term rush to judgment. that ultimately will observe the same paradox that we've seen play out time and time again, where the demand will increase disproportionate to ⁓ the current anticipation.

⁓ One of the examples of Jevons paradox that I think is an interesting one and kind of applies to this ⁓ is ⁓ Is the internet so newspapers used to account for hundreds of thousands of jobs across the country and you had some proportion of those people that were journalists ⁓ and the journal like journalists by and large that job for newspapers has shrunk substantially tens of thousands of journalists no longer have jobs, however The internet allows anyone to become a journalist. ⁓ so content creation teams that used to be teams of hundreds are now teams of less than 10 making content engine or content production for their creator studios. And they're making great livings doing it. They're having a lot of fun and the amount of content that has been created has fundamentally exploded. And so you have Jevons paradox. have one industry that shrank, but you had a new industry that spun up right beside it that largely is doing the same stuff. some entertainment value, some newsworthy, whatever else. you know, the same is true of Hollywood, right? So newspapers and Hollywood are both great examples.

⁓ I'm thinking movies and ⁓ movies, TV ⁓ and music all used to be controlled by these massive organizations. The only way anything got produced was it had to go through one of them. And then you, you now create this exposure where unfortunately, SoundCloud rappers can exist. ⁓

Yep. Yep. ⁓

Correct.

But it's it's Jevons paradox, right? So something that shrank in that Hollywood productions have shrunk. The people that had jobs in Hollywood 100 % have have gotten smaller. But the number of people working in entertainment of some kind, which is Hollywood adjacent, has exploded with the creator economy. And so I think with AI, we're going to have something very similar. We don't know exactly what it's going to be yet. But I think AI is going to do largely the same thing. It'll it'll displace

All right.

some ⁓ workers in resourcing, but it will create 100 times more opportunity that we don't know how to define yet.

The interesting thing too is in all of those examples what it did is the I don't think anybody's losing any sleep I mean ⁓ there's a there are a few executives who are losing sleep over those organizations and the industry is being disrupted, but I don't think by and large the populace is losing any sleep over those being disrupted, right? Being able, what those did is they diversified everything, right? You used to have content creation consolidated within a dozen organizations ⁓ that were owned by some people and influenced and controlled and whatever. ⁓ And you went from that to ⁓ a complete democratization of content creation. because anybody can create content, anybody has an avenue to get content to people, ⁓ and as long as you have an audience, you're appealing, can, I mean, this is a good example, 50 years ago, pre-internet, this doesn't exist. Not one person ⁓ sees or hears our voice, like, they don't hear our voice, they don't see our face.

Mm-mm.

But because we have the ability to ⁓ purchase a camera and some lighting equipment and ⁓ take, spend some time, record something that, you know, at least some, number of people find interesting and publish it on the internet and disseminate it through distribution methods. We can do that. And we don't need, ⁓ you know, ⁓ a large organization to sign off on our idea or sign us to a record label or pick up our movie idea. We can just go fuck and do it. ⁓

Yup. ⁓ So how do we wrap those up?

I think that's ⁓ it.

Is it till next time? It's gonna do it. I'll end it there. ⁓

That's a good point to end on is just go fucking do things ⁓