---
title: "S01 E16 - The Authenticity Paradox: EU AI Watermarks vs. Human Flaws"
shortTitle: "The Authenticity Paradox: EU AI Watermarks vs. Human Flaws"
season: 1
episode: 16
slug: "the-authenticity-paradox-eu-ai-watermarks-vs-human-flaws"
publishedAt: "2026-08-21T13:00:00.000Z"
duration: "00:46:11"
canonicalUrl: "https://bloodsweatandtokens.xyz/episodes/the-authenticity-paradox-eu-ai-watermarks-vs-human-flaws/"
audioUrl: "https://api.riverside.com/hosting-analytics/media/5bb572c62ee73c8a6cfc861394904202725f475d7d3947a29ff049c966e66022/eyJlcGlzb2RlSWQiOiJhOTE1ZWI5MS1mNTEzLTQyMTItOTQxMS0zOWYzNzBmNGFiZDUiLCJwb2RjYXN0SWQiOiI3ZmE0NDhkNi0xMzhiLTRkOTktOTFhMi1lYzBkNGExMWI2MDYiLCJhY2NvdW50SWQiOiI2OWE1ZmJhOTRkZmVkYWY3NDViZTk1YTMiLCJwYXRoIjoibWVkaWEvY2xpcHMvNmE4NzJjMTNkNDRiNmJiMDc3NGRjNGY4L3RheWxvci1tYWNkb25hbGRzLXN0dWRpby00VE9qRS1jb21wb3Nlci0yMDI2LTgtMjBfXzE4LTMyLTE5Lm1wMyJ9.mp3"
youtubeUrl: "https://www.youtube.com/watch?v=vcPdFWKK1m0"
---

# The Authenticity Paradox: EU AI Watermarks vs. Human Flaws

## Summary

The EU AI Act just went into enforcement with mandatory watermarking for AI-generated content, but Taylor and Sean question whether this regulatory approach actually helps anyone.

## Show Notes

The EU AI Act just went into enforcement with mandatory watermarking for AI-generated content, but Taylor and Sean question whether this regulatory approach actually helps anyone. They explore why audiences might prefer human mistakes over algorithmic perfection, debate whether education needs complete overhaul when AI can solve most problems, and discuss their major pivot away from building another CMS toward letting clients "vibe code" websites with natural language.

From Taylor's confession about hacking college quizzes to Sean's AI-powered gardening success, they examine authenticity versus transparency in the age of synthetic content. Plus: Taylor's "Bread and Butter Connects" sandwich project that's revolutionizing team culture at Ample.

**Show Notes:​**
• Social media authenticity struggles and content creation challenges
• Consistency vs. quality debate in AI-powered creation
• EU AI Act watermarking: consumer protection or creative friction?
• Students gaming AI detection systems in schools
• Rethinking education when knowledge becomes commoditized
• CMS project pivot: embracing simplicity over complexity
• Natural language website creation replacing traditional CMSs
• Team building through Cincinnati sandwich exploration

**Resources:​** EU AI Act documentation, Anthropic watermarking announcement, [breadandbutter.wtf](http://breadandbutter.wtf)

## Transcript

Taylor MacDonald: Hey, hey Sean, welcome. Welcome back, Mr. Davis, to episode sixteen of the Blood Sweat and Tokens. How you doing, man? Where you been for the last couple of weeks?

Sean C Davis: I've been I've been everywhere and it feels like I've been everywhere, but I haven't left the Midwest. In fact, after it was two and a half weeks and we were like, We've only been in three states, but I'm tired. I need a vacation for my vacation.

Taylor MacDonald: Yeah, from that happens all the time. Are you the type of person that schedules an extra day at the end of your vacation for just to recoup? Or or do you like drive home on a Sunday night and you're in the boardroom at nine AM the next morning?

Sean C Davis: Yes, yeah. Al almost always I learned this a couple of years ago. You add that extra day, it's it's invaluable. Do the unpacking of both the clothes and the brain and it's it's great.

Taylor MacDonald: Yeah, I completely agree. I factor that into all vacations. Highly recommended. You gotta have a day just to like, you know, get back in the routine, do some laundry, take a nap. It's the best. Anyway.

Sean C Davis: Hundred percent. Yeah, yeah.

Taylor MacDonald: Well, so we haven't seen you in a minute. We've had a a couple last couple of weeks have been interesting, we'll say. So I met two weeks ago, met with my business partner. We interviewed Kevin Comer, talked a little bit about strategy, the death of Creative Services and and so forth. And then last week you and I did what I think was a cool format, but turned out to be one of the most painful post-production efforts of my life because I'm inept and I'm old. but I eventually figured it out. Anyway, that was an interesting format. Have you done that? Sorry, the format itself was that Sean sent me a pre-canned reel of questions. And it was hilarious because in between each question, he would just sit there, like stare at the at the camera, which I thought was so funny. But it took me forever to like figure out my workflow and how I should be responding to you and make it natural sounding. So if it sounded a little unnatural, my bad. have you done that before? Has that worked out for you and other you've you've you've been the host of several podcasts, have you not?

Sean C Davis: Yes. Yeah, I actually r I really like it. I usually I my failure was I didn't specify my my approach ahead of time. Cause yeah, usually I've been had this thing I was doing for a while that was worked really well and I was like, here are three questions and I send them to people and I'm like, Yeah, just record yourself and send them back to me and then would cut them up and and post them, which I thought was really cool 'cause this feels like You know, for some of these shows it can be hard to get someone nailed down for thirty or sixty minutes, but if you're like, Hey, can you just record ninety seconds of you talking into your phone? It's great. It's great.

Taylor MacDonald: Yeah, I feel that. this whole world of social media and putting myself out there is really challenging a lot about myself. I don't I d I never realized I don't think I'm an introvert necessarily, but I'm I'm kinda shy a little bit. So the experience of making TikToks or talking to myself or I've got this so on LinkedIn the team at Ample is helping me like learn how to be a little bit more social and put some of the stuff out there a little bit more to drive some brand awareness about the company and about the podcast and things like that. And they keep I don't know, they keep like like the they'll draft a post for me. And then I go in there to to you know hit submit and it just feels so canned and so

Sean C Davis: Yeah.

Taylor MacDonald: yucky. You know, it's like I don't know. So I'm struggling to figure out that that momentum. Let me ask you this. As a social file such as yourself, do you are you constantly seeking out ways to generate new content and kind of get your ideas out there? You've been blogging for years. Is that more of just a therapeutic exercise for you? How do you muster with everything you got going on, you got kids, you get the work, you got, you know, the house to deal with, all that stuff, how do you muster the space and time to make this a priority?

Sean C Davis: Yeah, and that's that has changed dramatically. It's a it's an ebb and flow kind of thing for both the blog, the videos, social media. I feel like the one thing that people say is really important and that I haven't done is been consistent. Like, can you be consistent in this one space? That's how you'll build that following in that space. And so that's what I'm I'm really trying to do now. I I feel like like the other day. I've gotten old enough, I need to I need to be able to focus and to also admit what I'm good at and what I'm not good at. And until you you call me call me what social social file. I hate social media. I hate social media. I'm I've n it's always felt like a thing I should do and and and it's been a forcing function and and you know when schedules get packed the thing that you're forcing is the easiest and first thing to go. On the other hand, I really like recording videos and I don't know why. It's just a it's a thing I've enjoyed. And so lately I've been like, okay, here's what I'm gonna do. First, I'm gonna get consistent on putting videos out there strictly on YouTube and as frequently as I can. And and the way I do that, by the way, is just like sitting down and having a series of topics and then just recording for sixty to ninety minutes and then they end up being four or five videos or something like that. And then I think this I you know, social media for me has always been kind of a a megaphone more than it's been a way for me it's not how I like to connect with people. I prefer to connect with people either in person or direct messages or and that sort of stuff. And so it's become it's more of a megaphone. And I think that is becoming easier to automate. But I was like, I'm gonna I'm gonna solve this the the consistency thing first and then we'll come back to the how can we automate some of these posts but still make it feel like me.

Taylor MacDonald: Is consistency more important than quality?

Sean C Davis: This is a that is a fascinating question. And I feel like that is a that is a ripe conversation in the age of AI too. Because I was I was just talking about this yesterday. Because with what we're seeing today with engineering is the answer seems to be yes. Like well, at least not consistency, but speed. Right? We're like, put it out there as fast as you can, build as fast as you can. It's okay if it's a little bit broken, and we're we're having to shed some of these discomforts we've had in the past of like, this isn't as polished. I can't release it. And there seems to be more of more tolerance for bugs for whatever reason today, when it you you would think it would be the opposite. And I see content kind of inheriting the same sort of idea with that, like, yes, it's more important that you put stuff out there than it is that it's perfect.

Taylor MacDonald: So, okay, we've been having this debate in our house for a long time about, particularly in the in the vein of politics, you know, like how does somebody like Donald Trump rise to the position that he had? And Crimson's had this theory for a long time that it is authenticity. Not that he's the most authentic person, but he has presented himself that way, and that human beings naturally gravitate towards authenticity, which is all the more important today when you have synthetic content popping up everywhere. And what you you said something You know, it's people are more tolerant of things being broken or making flubs or whatever. That's one thing that I keep trying to remind myself. You know, I'm a kind of a perfectionist. And when I first started, I did it was actually for Netlify back in twenty eighteen, I think. Twenty eighteen, twenty nineteen, whatever that Jamstack conference was that was remote. it was remote. So I guess it was twenty twenty 'cause it was during the Yeah, yeah. So

Sean C Davis: Yes, I remember that talk. Yep, yep.

Taylor MacDonald: what I did was I recorded you know, like a reel that I sent in and that was my presentation. For the conference. And at that moment in the conference, because it was all remote, they just put it up on screen, they shared the screen, and it was me. And I was reading off a script. And it was the most horribly awkward thing in the world. And I remember going through a thousand iterations of this script, just trying to get the lighting right and the position right. I messed up that word and all that stuff. And I think Crimson's on to something with regard to this authenticity argument. And I think it's being played out. And what you just described, it's like, You know, just creating a video, it doesn't have to be perfect. The flubs, in a way, those make you human. I wonder if those blemishes, if those things, those imperfections that are make us uniquely human are gonna be the currency that helps differentiate the content that we're creating, you know, today and putting out in social media versus the the perfectly manicured stuff. It'll be interesting to see if AI agents start introducing blemishes. into their output too to generally approximate hum humanity.

Sean C Davis: Yes. Okay. This is I would say my brain wants to say like ten different things. I'll say two things. One is that I've hosted a a number of virtual conferences, particularly in those COVID years. And so I was dealing with a lot of this seeing people present in real time, but then also pre-recorded videos. And it came to be I've I found it very easy. To determine whether somebody was reading or whether they were speaking, just speaking naturally or working from notes or something like that. And I have found that people in general respond way better to somebody stuttering or like, I need a minute. Let me take a drink of water and think about what I'm gonna say, or I lost my place. And people

Taylor MacDonald: Mm-hmm.

Sean C Davis: do not mind that, but the the reading comes off a little more canned. And I've thought about this recently. So thing number two is, and I've been w when now it's like you're reading something, you're like, is this AI slap or not? And on one hand, you say, okay, well, if there are no typos, it's it's probably an indication that an agent wrote it. And that was what teachers have used in these early days of AI to catch whether or not something was generated by AI. So I've got this buddy who is a dean of students at a high school. And they were catching cheating, meaning like AI agents were writing essays for these kids because they were looking for basically like perfect grammar, perfect spelling, because that's not how ninth graders write essays.

Taylor MacDonald: Hundred percent. Yeah.

Sean C Davis: And what the kids did is they learned how to prompt agents to say, Right at the level of a ninth grader. You should have a typo every X number of words or whatever. And it's fascinating

Taylor MacDonald: Man.

Sean C Davis: because it's kind of this that's how humans work too, right? This push and pull of like, how can I get it?

Taylor MacDonald: All right. This is okay. I have a confession to make and and also a I mean I appreciate that. I think that there's real if you have the ingenuity to work around the system that is designed to like enforce some sort of behavior, I you know, there's some creative thought there. And having kids in a in school, especially in the age of AI, right now. If my son could figure out how to do something by working the system as it exists today, as opposed to like following the str the strictures and parameters of what life used to be like, should he be penalized for that? You know, it we got a whole lot of interesting stuff to talk about today along these lines. But my confession is that back when I was in college, I had a I I was it was my senior year of college. I was almost done. I had to graduate. And I figured out how to reverse engineer the quiz on Blackboard.com. So we used Blackboard, which is like this remote thing. And it was a

Sean C Davis: Nice. Yeah.

Taylor MacDonald: a weird it was like in the a day and age, way before COVID, way before most of this technology. So it was brittle and it obviously was not very secure. I mean, it was an immature system. And so I forget it was a nutrition class. Not that I don't appreciate all that stuff, but it was so tedious and so mind-numbing. And I wasn't being a nutritionist. I was literally trying to fulfill this credit hour so I could graduate. And so what I ended up doing was just like I was building websites a lot at the time, and I viewed source on Blackboard, and I just dug through all the JavaScripts until I found basically a way to reverse engineer the answer for every question on the test. And okay. I'm not, I don't feel I'm not proud of this, but like so on the final exam, this is great. I feel like I can find I'm finally unburdening myself from this. This is fantastic.

Sean C Davis: Yeah. yes.

Taylor MacDonald: on the final exam, I totally cheated my way all the way through there. And at the very tail end of the exam, my power went out. And I got kicked out of the system. And when it all came back on and I tried to lock back in, it wouldn't let me because the system was already in session, but because I never completed. Test ultimately resulted in me failing. Now, I probably deserve to fail because I straight up cheated. I was super dishonest and I did the wrong thing. However, I feel that my ingenuity of trying to reverse engineer of figuring out how to reverse engineer this complicated system should afford me some grace in this situation. Well, what do you think? Fairly self-serving.

Sean C Davis: Yeah yeah, yeah. I mean it I I would guess that they I mean, I'm not an educator, so I don't know how they write these rules or whatever, but I you know, if if it's not written down that you can't do it, you can do it, right?

Taylor MacDonald: Well, the no, it's not even that. It's like the question is, knowledge is now a commodity. It's every it's like it you can

Sean C Davis: Right. Yes.

Taylor MacDonald: all you have to do is yell at something and like, here's a very accurate answer. I recognize that and a lot of the criticism of this technology is that we're degrading our interest or our ability to learn, retain information, think critically. These are all 100% accurate statements. However, why do we need to do that? Do you st Let me ask you this. If I was like, hey, I want you to do this factorial on these two fractions or whatever and come up with the binomial. I I'm making up words, I'm not a mathematician, obviously, but like, I want you to do this complicated math pro proof. Sean, if I put that in front of you, you'd be like, what? It there's no point. You don't need to know how to do the mathematical proof at this point in your career or at this stage of the game. Now there's value in learning how to learn. And and my problem with

Sean C Davis: Mm.

Taylor MacDonald: it is like kids could use this technology to cheat their ability to lean into the educational challenge. I mean, learning something, being coming educated on any topic requires diligence and effort and and responsibility and and curated time management and all these things. These are the things that you're learning. It doesn't matter what's in the book.

Sean C Davis: 100%. And I think so. Education it needs to be totally overhauled because of this. I mean, I because I would I would argue the opposite that I don't think we're degrading our ability to learn or to retain information. I think we're we're able to figure out what we need to and want to retain faster. And as a a case in point here, I Gardening and and caring for houseplants is one of my hobbies. And one of the things that I did this year was I set up this agent system where I could basically go around and would take photos and I would talk and record a voice memo of like what I was seeing and what I thought was happening. And then I'd feed it all into this agent and it would be like, you know, this is This is what you're doing wrong. This should be your next step. Here's how you fix this plant. Here's how you get some tomatoes to grow. And I failed a lot of the time, but I also created a pretty decent garden. I had I had people stop by the house for the first time ever and be like, I love your garden. And so now

Taylor MacDonald: That's awesome.

Sean C Davis: I want to fill in some of these spaces, right? And I'm like, all right, I don't I see plants I like. I don't know which plants I can I should add. And without AI, I'd be like, okay, I like these and I'd put them in the ground. And it might take me years to figure out like what's gonna work, what's not gonna work, why. Here I can just be like, here's my yard, here are some photos of gardens that I like, help me figure out what I can reasonably accomplish next year. And so I feel like then when I I guess my point is here, okay, so everything I learned this year about taking care of the plants that are in the garden. I don't want to take a photo and ask somebody what's going on. I want to just enjoy being in the garden and

Taylor MacDonald: No. Yeah.

Sean C Davis: so I'm learning faster than I wa would have otherwise.

Taylor MacDonald: That makes sense. well okay. I I'm I've got like 12 different thoughts in my head, so let me just like settle for

Sean C Davis: Ha ha ha.

Taylor MacDonald: a minute here. one thing I do want to talk, I want to spend some time talking about is the EU AI Act. So while you were out of town, I think, back earlier this month, the European Union adopted well, I guess the legislation itself actually passed Holland Fund. It passed back in 2024. But on August 2nd, the EU AI Act became like like hit a milestone. it is now applicable. EU national and like EU regulators have started to begin enforcement on this. New transparency rules require certain AI interactions and AI-generated content to be identifiable. Now, this is relevant to our prior conversation here about education. But it's it's curious. So one of the big updates is immediately following this act. and what it basically means is that if you are generating, well, here let me think about it like this. If if a machine is talking to me, I as a consumer should have a right to know that it's a machine. So if you're showing me a deep fake, you generally need to disclose that. And if you're building the foundational models that are powering all of this, regulators can now demand documentation on what was built, how it was built. How you're managing risks, that kind of thing. pretty quickly thereafter, Anthropic came out, made an announcement that they are going to sign on to this, adopt this, you know, regulation, obviously, for reasons of doing business in the in the European Union. But the more interesting aspect is they're going to be applying these same kind of watermarked content output globally. So that means that. There's going to be digital signatures attached to content, images, or other assets that you generate inside this tool that make it identifiably so. Now, as a matter of the human centric piece, I have a right to know you you you have an obligation to disclose that this is generated content. That totally makes sense. It's very similar to the GDPR law, right? Like I have a right to know what you're doing with my data, I have a right to know whether or not I'm being manipulated. I saw a Example of this related to like food safety standards, you know? It's like a nutrition label basically for content. It's basically telling you what went

Sean C Davis: Mm, mm.

Taylor MacDonald: into this so that you have a right to understand what how it's might be impacting your or influencing your receptivity of that content. however, as a business that's paying for the content that's being generated, is it appropriate for them to do this? And particularly is it appropriate in light of the Seeming hypocrisy. Now, this is what is going crazy on the social sphere, is that for the organization, a lot of these frontier models trained their data on publicly accessible stuff, right? They are now planning to take all of that corpus of knowledge to generate content for you, but put a watermark on it. You have paid for this content, it should be your content, right? So is that appropriate? Is it not appropriate in light of the risks for children all the way up to, you know, corporations? Where do you land on this? Have you have you heard about it much?

Sean C Davis: I've I've thought about it not in the context of the EU Act, but just in in general. You know, we've we I I talk about this a lot at Netlify as well. In fact, one thing that I'm working on that's you know really interesting, I think, is that we're we're building this pipeline so that we can make sure that all of our content is accurate and consistent across our properties, particularly those that the that other agents are going to use like the doc site and the skills and and llms dot text and it's really it's really you know it's like okay engineers a lot of engineers are trained to r release their feature with documentation and depending on the size of the company, you might have somebody who's handling the documentation. It might be on the engineer's responsibility. But now there's all of these other properties that need to consume and reflect some of those changes. So how do we make sure that all of that is consistent? Well, we can I I mean we're we're close to implementing this to be driven by agents. you know it's verified by humans But it's largely built by agents. And so I'm just trying to like

Taylor MacDonald: Well, hold on. That's that right in and of itself is interesting. Like, do you have an obligation to disclose that this is synthetically I mean, you like, okay, so the agent let's say the agent you're using generates content and it has a digital watermark. You guys are gonna copy and paste that onto social media, right? Are you are you

Sean C Davis: In theory, yeah. Yeah.

Taylor MacDonald: should you be obligated under this law to notify the public that this is content generated by an agent?

Sean C Davis: Right, that's I

Taylor MacDonald: And and then think back to the prior conversation we just had about authenticity is what sells. This is what you know, the authentic human, somebody who is flawed in a way, you know, that makes the mistakes, has the typos, the errors of statement. That you told me earlier that your assumption was that people actually are quite endearing to some of those foibles. And that it's better to be a human than it is to read off a cue card and get it exactly perfect.

Sean C Davis: I agree. I think it's just a matter of It's it's probably a spectrum and there's a line somewhere. And what I don't feel like I understand is where is that line for me and where where should it be for most of us in the sense that if I Okay, let me give you an example. I just recorded a video and the re video was me talking to the computer, and then we transcribed that audio, which is my like literal words that were pulled. From the video. And then I tell an agent to create a a blog post or blue sky post, Twitter post, whatever, from that, and it creates it and it knows my voice. And so it creates it in my voice. I look at it and I say, hmm, yeah, it kind of sounds like me. And then I post it. And so, okay, I didn't generate that, but it was generated from my work. Like, is wh which side is that on?

Taylor MacDonald: You. It was your originally your intellectual property.

Sean C Davis: Right, exactly. But is that should

Taylor MacDonald: Great question.

Sean C Davis: that get the watermark? I don't know.

Taylor MacDonald: How about this? every single printed ad has been retouched in Photoshop. Every single one. Literally every

Sean C Davis: Right. Yes.

Taylor MacDonald: raster image you've ever seen in every magazine has had somebody going through removing pimples and moles from the model's face. Should that be disclosed as computer manipulated?

Sean C Davis: I mean that would be weird, right?

Taylor MacDonald: It it's ridiculous. I completely. I just I don't know. You know, at the end of the day, I think regulation is extremely important and there has to be a healthy balance. And I I actually applaud what the EU is doing because the irony is, I was as I was prepping for this conversation, I was asking about like, you know, what is the impact worldwide for this legislation? Like so GDPR is the corollary. GDPR is the global privacy act that was implemented by the EU for all I'm probably misstating this for what it's worth. So my general understanding is that it it applies to all you know citizens of the European Union and that if you are accessing a website as a sovereign citizen of the EU, this act applies to the website, to the controller of the data. Meaning if you come to my website and you're You know, sitting in I don't even know, like Belgium or something, you access my website and I do analytics or prospecting or any sort of like marketing analysis or intelligent like business intelligence stuff based on your IP address or your, you know, digital footprint or whatever, that that actually puts me in violation out of the gate of this privacy act. You as the user, it there's a bunch of different things, but you as the user should have the right to opt in. To me using your data for these purposes or retaining your data, you have the right to opt out. You have the right to basically call the company and say, I want you to delete everything you know about me. This applies theoretically, you know, their the jurisdiction lives in the EU, but like when this happened, there was a bunch of questions. This is like three or four years now, but like when this happened, all these websites in the United States are like, wait a minute, what? Like I we're not operating over there, so it's not that big a Deal, but the truth is they were still theoretically governed by this legislation. Now, whether or not somebody would

Sean C Davis: Mm.

Taylor MacDonald: have standing to bring suit, that's a whole different question. I'm not sure if that's been adjudicated yet or not. But what has happened since is that a lot of independent states in the United States have started adopting their own similar privacy laws. California was arguably the most notable. They created an act called CIPA, which is like the California Invasion of Privacy Act. I think there's several others. That kind of there's a wiretapping act from back in the 60s that is being invoked for certain legislation along these lines. But again, the problem is there's all the it's so fragmented, right? Like we live in a in a conglomerate of US states. And if you're doing business in the United States, you've got apparently under this circumstance 50 different 50 plus different regulations to worry about and like the gradation. So it's a massive burden on business owners. But the bigger question in my mind, getting back to the To the AI Act was how's this going to impact the rest of the world? And the argument was that regulators will follow suit, right? So the regulators in the United States will eventually adopt similar conventions. I hope that's the case, at least to some degree. But

Sean C Davis: Mm-hmm, mm-hmm.

Taylor MacDonald: it would be nice if it could happen at the federal level, because then you could reduce that granular, like kind of compliance burden on the small business owners. However, I have very little faith that's gonna happen. And I'm not necessarily in favor of big federal like mandates either. It's just it probably does need to happen, but how does it happen and and how do we roll it out? How do we comply? It's a complete disaster.

Sean C Davis: Yes. And so in the with GDPR though, it's I think it's pretty obvious that the if I say what are what are we protecting against there, it's individual privacy. And if do do you have an understanding of what we're protecting against this EU Act?

Taylor MacDonald: Well, again, I mean it's like if you go to the co-op and you buy some granola and that granola turns out to not be organic. Or or what you what I mean, or like like the label says something or there's a an air of of manipulation on behalf of the producer, I think is the is the argument, right? I mean, in in the you know, most mild outcome, okay, well you didn't eat organic oats, like your crunchy ass is now You know, contributing to the downfall of society. On the worst case, you know, you you buy the product thinking it doesn't have dairy in it, and you have some horrible anaphylactic shock and end up in the hospital. So there's real implications to being able to understand what you're working with and like what the services are providing, where it came from. I think that argument, like if this is a deceptive tool or something, that that seems reasonable to me to To have a label that tells you it's like cigarettes, you know? Yeah, you can smoke cigarettes. If you read the box, it's not good. I mean, the federal government is like, yo, do you

Sean C Davis: It's not good for

Taylor MacDonald: should not be doing this? Of course, it took, you know, how many years and how much legislation to get that stupid label on the box, and nobody reads it any fucking way. But I am curious.

Sean C Davis: Yeah, but I I think I think what I'm saying is so in those two both cigarettes and the the organic oats, those are protecting against health risks or health implications.

Taylor MacDonald: Maybe. I'm not sure the organic is more of a crunchy environmentally forward type thing, isn't it? I maybe

Sean C Davis: yes. But I

Taylor MacDonald: not. I don't know.

Sean C Davis: well, we could say maybe the ingredients list, for example. i or, you know, you have to say when something is made in a facility that also makes peanuts or something like that.

Taylor MacDonald: Okay, how about this? What about have you ever looked at a label that has like a really rosy picture of the calories or something? But it turns out that that's like four servings of the thing. Like, like there's there's all this constant manipulation.

Sean C Davis: Yeah, right. Yeah. Yep.

Taylor MacDonald: I remember when Michelle Obama tried to roll out like these updated nutrition guidelines. It was really smart. It was intended to like equalize a lot of this stuff so that like you're comparing apples to apples. But this is the whole thing like a nutrition label, great. Like what is it actually telling me? I have to do math to understand what the impact of that is? Like give

Sean C Davis: Yeah, right. Yes. Yeah.

Taylor MacDonald: me a break. That's so stupid. That completely defeats the purpose. You're putting more burden back on the consumer simply to satisfy the regular the regulators. What how is this going to be any different necessarily?

Sean C Davis: Right. And what I'm I think I'm trying to wrap my head around okay, if I see an advertisement that was generated with AI, I might feel differently about it. But if it's if we get to a point where it's hard to tell, I don't why do you know, you can animate an advertisement today or you can put you can have a live action advertisement in a video. I don't know that I don't know that I care. No, I don't I don't think it matters.

Taylor MacDonald: Does it does it matter? Does it matter? I you can watch a movie and think it's great, and then I watch the same movie, and I'm like, that sucks. I don't like that. Like we all have personal preferences and taste. What difference does it make? Really? I'm I mean the mask is off. Like people can create whatever the hell they they're already doing it. I mean, the irony is, here, let me show you. This is the funniest. I haven't laughed this hard in a minute. if I can find it. There

Sean C Davis: While you're looking for that, did you see the the guy who created his his interpretation of the Odyssey? And instead of doing it for like hundreds of millions, he did it for I think I think it was like twelve hundred dollars or something like that.

Taylor MacDonald: No, I love it. I think it's great. It's not terribly dissip dissimilar to like when Deep Seek released their R1 model and it was built at like, you know, what it was like ten million. OpenAIs over here spend it hundreds of millions to do the same thing. It was just like

Sean C Davis: yeah. Yeah.

Taylor MacDonald: not not a good look. All right, hold on, let me show you. So if you're on the if you're listening to the audio of this podcast you're gonna see this image, but this is this is a picture of Dario Amodei. on Twitter that somebody obviously used AI to generate a picture of Dario holding a a a glass of white wine, wearing a very fluffy turtleneck that has like twelve layers

Sean C Davis: Ha ha ha.

Taylor MacDonald: wrapped around his neck. It's so funny. And the caption says how it feels to steal content from the internet and put your own watermark And we're back. Sorry about that.

Sean C Davis: We're back. Okay.

Taylor MacDonald: My bad. I tried to hit the stop share button and I hit the stop record button. I'm struggling these days. Anyway.

Sean C Davis: it's fine. It's fine. So okay. I but that that is an interesting point. Like for for the case of, you know, I don't know, defamation or deception or things like that, I think it it makes a huge difference. that so that's interesting because it's like this it seems like it's this blanket act. But it's only gonna matter in a few cases. And it makes me my brain goes right to how how anthropic has changed the rules for Claude over time because it's really annoying to just hit approve, approve, approve, because one out of a hundred times that you know that permission gate gets tripped, you should actually care about that thing. Other times it's like, well, I'm just changing into this directory and and running this command or whatever. It's like, I don't know, what does it all become noise and then we don't care or not? You know, it's like but but I don't know. Should people be allowed to generate art of politicians and people like that? They've been doing that for a long time.

Taylor MacDonald: True. I mean, political cartoons. Like what I mean, what was the the First Amendment case with Larry Flint and the political cartoon back in the 70s or early 80s or whatever that got him into a ton of trouble. He ended up getting sued, but it turned out to be this hallmark case in First Amendment, you know, freedom of speech rights. So I don't know. There's I here's the worst part. Well, never mind, I'm not gonna go there. I'll keep my mouth shut on that point.

Sean C Davis: Ha ha ha ha.

Taylor MacDonald: We'll save that for a different podcast. Anyway, this is great. This is a fascinating conversation. I did want to kind of circle back. I know today we're a little bit out of our normal format. Sean, could you shift gears a little bit and maybe talk to me just a tad about the state of our current product, where your head is on that, and where you think that might be heading over the next, you know, couple of weeks or so.

Sean C Davis: For sure, yeah. So I I didn't make any progress while I was away. It was pure chaos for me all the time. but and Taylor and I have we've we've talked a little bit, but I had started thinking more about this and the problems that we were trying to solve here and thinking is okay, so let me let me back up a minute. Our our season theme for season one. Is delete your CMS. Everybody hates their CMS. And I don't think that's changing in any time soon. But we started drifting toward some new age version of a CMS. And what I think is potentially quite a bit more valuable is instead helping people solve specific problems. With the infrastructure they have or setting up new infrastructure or whatever that may be. And so we've been talking a little bit more about that. And now I so it feels to me like we're probably not building the right thing. And I want to make sure we're giving that its due diligence to make sure that it is actually not the right thing. but the the place we left it after episode 14, 15. Thirteen, thirteen was that we want to actually build in voice controls or speech to text within the application itself. And so I want to stand up a prototype with that feature. And while I'm working on that, I think we'll you and I should keep going back and forth and figuring out, hey, is this is this the right

Taylor MacDonald: Yeah.

Sean C Davis: direction or where are we gonna go?

Taylor MacDonald: So I've had an epiphany on this. I think I think you're right. I think w the direction so far has been wrong because it's largely just re-engineering the concept of a CMS with different controls and a different surface. And what came along with that was a fair amount of complexity to manage and keep track of as you were as you were vibing and all and all that kind of stuff. Here's my plan. I want to start selling clients on static websites that The agency or the client can vibe code the interior contents of the page. It follows a design system, it's locked to the same equity, but at that point, it literally we're just taking advantage of the tools, right? So we use the Netlify CMS. I'm sorry, the Netlify MCP and agent runners or some sort of product like that. And of course, you're it doesn't have to be Netlify. There's a million different providers I'm using that as the example because I'm looking at Sean's face and he works there. Also not a sponsor of the show for what it's worth. Not yet. Maybe one of these days. Maybe one

Sean C Davis: Not yet.

Taylor MacDonald: of these days. Anyway, so yeah, what if I mean you so most of our we have a lot of these brochure application or like brochureware sites, right? It is a Astro site, it's an X.js site. There's a CMS, they're expensive as hell. Like, and it's hard to iterate. You know, I want to do something different for this new campaign, and I can't figure out how to get that done in your brittle, you know, single purpose CMS. That has been a historical challenge with literally all CMSs for all of time. And

Sean C Davis: Mm.

Taylor MacDonald: and while some of these headless products have evolved to a more flexible and malleable state and have decoupled the front end from the back end, that has only added more complexity and more cost. So yeah, what if from here on out, like if you came to Ample and you're like, hey, I want you to build me a website for my business, I'm like, cool, I'm gonna build you a header and a footer, and I'm gonna build the plumbing so that you can just tell it what you want. You open Cloud and you say, I need a new page, landing page does this, this, and this. A couple minutes later, it gives you a URL, you're looking at it. You don't even need a CMS at that point. as long as it's orchestrated in such a way that, like, if I want to update globally, that that's efficient. You know, it's using proper includes, there's not a duplication of business logic down in there. There's some constraints around what the client can do. You know, you can't like, hey, I want to open this big secure gaping security risk. on my new landing page for this campaign next week, something like that. That would be bad. But at the end of the day, I think there's a world where the harness, the framing, the the structure, parameters and rails within which you know the client can work is simply as simple as saying, move that image to the right and drop this thing down. And there is no plumbing underneath. And we just rely on the non-determinism. We just lean into it. You know?

Sean C Davis: Yep, yeah. I mean it could be, you know, when when I was at Ample back in this was probably like twenty nineteen, twenty twenty, what I was a small child. What

Taylor MacDonald: You you were just a baby then. There's a baby mustache on that face back then.

Sean C Davis: we what we were trying to do was we were trying to build this starter kit where that would help projects move fast. And it's just a It's a perpetual, extremely hard challenge in an agency environment because you're just like you're moving so fast to build the next website that you just don't get the time to go back to the starter kit and it it fizzles or you know, it just doesn't get as powerful as it can. It it's almost like if we provided rather than more of this platform that we're talking about. Go back to that concept of here are the fundamentals of this thing. Here's the starter kit. Here's the first few components. Here's some skills that you can use with their agent. And then, you know, if we have maybe it's a couple of MPM packages or something like that that can get updated. So the skills can get updated. and then people can work wherever they want. They can work in cloud code, they could work in It could work in Slack or whatever. If we wrapped, say, Netlify or Cloudflare or something like that, then that could be that could be really interesting. It's lower level, but it maybe meets the people where they already are.

Taylor MacDonald: Yeah. So I've actually put this into practice already. and you're gonna love I don't think I've told you about this because this is a relatively new thing. Okay, I had this epiphany recently. Are you still a sandwich man? Back in the day, you and I used to have these long winded convers You're a sandwich man, you know? Like

Sean C Davis: Yeah, yeah, let's call me a sandwich I'm a sandwich man. Yeah.

Taylor MacDonald: what's your favorite sandwich? What's the best is a hot dog a sandwich?

Sean C Davis: w we don't have enough time for that.

Taylor MacDonald: Sorry, yeah, my bad. I'll let I retract. What's your favorite sandwich?

Sean C Davis: I I mean if we're going if we're going like gourmet sandwich, yeah.

Taylor MacDonald: You're on death row. You're on death row. And this is it, man. You got one last sandwich in you.

Sean C Davis: I'm gonna I'm gonna go Ruben where every piece was, you know, handcrafted.

Taylor MacDonald: Amazing. I love this. Okay. that's an excellent one. I actually almost asked you if a Ruben was your favorite. Okay, so I too like sandwiches. And it dawned on me recently. I do one on ones with my direct reports, of which I have two right now, which by the way, the work life balance on having only two direct reports. Woo! It's sweet. I encourage everyone to only have two direct reports, especially

Sean C Davis: Yeah.

Taylor MacDonald: direct reports isn't as cool and supportive and awesome and conscientious and professional as the two that I have. Very grateful. That said, so I meet Shay pretty frequently for one-on-ones on Wednesday afternoons. And so we were thinking, hey man, let's just start getting sandwiches. And instead of just making it a traditional one-on-one, let's start cataloging and storing the sandwiches every time we meet. Okay. And

Sean C Davis: Love that. Love that.

Taylor MacDonald: so then I put together a website, I vibecoded it, and I'm following this exact same model to keep it up to date. And it's awesome. So this is Bread and Butter Connects. The website is breadandbutter.wtf.

Sean C Davis: Yes.

Taylor MacDonald: I thought you'd like this. I was like, dude, Sean is gonna be so jealous. You're gonna have to come join us. So, anyway, so this is the website, and it's kind of cool. So, the idea here is that food is the connective tissue in all of our relationships. Now, an ample culture is super important to me personally, it's important to us. You know, as an organization. And so as you know, Sean, we spent a lot of time trying to build in humanity into the work, like trying to make

Sean C Davis: Mm-hmm.

Taylor MacDonald: it human, as broken and authentic as we possibly can. We're we're humans first. And we try to encourage that, right? And so what we ended up doing was thinking like, hey, well, let's just go around Cincinnati and we'll find different sandwich shops, we'll go experience the atmosphere, we'll share some conversation, we'll break bread. And it brings us closer together. And I'm firmly in the camp that this is a powerful force. We did our third instance of this yesterday. I I have a pull request that I'd need to go review. but it was great. And we actually expanded. So we started with just me and Shay, but yesterday Shay was unavailable. So we ended up doing with some other folks, Kevin and Becky and Sarah. We all ate sandwiches and it was awesome. And I it led to some really great

Sean C Davis: That's awesome.

Taylor MacDonald: conversations, such as what is the sandwich you would order on your deathbed? so anyway, if you're interested in following along or if you have recommendations on sandwich shops you're particularly passionate about in the tri state Cincinnati area, please come to our website and drop that information for us so that we can go check it out and we'll post the review summarily.

Sean C Davis: I love that. This is great. It's my new favorite website.

Taylor MacDonald: I know, it's awesome. It's so fun. All right. Yeah, keep keep it bookmarked, buddy. Maybe you can join me for one of these sandwiches one day. I'd love that. Okay. Well, with that, I think we're we're getting close on time here. Sean, any last words of wisdom, any experiments that you picked up while you were out of the office or anything else that you'd like to share before we call today?

Sean C Davis: No, not not not really. I'll I'm happy to leave it there and I'm I'm getting back into the routine, so I'm sure I will have a lot to talk about next time.

Taylor MacDonald: Yeah, look forward to it. All right. Well, until next time.