The 47-Second Attention Span Crisis & The "Marketing Engineer" Myth | GTMN 21

Is human attention disappearing entirely, or are we just adapting to the AI era?

In Episode 21 of GTMN, Pranav and Austin dive deep into the rapidly shrinking human attention span and the "Ozempification" of hard work. They also break down Anthropic's massive launch of Claude Design, Salesforce's pivot to "Headless 360," and why Meta's highly-anticipated "One-Click CAPI" is just smoke and mirrors.

Transcript

SPEAKER_00 0:00

All right, April 18th. We are on our 21st episode, Austin. The podcast is of drinking age now. Welcome back to GTMN. It can vote and drink. It can vote and drink. That's right. Well, you had quite an exciting morning. It's been a busy week for both of us, I think. But what what happened? What happened with your bike? Tell me all about it.

SPEAKER_01 0:23

First of all, people who don't listen don't understand that me and Pranab are constantly trading exceptions between the Thursdays when we regularly schedule this to Saturday and Sunday. It seems like something always comes up on Thursday for one of us. And so we we tried to do Saturday today, which is normally fine. And as you know, you've been training for half iron, I've been training for, sorry, half marathon. I've been training for an Iron Man. And different levels. We're all here together, and I'm gonna be running with you in November.

SPEAKER_00 0:50

Yep.

SPEAKER_01 0:51

So I had this amazing bike ride planned, and I was 40 miles into the 60-mile bike, and uh there was this giant pothole, and my back tire went right in it, and uh I did not bring a tube with me. So I thought, well, first I tried to see if anybody had a tube, but the problem with these tires is or the frames, like the rims, are really, really thick. So you need a very specific type of tube to go in the rim, you know, because you have the tire, which is the you know, the thing that you think of on the outside of the tire. Then you have the tube, which goes in the tire to inflate it, unless you have tubeless tires, which is a separate thing. And then you have the the rim, kind of like a car, you know, you have like the rim, you have the tire. The only difference is the bike you have a tube in the middle. So I didn't have the right tube. And uh my buddy was riding with me, he went to actually a bike stop that was like uh a mile away and tried to get one and it wasn't the right size. So after like an hour, hour and a half of screwing around, I gave up and called a lift, and I sent you a photo from the lift of my wonderful bike in the back of the car. So it was just it was a longer than planned. Oh man. It was supposed to be like a four-hour afternoon. I was supposed to be back by 12, plenty of time for our 2 p.m. meeting, but then you know, now it is it's three o'clock, so I'm a couple hours behind.

SPEAKER_00 1:57

Well, you you got a nice little workout done, and sometimes, you know, it's nice to run into real real problems and figure things out, you know. So, well, I'm I'm glad you did not get hurt because um, you know.

SPEAKER_01 2:14

I can tell you how to change a tire, but I can't change a tire.

SPEAKER_00 2:18

All right, okay. So, I mean, it's actually interesting that we had to do this on Saturday because some some big news dropped yesterday, um, including Claude Design. Have you played around with it? I barely got like five minutes to look at it uh yesterday afternoon, but Figma stock is down 7% yesterday, and you know, kind of crazy. You know what was funny? I saw the founder of uh Deal uh make a post on X about hey Dario, can you please not uh do anything with payroll? Can you let us uh be? And it's because Claude has a job post out for people products and a and a PM or an engineering leader to build people products. So, you know, it's it's crazy how much we talk about Claude on this podcast, even though it's all about go to market, but like they are frankly just just kind of bulldozing through a whole bunch of uh verticals. So I have lots of thoughts on the implications for go to market for claude design, but I'm curious what you thought of it. Have you played around with it? Have you had a chance to kind of think think deeply about this?

SPEAKER_01 3:27

Started to like we're we're living in interesting times, right? The question is who is responsible for design and where does that happen? You know, recently in the marketing teams that I'm running and working with, there's a very common workflow, which is like design and other people work in Figma. And now there's this ability to pull Figma designs and recreate it through HTML or other sources because they kind of opened up a node-level understanding of a Figma design. So, like what's pretty cool is for teams that have been using Figma historically for a long time and are versed in Figma, the ability to go in and pull designs, really, really functional, great designs, and turn that into code is super easy. I think the example I might have given a couple weeks ago was I'm doing some fractional CMO work at A to B, and our team has taken like massive canvases of work for our partners that we work with in the transportation sector, and you know, Claude just turns that into HTML and publishes it to HubSpot. And I think like the same thing will happen with anthropic, you know, Claude design. The question is like, will designers switch? You know, is there a material switching cost? I imagine if they build like a uh a Figma to Claude converter that takes your entire asset catalog and converts it, then probably, yeah, the activation energy is so low that it doesn't matter. But right now, I I'm kind of of the mindset that it's like I I'm myself am not a designer, and now increasingly I'm designing with AI tools. So I'm actually kind of tool agnostic. Like a another example would be like I I often use this tool called ExcaliDraw to like do my diagrams for substat. Yeah, it's like, do I use ExcaliDraw? Do I use Claw Design? Do I use Figma? So yeah, I mean, probably not great for Figma because it's just like another competitor, and it's not very defensible to say, like, oh, you should use Figma because it's the place for designers to work. But guess what? You also AI is converting all this stuff in the back end to like code because that's ultimately what happens. So I don't know. I think I don't know what happens to them. I understand why their stock is down, but I'm curious to get your thoughts. So, like, there's a good post I saw about this, which is like as Claude creates more and more tools, at what point does the quality of these tools start to drop because there's just too much surface area? Or will AI actually help for the first time ever make an AI company that can manage 30 different surface areas because the administrative burden is actually like not human anymore? You know, Claude can manage itself. So for all the administrative upkeep of keeping tools alive and working and fixing bugs, all that can just be federated to AI. So I don't know. I mean, like what's to stop them from building into every vertical.

SPEAKER_00 5:53

I I th yeah. My take is the the latter, to be honest. And for for there's gonna be a surface level that's just developed in Claude that like gets you through 80, 90% of the work. And in some cases, you'll publish from Claude to Figma or to Canva, and they built those connectors, right? So they uh along with the launch that they announced yesterday, they do say that hey, you can take the output from Claude Design and push it into Figma or to Canva to take it to completion. And so, you know, that's I think what's gonna happen. It's kind of similar to their you know strategy with Excel and with slides, right? They they create, um, they they let you like work in Excel and then have like some superpowers within Excel, right? Same thing for docs, same thing for PPT, now same thing for Figma and Canva. So I do think that they're gonna go pretty horizontal on the types of things that most knowledge workers are doing and get you 80% of the way there. So it's interesting. And and you know what I'm gonna do for our company, so we have a design system for our you know customer-facing documents, right? So whether that's an existing customer, whether that's a sales or a prospect, whether that's uh, you know, office hours we're doing, like any type of stuff. And this makes it possible for me to actually just set up a design system for the whole company. And now I don't have to worry about what it looks like. I know that it's gonna be you know well designed. And we were kind of already doing it with a skill in Claude, right? And now they've made it a little bit more structured, and I think that's how people are gonna use it. So it is gonna have a meaningful impact because without this, you would have you know used Figma, Canva, what have you. And I think a lot of the you know basic stuff is just gonna happen directly in Claude. So going from an idea to an asset or you know, high fidelity like sales enablement, or you know, even like very basic marketing presentations, lots of lean execution is now possible for very, very small you know, marketing and sales teams.

SPEAKER_01 7:55

100%. And uh they also, I don't know if this was on your on your uh list, they also reduce uh release routines. Did you hear that?

SPEAKER_00 8:02

Yeah, I didn't I didn't fully understand routines and I know what was the news there? What tell me more about it?

SPEAKER_01 8:09

Yeah. We have a growing collection of I would what do we want to call them, like clawed surface areas or clawed attributes. And what I think is really interesting is that all the things that Claude has released help Claude be better and are these primitives that if you understand and use, you can like wield AI way more powerfully. Like, for example, in the very, very early days, there was this concept of context, and like people knew what context was, but like we didn't know how to manage context. And then came along the Claude MD file, and then came along memory files, and then came along skills. Skills are just markdown that has a predefined, you know, rule set or predefined thing to follow so that when you're repeating tasks, you don't have to type a bunch of stuff in there and it can reduce the context window. And then more recently we talked about remote control, which is the ability to like control Claude from Claude code, and here there's dispatch, which is the same thing for cowork. So I'll talk to our routines in a second, but like they're they're releasing all these structural ways of kind of thinking about AI, which are totally novel. Like none of these things like existed in AI three years ago. They've released them, they've taught they've kind of taught their users how to use them as a way of getting more value out of the tool. And routines is really interesting because if you think about it, you know, you and I come from the world of probabilistic and deterministic. And routines is actually in this model, right? So in attribution, deterministic is like you click the thing and like we know who you are. And probabilistic is we don't really know who you are. So we're doing some type of matching, or you know, we're trying to figure it out through IP address sort of thing. So, you know, in one case, you're very certain of the outcome. There's something that is physically being done. And in other case, there's something that kind of is being done, and you're taking some liberty in getting to that assessment. And AI follows the exact same mental model, which is I think part of the reason why we talk about this a lot on our show, is because the same kind of mental model for how we talk about attribution is the mental model you should use when thinking about AI. And AI is a probabilistic model. It is like, it is literally like a random walk, except it can be less random depending on the things you give it, and it can be more random depending on the things you don't give it. And routines and workflows and crons are are kind of representative of this, right? You can imagine if your computer every day you were to write in code, hey, run this script, this Python script that does something. That is a very deterministic outcome. It's gonna do the exact same thing every single time and it might fail, but presuming it works, it produces the exact same results every single time. So, like if you were like, hey, hit the Salesforce API and give me a list of the first hundred results, like that's a deterministic outcome. And you know, if you ask AI to do that task, it's gonna go out and write code or kind of hit the API in a different way every single time. So it might do it differently on each session of the AI. Skills make that better because skills can be combined with a script that can say, hey, when you run this skill, also run, you're just running the script that we've already built. So now it's a blend of deterministic and probabilistic. The probabilistic piece being the session calling the script. And so a routine is really fascinating because it diverges from what everybody else is doing right now, which is a cron. You know, people are managing workflows largely through crons, which is just like, you know, a time-based system that triggers something. A routine is a time-based Claude session. So it took the kind of understanding of a cron job with a deterministic outcome and applied the exact same thing to Claude. So now a routine can kind of wake up, it can call a session of Claude and perform a task at any given time and then write the results to any different place that you want. The thing it can't do yet very well is it you can't do call and response. So I can't like, hey, wake up and like run this routine and then wait for me to come back to you. It has to like produce a result.

SPEAKER_00 11:54

But where will you ask why why why is it different from a scheduled task and co co co-work and code?

SPEAKER_01 12:02

These are kind of all the same things, by the way. I I think you're it's a great point. Like, how is this different? How technically a cron that calls a Claude session is a routine. But I the point I'm trying to make is that I think Claude is trying to define ways of using AI in really structured terms to educate their audience on how to do things. Like we're all trying to figure out how to build together. Like a really relevant example is Claw, Clawbot, right? They had they invented this idea of like a soul MD and a memory file system to give agents life, right? So that they feel more realistic. These are just like ideas, right? And so what's interesting is like we're gonna see more ideas coming out of Claude that will make AI better and even easier to use. And this is one of them.

SPEAKER_00 12:47

Got it. I gotta go play with routines. And frankly, I feel so behind on open claw. I still haven't set it up. And we talked about perplexity computer last week. Like there's so much I need to play with. I I I keep thinking I need to run Claude, not on my machine, but on, you know, a system that's always running. And I I feel way behind, and and I'm definitely getting some FOMO from people who have it all set up and it's doing wonders for them. Okay. I want to pivot the conversation for a little bit to a to this topic that I've actually been thinking about for the last week or so. And, you know, there's we we we as an entire industry, we talk about cloud and automation and AI so much, so much, right? And it's changing so many things about sort of marketing and go-to-market and advertising. I get it. But then I got curious about, you know, how do people, how do human beings spend time, right? You went on like a bike ride for whatever, four hours this this morning, right? Did were you listening to music at that time, or you you prefer not to like partake in anything?

SPEAKER_01 13:50

You're just enjoying the world. I I do sometimes sometimes I do, but not today.

SPEAKER_00 13:56

Okay, okay, okay, cool, cool. So I I ran some research just, you know, uh over the over the the course of the morning today, and a couple of things stood out to me. So, first was the average attention that or or the duration that consumers focus on a single digital task has continued to come down over the last like two decades. So in 2004, the average focus on a specific task lasted for about 150 seconds. Now the average, you know, uh in 2012, the average was 75 seconds, and in like 2026, the average is closer to like 47 seconds. So that was one interesting finding that just like you and me sitting here and recording this podcast for 40 minutes, we're already way past the average, right? Like the average that a human being just does one thing on a digital screen is like 50 seconds, and then they switch to something else, and then they switch to something else, and then they switch to something else, which is like wild. All right, but then let's keep going. So consumers obviously prefer this bite-sized way of thinking about life, right? Whether it's work or whether it's personal. Short form content, so TikTok, reels, what have you, accounts for 90% of all internet traffic. All internet traffic, 90%. And then what's happened is media, so you know, if you think about music videos, you think about films, the average length of a shot in those formats is has shrunk to four seconds to maintain engagement. So this like every single thing in in in in a in a digital format has shifted to this like shorter and shorter time span, which is kind of wild, right? And you think about it, like our our show probably has more views on our shorts than on the full sort of episode, right? Which is which is what this is talking about. And at the same time, I'm gonna keep going and I'll I'll pause for reactions from you. There's research from the American Psychological Association that shows that frequent switching between these tasks increases stress, well, obviously, and reduces the overall productivity by about 40%. So, and I'm thinking about after after the segue here is like we're talking about all this AI stuff, right? And e even the AI stuff is like we run a cloud session and then we go do something else, and then we come back and then we go do something else, right? It's like it's we're constantly like setting off these AIs to do different things, and Slack and Linear and email, and then you know, meetings and Zoom. And I feel net net, while we're getting more productive, we're also just unconsciously, subconsciously getting less productive because of all this switching and this, you know, the the space to really just think and marinate and and you know increase the flow state that we all want to be in. And you know, it's just going away. So I'm just I'm gonna pause there, see see if you have any reactions, and I want to like talk about what this means also from a marketing standpoint.

SPEAKER_01 17:03

Well, I I don't want to make this political, but there was an amazing New York Times guest piece that I read this last weekend. And I'll give you the the headline so that people can find it, but I don't want to talk about the headline. The headline was why aren't the kids out protesting Trump? But the whole article was not about just that. It was about what is happening to our society that's leading to these conditions in which people are indifferent. And I think a huge part of it is the fact that you just described, the fact that kids are suddenly losing their attention spans and are cognitively declining. And there's a lot of research that that shows this, by the way. This isn't me just making stuff up. Kids and young adults and even ourselves are losing our capacity for deep work and cognitive, you know, cognitive capability because of social media and the infinite scroll. And there's like a lot more podcasts out there, there's a lot and more news out there on trying to stop the cycle and the habit of scrolling and short form media. But you're right, like our work is now becoming more and more like that. And so I think, I think in a way, there's this counterculture that might develop. I think it is developing around getting out of that mindset, getting out of that routine. We talked about this maybe a week or two ago. Substack is maybe becoming, in an interesting way, a place where people who are kind of protesting social media and protesting the mind rot that's coming with AI, where they go, like the power of writing is becoming a weapon to combat this, right? Like if you can't think for yourself, then you're not really human anymore. But, you know, like the tobacco industry before this, and now like social media, I think the distractiveness that we have is just another civilizational problem that humans are gonna have to encounter and deal with. And I think many people will not be able to and will be manipulated by corporations taking advantage of them. And a lot of people will figure out that these things are harmful and they'll figure out a beta better regulate it, getting the benefits of AI while not letting it take too much over you. For for what it's worth, like I feel this really personally right now, where it's like, on one hand, Claude is so exciting and AI is so powerful, and I just want to spend all my time doing stuff because it's so fun. It almost feels like a video game. At the same time, that's deeply scary. Because I know, for example, my my early 20s when I was like playing video games, I was like, wow, this is like a huge mind rot. Like I am on this thing all the time. I gotta regulate. So, you know, everything in moderation. And so it'll be this question of like what, you know, what will people do to regulate how much that takes from them? My my hope though is that in the next call it two years or so, the technology continues to improve where it doesn't require us to be as distracted. Because right now, I'd like we we've talked about this. I have like when I do claw work, you know, AI work, I have like six terminals up on each screen, and I am literally uh attention deficit maxing, right? It's like run one, run the next one, check the other one, go to the next one. And I'm doing so much work, but it's also at the end of the day, I feel deeply depleted and like, what have I done? Like my life is is ending from this. So I agree with everything you're saying. I think it's really complicated. We're in a really early time. So you know, it's almost like the same, the same thing's happening with Ozempic. Like everybody's everybody's on it, everybody's trying it, and we're all gonna learn together where this goes.

SPEAKER_00 20:32

No, it's a very interesting point, right? It's like instead of solving the root cause, we're kind of taking the easy way out with with GLPs, at least like the mass population, right?

SPEAKER_01 20:43

I'm not talking about people who have real sort of health issues, but it's uh, you know, there's an amazing article on Substack called The Isempication of Everything, and it talks about this, which is like all there are a lot of themes that are related here, like taking the shortcut in AI. I think society and America as a whole wanting to take shortcuts to wealth and generation because you know the prospects are hard and it's hard to find a job and it's hard to make money and prices have gone up, wages have decreased. And so the isempication as a theme for like kind of cheating your way out of fixing your body is the same thing for cheating your way out of making money the hard way, but that's because of the system and the constraints, right? Like wages are no longer enough to support a middle class lifestyle. And so people are looking for these make it rich kind of schemes. That's why VC is taking off. That's why Calci is such a big deal. So it's like it's fascinating that like this idea of the zempication of stuff actually pervades not only what we're talking about, but many, many layers of how society as a whole is operating right now.

SPEAKER_00 21:47

Yeah, it's interesting. You know, the the the other sort of two line to this conversation was if you now if you think about how do people spend their time digitally, right? So there's this, if you look at 2025 and 2026 data. On average, in a given week, video consumption, so across all devices, including connected TVs and mobile and everything, is about 43 hours, right? 43 hours of video consumption. And television out of that, which is both streaming or CTV, is about 20 hours. Smartphone video is like six hours. Tablet is like four hours. Social media is like 15 to 16 hours outside of like video. So roughly about two hours every day. Gaming is 12 hours. And then you have all the general browsing and tasks of like 35 plus hours. And if you think about all of these different things, right? Like how much of is how much of time is spent like actively, you know, with AI. And what I mean by that is like a direct interaction with a Chat GPT, a Claude, a Gemini, or whatever. On average, it's not more than about 30 minutes a day amongst all of this other time that is spent on entertainment broadly speaking, right? And the reason I'm bringing this up is we we talk a lot about how AI is going to change everything, right? It's AI is going to change shopping, AI is going to change productivity, AI is going to change like how you do discovery and so on and so forth. But like if I take a step back and think about do you think those 43 hours of video consumption, those 12 hours of gaming consumption, the 15 to 16 hours of social media consumption, like that's not going away because of AI, right? People are actually going to maybe spend more time escaping from the reality and spend more time on video and social and gaming and getting out there in the real world. And so, as marketers, the lesson for me is like, hey, don't just keep optimizing for AI and agents. You actually are going to realize that you have to optimize for video and gaming and social media and the real world more to be able to grab the attention because that's where the attention's going. And I think more and more human beings are going to sort of gravitate and escape from you know their general sort of work life to immerse themselves, even though it sounds very like negative and dystopian, but that's the reality that I'm I'm you know thinking about that may maybe the implications aren't as uh, you know, go go AI first and optimize for AEO and GEO and all that fun stuff. I I don't know that that's gonna be the real needle mover for marketers. I don't know. What do you think?

SPEAKER_01 24:23

Yeah, I mean, you know, as a marketer, I think AI will help marketers be more efficient. I think that I don't think video is going anywhere.

SPEAKER_00 24:33

Yep.

SPEAKER_01 24:34

And I think there's a relationship between stress, anxiety, and your human condition and how much you rely on these as a tool for coping with that hopelessness and sometimes that stress. And, you know, it's really sad, but especially in like lower socioeconomic circles, that's where like, you know, infinite scrolling is really, really harmful. And so it is actually biased towards the people who actually would benefit the most from not having to be on those platforms. And it kind of reinforces the cycle where you know you have some hopelessness or it's hard, life is challenging, and you're not able to get things you want, and you have a tough job or whatever. And then in order to kind of feel better about that, you rely on social media, which then makes you feel worse about yourself. And I think that's a big reason why we see huge depression rates in the United States. I know, and and and like, look, this this is kind of the thing that all marketers have known for a while, is like our job is to help optimize capital or optimize an outcome for our business. And oftentimes that means taking people's attention, and that is the name of the game. So, how like there, I don't know if there's a way to like ethically market right now. You know, all the time and money and energy that you spend trying to get somebody to pay attention to you is what causes this cycle. But I would argue, you know, a big fraction is not corporate media. It's what I'd call like the influencer maxing media, right? Like I'm I'm watching videos of cats doing flips, I'm walking AI videos of random, you know, BS. Like that serves no purpose. I mean, it's it's good, I guess, for TikTok and YouTube, because influencers or people are creating that and that content's being viewed and they're paying those creators, but it like provides no net value. There's no like there's no outcome out of it. I a great a great example is you know, there's this chef I follow on on YouTube, and she's probably made like 20 million dollars. And she like she was a chef before and she's a chef now. And have I like used any of her her recipes watching her videos? No, I just like you feel good watching them. Exactly.

SPEAKER_00 26:37

I know, yeah. Entertainment, like that's what we're all gonna like. We're really gonna become WALLY, WALL E style, bloated human beings on spaceships. Well, okay, sorry, that went a little dark. Go ahead.

SPEAKER_01 26:51

The thing that I think about a lot is like, okay, I don't know, 10 years ago, 15 years ago, when you're gonna have short form media, you're watching TV shows, sitcoms, tele like actual, you know, movies and podcasts, maybe, and all those could consume a lot of time. I think the big difference though is that because short form media is so quick and easy, it's filling in a lot of the gaps where like watching a show or watching TV wouldn't have been okay. But I don't know. My I growing up, my parents watched a ton of TV, like probably as much TV now as a human would watch on YouTube or TikTok, but we just weren't measuring it. So I also just wonder about how much of this is the fact that it's like short form and not necessarily about like the intensity and frequency of watching, but like the style of it because it's it's drawing people's attention away and like declining their cognitive abilities.

SPEAKER_00 27:43

I mean, think about it, right? I I don't know if if this happens to you or not. I was watching a Warriors game last night, and we lost, unfortunately, but uh didn't didn't make it to the real playoffs. And throughout the game, like I had this urge of like picking up my phone and like doing something on my phone, right? That that's the addiction that we all have to our devices, and that is the cost that you are exactly right that 20 years ago we were still spent maybe spending as much time on screens, but this like constant, you know, switching from this thing to this thing, and then on my iPad, and then my work, and then let me check my Slack, and then open you know, it's just like the all the it's kind of crazy that we have become this generation. Maybe it's just me, I don't know. And I'm I'm not even that like you know, I don't have TikTok, I don't do you know, reels on Instagram and what have you, but still it's it's quite interesting how attention is so fragmented. Well, on that front, a couple of other pieces of news that are happening this week. So apparently Meta is now going to overtake Google in ad revenue. So this has taken them what, like 20 years to get to this point, but they are forecasted to reach 240 milli billion in net worldwide ad revenue, and Google's forecasted to be like close to 240 at the end of this year. So all of the video, right, short form video from Instagram and all of the short form stuff from social media is finally taking over all the search stuff, which is kind of interesting. And I think YouTube is probably growing faster than even Meta in terms of like the overall growth rate. So at some point in the next like five or 10 years, maybe YouTube, you know, helps Google overtake Meta again. But that's uh sort of an interesting, you know, landmark moment for the advertising industry.

SPEAKER_01 29:33

Crazy. I mean, I I I don't give Meta enough credit, even though as a uh you know a marketer, I I know how much companies spend on them and I know their platform and I know they work, but it's still like, I don't know, isn't it in my mind when I think of meta, I think of Facebook. I'm like, it's all boomers on there now, but it's not. They got Instagram, they have so many surface areas, and so when I think of face, I just think of Facebook and I'm like, how are how is it so big? But it's not. It's all Instagram. It's all Instagram. Yeah, yeah. And but boomers do we've got a lot of Facebook, so they're they're propping up the old Facebook channel.

SPEAKER_00 30:07

Totally. And you know, they they launched something new this week, which I was doing my research this morning. So Meta launched this thing. You you know what conversion API is. So this is basically like uh, you know, server-to-server integration between your first party data and Meta's first party data. So you send signals to say that, hey, here are you know the email addresses and the names and the IP addresses of all the conversions that we've seen. So you can, you know, find more people like them and also do better attribution if I send you this CAPI data. They announced a one-click CAPI because CAPI is like still kind of complicated and technical to implement. And so they launched this one-click Cappy, and I was like scratching my head, how the heck would a one-click Cappy even work? Like you still have to send some data from your server site to Meta. And so I ran my research this morning, and it's all smoke and mirrors. The one-click is basically just taking your FBCL ID, so your click ID from Facebook from your browser session, and then just passing it back uh uh from Meta. So that's all it is. Uh there's nothing else to it. So it's just picking it up from the Pixel and sending it server side rather than you know just the the browser side. That's all that was. And I'm like, really? You launched this as a feature? You're calling it one click cappy? Like it's complete nonsense to call it cappy because it's not server side at all. So uh, and a lot of people on Meta on Twitter and on LinkedIn are talking about, oh, this is gonna be so amazing. And I'm like, no, it's not, it's gonna be pretty bad. So yeah, that was interesting.

SPEAKER_01 31:42

It's gonna be bad, but it's not gonna look bad because frankly, this is just a great way for Meta to take more credit for things that they don't deserve. I mean, the whole purpose of Cappy is to have a lot of signals, not only so that Facebook can get the right amount of credit, but so that you can optimize towards the right people. And so if if they're saying, hey, we have a single-click Cappy program with just the FB click ID, I don't know. I mean, they're basically saying if we have this FB click ID, we don't need any of the information from what happened from when they clicked to when they purchased, because we can kind of reverse engineer it on our side and figure it all out, which means like they're probably gonna use some type of browser app session to figure out the device and figure out the IP address and location, all that stuff, which, you know, is a technically reasonable feat. But I think the whole purpose of sending first-party data back to Meta was not only to make it better for them to find more people, but to be really, really clear when attribution is given. And I have now implemented, you know, Cappy programs for Meta, for Google, for TikTok, for Reddit. Um, so I understand all these really, really well. This is like some of the core of marketing technology. And I just I don't buy it. And and Metacapy is hard to implement for what it's worth, but I would I would argue like doing it the good old-fashioned way is way more reliable, and it's never been easier with tools and technology. Like this used to be super, super hard a couple of years ago. I I tell the story all the time. You know, me and McComer at Ramp, we had a data science scientist, we had two engineers, and like we worked the entire summer to set up this Metacapi program at Ramp. And then, you know, like it took a lot of work to maintain. There were always problems, and we were using DBT and high touch. We had segment to collect the data, we're modeling in DBT, and then we're fettering it out with high touch. There's still a lot of work to manage because all these things break, or like a model breaks, or some other instrumentation breaks. However, today, you can do pretty much all that. The only thing that stops you from completely using AI is mental switching costs, which we just talked about. Like so, with one team I'm working with, we're we're implementing Metacapi, and we have a data engineer who kind of can write DBT code, but can also write some front-end code. And even though he's not working on Metacapi exclusively, having him there is really, really helpful because he just understands the DBT surface layer way better than I do. So, like if I want to go fix something, I can make small changes to existing models and I can do that with Claude. But I'm not really qualified to like build our entire dim model, right? It's the same thing on web. Like we have a web engineer. I can write a PR if it's something small, but he should review it. So, so what we're seeing now with Metacapi is like you kind of still should have the same team players, but you really only need them for like a fractional amount compared to what they used to be. But anyways, I like this.

SPEAKER_00 34:34

I completely agree. On this news, I wish they had just called it like something, you know, like Pixel Plus instead of calling it Cappy. Like just don't dilute the importance of what Cappy is, and they have done pretty much that, which is kind of frustrating. So lots of people are gonna go implement it, fantastic, and they're gonna think that they're actually sending clear server-side signals, and they are not, and they're gonna give up on like the actual Cappy implementation. That's the downside of this launch, in my opinion. Yeah, okay, on that note, so OpenAI is finally launching Chat GPT's ad manager in a self-serve capacity. Um, and they've brought the pilot, the floor down from$250,000 of investment to$50,000. This was reported um on April 10th. And it's uh they're also launching CPC and CPA-based ads, so cost per click and cost per acquisition based ads, not just CPM ads. So all of this is you know probably gonna go down in the next like month or two. If you haven't signed up for this, go to chat GPT or openai.com slash advertisers, go sign up, get in there. There's only 600 advertisers who have uh gotten access up until now, apparently. So more, more coming soon. Yeah, okay. And then a couple of other pieces that switching a little bit more to like B2B and maybe B2C as well. So Salesforce got some attention this week where they have launched this thing called Headless 360. Yeah. And this is essentially uh Benioff saying, Hey, we've killed the entire like not killed it, but like they've created a version of this thing where you never have to interact with the GUI, the the graphical user interface, and the whole thing is just gonna be API'd, which is uh you know kind of leaning into the whole AI narrative in a in a pretty big way. Thoughts, reactions, given your very extensive CRM experience.

SPEAKER_01 36:30

I think like this was a long time coming. Uh one, I I think it makes a ton of sense. One of the biggest complaints about Salesforce, I think, is distilled down to how slow it is. No matter how much they've changed the UI, it's slow. It takes seconds for things to load and it's complicated. You have to find the field that you're trying to update. And as a RevTech manager, it's also hard to manage because you have to manage the fields that people want. And people can't create things, and when they want to request it, they have to go talk to a human. So now Salesforce really is just a database and the database can be updated by agents. Sales probably doesn't even need to know about Salesforce anymore. They just need to know that the tool is a database with the information. And if the Rev kind of operator or this the head of Rev wants them to update their prospecting pipeline with information, then you define it in a markdown file and everybody gets that skill. And then at the end of the day, you can update all of your calls to the information you need automatically. So we're kind of moving away from the world where that started Salesforce, which is where you have reps that are logging into this thing to find and search for information and information. And now it's just a database, which, you know, I think like means a lot of things. Like one, I think for companies where they said, hey, Salesforce is dead, it might mean that actually Salesforce has new life. Because maybe it's like less expensive, easier to manage if you don't have a UI.

SPEAKER_00 37:47

And just think about like the number of custom apps that you can just develop on top of that, right? So that's gonna happen. And I have a related story because this week, um, somebody on my team shout out to Rom, Rom Bachu, who's uh uh a GTM engineer for us. So we pipe everything into BigQuery and he built a Google App Script, I think that's what it's called, on top of all that data. And he gives a weekly score to all of my sales calls, right? With like, here's everything that you did from MedPick, here's what you should do in the next call, here's the probability to close, here's the estimated day that this deal is going to close, right? Just based on all the call transcripts. And just it's it's a custom app. I've I have it bookmarked, I call it sales intel, and I can just every week on a Friday open that up, look at what happened in the last week, and prepare for the next week, right? Like you can create all these custom apps because all of that data is like neatly flowing into, you know, a warehouse, and you could have a gazillion different apps. That was like what coaching app. Maybe there's a different app about, you know, uh custom landing pages for every one of those customers, like whatever, whatever you can think of, right? It's just, and I think that's where the the world is gonna go. Like Salesforce admins are gonna be GTM engineers who are hooking up all these different things and and building custom apps.

SPEAKER_01 39:09

I mean, there's a million ways to do things now, right? You can build a custom app. You could still have had a Salesforce dashboard, you could put it in SQL and have Cloud write a SQL script, you can put it in hex and have AI write a SQL script for you. There's just, I think like people I need to dispel with the idea of like the right way or the best way of doing things and think about like what is going to be easiest for me to do in the long run. And actually, this is where I think in the very long run, architects, as people who think beyond the initial entry point of like, okay, this just works, to how is a team of X people going to be able to achieve the same outcome? Because like the then the micro app works for you, right? And you're not really as a small growing startup, you're not thinking about more than PRNOV. And even if Pranov plus five, it doesn't really matter. But eventually, if Paramark's a 400 person company, if you're company, now you it actually does matter how you do things and you have to make things accessible and easy, and there has to be knowledge about it. And like if you can do it, everybody has to be able to do it. So I think like, you know, what's kind of amazing and powerful right now is that for small businesses and solo operators, you the best way to do something is the way that works for you. And there are a million ways to do things really, really well. I think for later stage companies, they're gonna find out that go-to-market engineers and this new term called marketing engineer, which I totally want to talk about. Those folks don't know how to architect systems for thousands of people. And the stuff that works really well for growth isn't gonna work well at scale. And so I I think we will inevitably run into that problem.

SPEAKER_00 40:44

Yeah, yeah, yeah. No, I think and I think that's you've described law's opportunity and risk here really, really well, right? Which is horizontal software for very like, you know, it's an inch thin, but it's a mile wide because you can kind of address these 10%, 20% use cases for like a you know, large audience. And when you need real software, you're still gonna rely on real software. And you need real architects and engineers and designers and PMs to you know make that make those things happen. So I'm I'm I'm totally with you on that. Let's talk about marketing engineer. Like, did Profound come up with this term? Is that it at their event?

SPEAKER_01 41:24

I think a head of marketing realized that that Clay did this exact same thing with go to market engineering and said, what if we do this? What if we're smart? For what it's worth. I used the term marketing engineer in 2019 and everybody thought I was dumb. Um and before that, marketing technology existed as a term. And before that, growth engineering existed as a term. I think like, again, look, I I'm not contesting maybe.

SPEAKER_00 41:51

Kudos to Profound, because I will tell you, like, I heard the term like five times last week after their event. And uh people around me were like, hey, like this term is awesome. Can I use this term? Like, should I be using this term to define who I am? And I'm like, go for it, man. Whatever floats your boat, right?

unknown 42:08

Exactly.

SPEAKER_01 42:09

That's exactly my point. I think that like it was great advertising, it was well done. His piece on LinkedIn was good, even if I disagree with him being the first marketing engineering ever, you know, like a little too high on his horse, in my opinion. But I I think it was a great use of the term. And probably what's great for both profound and for operators is that like if you're one of these people who's like a marketer who likes to vibe code and you haven't really been able to describe yourself to people, it's gonna be an amazing way for you to describe that you're not a full engineer, but you're also not a pure marketer. You're something in between. You're this pseudo-technical person, you're a ninja, you're able to deploy yourself and use tools and technology and AI to get work done.

SPEAKER_00 42:49

I'm a little bit worried about it because I'll tell you why. I think when I think about somebody with an engineering title, I really do mean that, you know, the the app that I just described, right? You have the ability to actually publish an app end-to-end. And if you're not, if you're not equipped to do that end-to-end, I think putting an engineering title is a little bit dangerous and is going to cause like just the hiring managers to to make mistakes in hiring because you're like, oh, I'm hiring a marketing engineer thinking that they can like go all the way to production with X, Y, and Z, and all it is is, you know, a lovable app. I'm sorry. Like that that's not it, right? So this is the risk.

SPEAKER_01 43:31

This is the risk. For what for what it's worth, I mean, all this comes down to the fact that people, there will be some people out there who will overuse that term and they will burn other companies and they'll burn themselves. Right. Like I I have deployed into many client organizations where somebody hired a go-to-market engineer who was not a real engineer and deployed a bunch of stuff and didn't think about how to connect it. And it's an architectural next to fix afterwards, and it costs them$100,000 or more between tools, technology people, and contractors. That will continue to occur, and that's going to be exacerbated by people using the marketing engineering term. At the same time, you know, if you actually are pretty technical, you know, like you and I are fairly technical. I used to describe myself as a really bad engineer because I actually have an engineering degree. I went to a coding academy and I learned how to code, but I'm not a very good engineer. And like in any version of the world prior to 2019, I would never have called myself a software engineer because I would be shamed for it. Like I would literally be rocked. People would pelt me to death for doing that. But now with AI, I'm actually pretty close to being a good software engineer because the thing that I could never do well is I didn't have seven years of writing software experience. But I knew APIs, I knew object orientation, I knew how servers worked, I knew how HTTP works in JSON and all the core concepts that you want in a basic software engineer. However, I think like a lot of people who are using AI right now don't have basic software engineering understanding. Correct. And they're calling themselves an engineer. And this is also the reason why I'm like all the stuff that I teach, I always start with. Let's just start with the basics of coding. Because if you understand the basics of coding, you're gonna actually be able to call yourself an engineer a lot and have like more confidence in yourself because you have the primitives there.

SPEAKER_00 45:11

I think that's the that's the difference, right? I I I actually think it would have been easier if they, or or maybe more effective if they called themselves like a marketing coder. I don't know. Something, something, something that that's less engineering heavy, but anyway. Exactly. Hacker, marketing hacker. Oh my god. Uh all right, okay. One other piece before I want to wrap. So this is a bit of a long episode. Uh, Canva launched a whole bunch of stuff, and I didn't realize they've they've had five acquisitions in Q1 of 2026. Syntheary, Magic Brief, Mango AI, Duli, Cavalry, and then this one you're gonna like, a company called Orto. Orto is basically, I think the previous name of this company was Autopilot, and they were essentially competing with like the HubSpot and the MailChimp for like lifecycle marketing and and email marketing. So, and they launched a whole bunch of new like AI stuff and Canva Code 2.0. It's just, I didn't realize how much AI Canva has been building and launching. And there are people who like really uh swear by Canva as like their main way of operating, like it's got sheets and you know uh docs, and I'm just like Canva Code. I'm like, what? This is a thing? And they're doing like 4 billion in ARR, they have like almost 300 million active users, uh, monthly active users, like bonkers.

SPEAKER_01 46:38

I I want to ask you an honest question. What do you what have you used Canva for?

SPEAKER_00 46:43

In the last three years, I haven't used it at all.

SPEAKER_01 46:46

Okay.

SPEAKER_00 46:46

So what do you and maybe before that I was using it for like very basic like social media asset or something, and then I gave up on it because I was like, I'll just use Figma. I like Figma more.

SPEAKER_01 46:56

Yeah, yeah, yeah. What's interesting is like I feel like Figma is for design, is designed for designers, and Canva is designed for marketers. And like this this um happens in every organization is that you know, oftentimes marketing actually needs design assets, but they're not the same design assets as the product design assets. And so design with kind of you know apostries be on each side, is like they're designing the product in a tool like Figma. And then marketing will like need stuff, and and my experience has always been design complains about and never wants to help marketing do their job. So then marketing has to go out and hire contractors and other people to help them do the design work for marketing, which is ads and HTML for emails and social posts and all the things that go into the external world that are not inside the product. And so I've always thought of Canva as being really a tool for those people. But what's interesting is like it's a piece of non-technical marketing surface area. There are other pieces of technical marketing surface area, like high touch, I think is is technical marketing service area, and they're doing a lot of AI stuff around, you know, like let me pull all your ads from Meta and then help you resurface existing one. But you have to create that somehow. Where do you create it? You create it in a tool like Canva. And so I think it's interesting that a lot of these companies are basically starting to cover big swaths of the marketing almost like, you know, kind of surface or the the area that encapsulates marketing. And in some ways, they're like doing the things that that HubSpot just doesn't do. Um HubSpot, I would almost argue like it wouldn't, it would it make a lot of sense for actually HubSpot to acquire a tool like Canva. Because then you can build the project doc, you can create the design, you can deploy the ad, you can write the email all in one spot. They're covering this really important piece of what eventually goes into HubSpot. And HubSpot has like landing page editors and stuff, but like they don't have a design thing to start with.

SPEAKER_00 48:51

So, okay, let's let's look at this. HubSpot's revenue is 40 million, 40 billion, right? No, no, no, no. Revenue, dude, 3.1. Yeah, yeah. HubSpot revenue is 3.1, and Canva's revenue is 4 billion. Canva's making more than HubSpot. Oh wild. I yeah. That's at least what the public reporting is. So it would be a merger of equals than uh a real acquisition, which is wild.

SPEAKER_01 49:15

Which is wild to me that Canva is as big as HubSpot. Bigger than HubSpot. Yes.

SPEAKER_00 49:20

Yes, exactly. All right, on that note, um, we're gonna bring this to a close. There's a long episode. Thank you everybody for tuning in. Episode 21. Austin, what are you gonna do for the rest of the day? Are you gonna rest?

SPEAKER_01 49:32

I'm gonna rest. I'm gonna go get my bike fixed so that I don't have the same problem again. Need some food. It's gonna be an amazing weekend. And then tomorrow I'm gonna run a half marathon, you know, my typical prep. And then in November, we're gonna run together. It's gonna be great.

SPEAKER_00 49:46

That's indeed right. I have my run day tomorrow. I think I'm just like I'm way behind you. I'm gonna do a three mile, but time it. Um, and then next week we're gonna start ramping up. So it's gonna be fun. All right, folks, have a great rest of your day whenever you're listening to this, and tune in again next week for another episode of GTNN. See y'all.