ChatGPT Work For Marketers (& The Trap To Avoid)

The interesting part of a new AI model is no longer just what it can answer. It is whether it can hold context, work with the files that matter, and take useful action without making us repeat ourselves all day.

That is why the shift toward AI workspaces and agentic apps matters for marketers. In July 2026, OpenAI had brought much of Codex's way of working into what it called ChatGPT Work. The names, models, and interfaces will keep changing. The larger change is more durable: AI is moving from a chat box that gives an answer to a working environment that can find context, create artifacts, and continue a task.

The harness changes how AI is useful

The model still matters, but the harness matters more than many marketers realize. A useful work environment can keep a library of instructions and files, find the relevant context on its own, and make the things you ask for. That could mean a document, an image, a PDF, a landing page, or an edit to an existing asset.

That is a different experience from pasting a prompt into a new chat and rebuilding the context every time. When the work has a home, the AI can leave breadcrumbs for itself and look back at what has already happened. The practical gain is not that marketers suddenly need to become developers. It is that more of the work can stay connected to the actual project.

For a large task, I want the AI to have enough context to make a good decision. I also want a visible place for the working files. That gives me something to review, improve, and hand off instead of a pile of disposable conversations.

Pick the environment that helps you move

There is no prize for loyalty to a particular AI harness. If your work is organized as durable files and clear instructions, you can move more easily when another tool becomes a better fit. Moving between Claude Code and Codex is a much smaller decision when both systems can use the same files and Markdown instructions.

That is a useful operating principle for a marketing team: own your source material, your standards, and your project structure. Let the AI application be replaceable.

The goal is not to test every new model. Most marketers do not need the most powerful option for every task. A middle-tier model is often more than capable of creating a landing page, organizing a campaign, or producing a first draft. Use more capability when the work requires it, and use economical options when a workflow runs repeatedly through an API.

Voice becomes a working surface too

Voice is becoming more useful when it can handle a longer backstory, wait for the actual question, and help think through a decision. I have found it especially useful during a commute or before a meeting: explain the situation, name the constraints, and ask for considerations or a role-play.

That does not make the AI the decision-maker. It gives you a faster way to surface options and prepare your own judgment. The same standard applies to any AI output: it should reduce friction around real work, not create more information to manage.

The trap is mistaking activity for progress

The biggest risk is not choosing the wrong model. It is becoming dramatically more efficient at work that does not address the constraint holding the business back.

Manufacturing learned this lesson decades ago. A production line can add automation and look busy while total output stays flat because the real bottleneck remains untouched. Marketing teams can do the same thing with AI. We can create more assets, launch more experiments, and automate more tasks while revenue, leads, conversion, or retention does not move.

The question to return to is simple: what is the one bottleneck that matters most right now? If lead flow is the constraint, more clever downstream content may not solve it. If applications are piling up unfinished, more top-of-funnel volume may only make the problem worse. Fix one constraint, then look for the next one.

AI can help map a workflow, organize the evidence, and speed up execution. It cannot replace the leadership work of deciding where the constraint actually is. Use the new work environments to make meaningful progress faster, then measure whether the line moved.

The conversation dives into the evolution of ChatGPT and Codex, the introduction of new models and voice features, and the impact of these changes on marketers. It also explores the transition from Cloud Code to Codex, the ease of switching harnesses, and the future of AI voice technology. The conversation delves into the future of AI in everyday life, the role of AI in marketing, and the importance of identifying and addressing bottlenecks in marketing and business operations. It also explores real-life examples of AI usage and the power of storytelling in AI-driven marketing.

Takeaways

• ChatGPT has undergone a significant overhaul
• Codex and ChatGPT have been combined
• New models and voice features have been introduced AI-driven marketing and the role of AI in everyday life
• The importance of identifying and addressing bottlenecks in marketing and business operations

Chapters

• 00:00 Future of AI Voice Technology
• 24:56 The Future of AI in Everyday Life
• 25:31 High Level: The All-in-One CRM System
• 29:15 Everyday AI: Real-Life Examples
• 34:17 Viral Post: Jeep AI Ad and the Power of Storytelling
• 38:27 The Bottleneck Problem in Marketing and Business Operations

Dan Sanchez: AI is changing quickly, and chat GPT has had a freaking overhaul. And I mean, like, it feels like the whole thing got overhauled. And it's so much so fast that a lot of marketers are left wondering, like, ⁓ do I need to change the way I'm using this app? So today we're gonna be talking about Chat GPT work and what's going on in the AI landscape. I got a ton of stuff because of course we're two weeks since the last episode, and there's just a lot's happened. In fact, so much has happened that we can't really cover it all. So I'm cover what's the most important things for marketers to know, specifically marketing leaders, of course, because that's kind of the new angle we're taking the show. So welcome back to the AI driven marketer. I'm Dan Sanchez, friends call me Danchez, and I'm joined by Ken Ferre.

Ken Freire: Yo yo yo, what's up, bro?

Dan Sanchez: Who, of course, is not my blood brother like Travis's, but is probably the closest brother that I have after Travis, like bro from another mother. You know, why do you say brother from another mother? Yeah, I messed that. That's okay. We're not here because Dan's a smooth talker. We're here because we want practical insights. Because I'm a decent marketer that's learned some things about AI, I guess. but today, of course, the topic is chat GPT work.

Ken Freire: You're a great marketer.

Dan Sanchez: I mean, if you clicked on this 'cause that was in the title or in the thumbnail or something and you're like, ooh, ChatGPT work. Yes, ch if you haven't seen it, it's all over ChatGPT now. It's in the mobile app, it's in the desktop app. In fact, they just completely got rid of Codex, pretty much. ⁓ they closed it down. And I think that's kind of the big news is I've been pushing like, dude, Codex. Codex is the new thing, and it is. It was. It and it kind of still is. ⁓ that was the new way to use ChatGPT. I was only using ChatGPT when I was limited to my mobile phone. And even then from your mobile phone, you could kick off things to Codex on your desktop. ⁓ codex was just such a more it was much more powerful way to use ChatGPT. And of course Codex was the like Claude code was for Claude, Codex was for ChatGPT. But recently ChatGPT combined them all, Ken. They're like, I thought this would honestly take eight to ten months, and they're like, No, we're gonna do it all now. I think this is still gonna go through a lot of iterations and get a lot smoother. There's still a lot of dust settling from this it this massive merge. But Chet OpenAI was like, screw it. We know Codex is the future. And I knew it too. I think I even called it on this podcast. Like a year I said a year from now, what we're doing in Codex will just be standard across every AI app. Because the one big th difference with Codex, and this is true for Claude Code or anytime you're using like a what they call a harness or a code harness, is that it had two different behaviors that just made the whole experience of working with AI different. One, it could make its own documents and then reference them itself later, kind of like instructions and custom GPTs or instructions in a project folder, except you had to update those instructions. AI might write it for you, but you still had to copy and paste it, put it in there. And then every time you wanted to change it, you had to paint, like not painstickily, but you had to go manually change it. Kind of a pain. Now AI can just manage that for you. And it's not just one set of instructions. It can keep a whole library of them. Leave little footnotes for itself. That's kind of a big deal. So the fact that it could do that. take action and make things for you. Images, graphics, PDFs, music, all c whatever. It can go make things for you. And the other thing that made it so different was that it can be proactive. You could ask it a question and it would think to itself, huh, I should probably go check the files to see what was what we did last time. And then when it find all its little breadcrumbs at least for itself and then go find it and figure it out. So it almost have this like memory because it was proactively figuring it out. And it was trained that way because that's how It needed to work in order for it to write good code. It would need to find the context of where that code you were asking for would exist. Just so happens that that worked for everything. And that's why I've been using Codex, even if I wasn't coding anything. It was just smarter because it was proactive and because it can make its own memory files. ⁓ and I know you're you're on the Claude Code team, and I guess my goal, Ken, is to like convince you to like stop using Claude Code and move over to Codex. Chat GPT.

Ken Freire: Well, genuinely, wasn't it the other day I t I sent you a text and I said, Hey, this is what ChatGPT recommends for my new tech stack workflow? And you're like, it was like cloud code for this and codex for this. And then you're like, what are you gonna do? And I said, Don't ask me good questions like that. I'm not ready to change.

Dan Sanchez: Yeah. Yeah. Every t every once in a while someone like you ⁓ will send me like, look what chat told me and I'm like, Cool, what are you gonna do with that information? Like, are you gonna do it? Like what are you gonna do? What decision are you gonna make? You you now get to discern if that was a good idea or not.

Ken Freire: Exactly. Which brings us to the point of our conversation today, which is what we're gonna talk about is we get a lot of information and it's like, are we actually being productive with the things that we need to do?

Dan Sanchez: We're saving that for the end. That's the teaser in the title of ⁓ The thing you need to avoid, the thing that's getting most marketers in trouble with AI. And when I say most, I mean most. Like almost everybody I come in contact with, whether they're pros at AI or just getting new, getting started with it, there's a major flaw in the way most people are using AI. We're gonna save that for the end because I think that's gonna be, I don't know, a game, probably it's been the most game-changing paradigm I've had in the last month and a half that I think is going to be probably more applicable to everybody as everybody starts leveraging AI. But let's finish it. Up with ChatGPT work. This is the most significant update ChatGPT's made. People are saying this. Most significant update they've made since launching four. Okay? Like, like, it's which was like years ago. So

Ken Freire: Pretty big deal.

Dan Sanchez: And I mean like the jump from 3.5 to 4. Because again, it's bringing that power of codex into everything. In fact, they got rid of the codex app completely and just it's just the new Mac Mac GP chat GPT app. And it works a little bit more like how Claude had like cowork and Claude Code, right? And now you you it's essentially codex. You open it up, I'm like, ⁓ this looks exactly like codex. But it's chat GPT work and then you can flip it to code mode. Which generally work the same, but if you flip to code mode, it shows you a little bit more of the little bits of coding UI that's just helpful so you can track changes that are happening to what and how many code line changes, or even if you're editing like a Word document and it's a bunch of copy, it'll tell you like, ⁓ which lines of the copy change. It's actually a pretty helpful interface. ⁓ it's more oriented towards code. I use it when I'm like when I'm dedicated when I'm going into a heavy code project, like right now, completely I'm I'm ditching words. I'm like done with WordPress. Done done done. Freaking done. Like I can't tell you how done I am with WordPress now. Because now I'm just re- I'm just having AI scan my whole danchez.com WordPress site, take all the content, and it's just gonna be vibe coded. No s no content management system. Stupid. Like it's gonna be vibe coded by codex or chat GPT work. And just put into Vercel and just recreating the whole website and changing the styles and fixing all the forms and I'll talk about that in a bit, but it's it's amazing. I'm like, we just don't need WordPress anymore when you have like a tw when a twenty dollar subscription to Vercel can be your host and a twenty dollar subscription of Chat GPT can get most of your coding and what web app needs done. It is it's like a complete game changer. But cowork allows you to do that kind of stuff and you don't have to know how to code to do it. ⁓ it also brought it into the multiple. Like do you do you subscribe to ChatGPT or are you just all in on cloud? You got Volve? Did you notice ChatGPT changed on your mobile app?

Ken Freire: No, no, no. I got I got both. ⁓ ⁓ yes. And I'm I'm still trying to play around with it, but I feel like you're gonna talk to us today about how to master it a little bit. But it definitely changed.

Dan Sanchez: A little bit. I think we're all trying to figure it out because it's a little confusing, but here's here's what here's two really cool things that I've learned in playing around with these things. There's you definitely want to install the chat GPT, the updated codec slash chat GPT app on your Mac. You also want to use the mobile app and the web app. You know how everybody's running into limits with how much you can use these things? Well, guess what? You have two different rate limits. There's one for the Mac app, and then there's a completely different limit for your phone slash web app. Yeah. So what people are figuring out is if you got some big projects and don't want to upgrade to the hundred dollar or two hundred dollar version of ChatGPT or going into the API, is you do all your ideation on the web app. And then you can now

Ken Freire: No.

Dan Sanchez: kick off that whole conversation. ChatGPT is actually giving you like special links to like pass from one to another. So they actually thought about this. They're like, we want you to ideate here. And then when you got want to get the work done, whether it's coding a whole book or like writing a whole book or like ⁓ anything where you're gonna kick off multiple agents to go do some work for you. Then you bring it over to the Mac app where it has more power. And it has more power because the Mac app isn't restricted to just the app. It can actually use all these pre these These invisible tools you have on your computer, especially Apple computers, have all these invisible tools that you don't even know about that extend the functionality of the Mac app. That's what makes it powerful. If you drop a video like this podcast recording, I could drop a whole video in it and be like, go find the most intriguing parts and clip it together and make these specific edits to it. Add this one finisher three-second part to the end of every video to do this promotion. It can actually, using its code tools or whatever, it can actually do it. Not, it doesn't even have to. Design its own application. There are tools on your computer to do that that aren't Final Cut, aren't the Mac what is it? Not Garage Band, but you know, the Mac, the Apple, the che I mo iMovie, that's what it is. It does not using iMovie to make it. It's got like this code version that it has access to to edit videos and it will just go and tap into it. It's nuts. There's all these tools on your computer that I can use to do all kinds of crazy stuff. ⁓ and that's the power of the Mac app. So and it can like tap into APIs and all these different co like MCPs, plugins and all this stuff to work with your other tools and actually orchestrate a whole team of agents in order to get things done. It's extremely powerful.

Ken Freire: Yeah, I move you.

Dan Sanchez: Powerful. The they tried to pull some of that into the mobile app and the web app though. So if you notice at the top of your screen there's chat now. Used to have the model there. Now it says chat and you push it and it goes down to work. Now this is kind of the weird part. This like work, there's work on your chat GPT Mac app, but there's also work on your phone. So this is where it gets hairy because also from your phone they kept the response the thing where it used to be codecs over in the sidebar. Over in the sidebar, it used to have codecs where you could like essentially remote into your computer and access that Mac app called Codex, now called Chat GPT. It's the Mac app. ⁓ so you could still do that. You could still remote over to the more powerful thing and get it working for you, act answer questions as it's working for you, whatever, if you're out and about and want to keep your computer working. That's cool. But then we also have this thing. Called work. So what's that? That is some kind of weird hybrid they made for people who want some of the functionality of this powerful app on your computer, but still do it from mobile or from the web app. It essentially has this other place where it can create MD files and then access them later. It also has the ability to kind of press on and go longer and have that proactive mindset where I can leave breadcrumbs for itself. The problem with it is I don't know how to access the files it creates unless you ask it for it. I don't know where they go. No. There's they're on the cloud. They are not local.

Ken Freire: Yeah, I was gonna I was gonna wonder, do they save on the cloud or do they save locally?

Dan Sanchez: ⁓ but I don't know where they go. I'm like, where are these files? I assumed, like I started using it within a project folder. You're like, ⁓ so it's gonna exist under files in the project. No. They don't exist there. And you can't, you won't change any of the files in the files if you upload like a doc or something to one of the projects in the mobile app. ⁓ it can't change those files. It could still reference those files. But on the Mac app, like if you give it a doc X, it can go and edit the whole thing. It'll search the code, find the content, edit the doc. No matter what file you put into it. PDF. Shoot. I mean the Mac app's so powerful now. I'm like, hey, here's some raw content. I need you to turn this into a well designed little handout. And it will go through and design it and set the type in a way that's like, ⁓ this is cool. I'm just gonna send them this. It's like really cool what it can do now. ⁓ it's not as good as a designer yet. ⁓ Bec as far as like document layout and text text setting and all that kind of stuff, but it's getting there, it's gotten so much better. but that's that's the work part of the mobile app. It's better, it's certainly more powerful if you need to do some more work on the fly, but it's not nearly as good as the Mac app. But it's an interesting in-between that they're playing with, and I think we're gonna see a lot of refinement to this over the next couple of months. Sam Altman even teased everybody, he's like, Hey, we haven't been doing the best on On being cutting edge over the last six, ten months. But you're gonna be we're gonna knock your socks off over the next year. He's like, what we have coming, like be prepared. I was like, okay, okay. I mean he's the hype man and he's trying to raise money and k keep himself out of all the freaking trouble they've gotten into recently, which I'm not gonna go into here. Listen to other AI shows if you want the AI drama because it's not really relevant for your work. But at I'm I'm pretty sure they're right. They're they're cooking pretty good pretty hard over there. ⁓ to the keep stay try to stay ahead of Claude and Tropic. ⁓ but I think it's honestly I love it. I highly recommend all marketers start using ChatGPT work, the Mac app, and then experiment with ChatGPT work on the phone to see how you like it. I just think it's a better way of using AI. And again, it's just I think ChatGPT work on the mobile isn't quite ready, but I think the refinements they'll bring over the next couple of months, like I bet eventually you'll be able to just find the documents and the files in a project folder. On your phone. I bet you can even in the future sync your mobile your your cloud-based stuff down to your computer in interesting ways. Like it's gonna be more fluid to the point where you could just have a conversation with Chad as you're driving and it can get to work on files that you have in the cloud and then sync them back to your computer later and then set reminders for itself and do all that kind of stuff. Again, we're getting closer and closer to Jarvis, but it's getting freaking dang close to what Jarvis can do now.

Ken Freire: And and I think the b the biggest issue, you know, like 'cause I'm on the cloud code ⁓ train right now, everything because it's saved locally, if I want it digitally or on the cloud, I gotta put it on a GitHub and a repo. Like I gotta do all this other stuff where this is starting to like centralize everything. It makes it a lot easier.

Dan Sanchez: Mm-hmm. Mm-hmm. Yeah. I s I still I still back up everything to a GitHub repo and then I just schedule the task in Chat GPT to like, hey, every day I just everything in this folder I want you to push back up to the repo just as a backup case. That's always that's a good practice if you haven't done that yet. I will say, Ken, I really want it I want to like Cloud Code. I really do. I have a subscription, a nice subscription that my work gave me to use it because the IT department are all Cloud Code fanatics. But they're IT guys and they're really good at it. So guess they're doing? They're all doing it in terminal. They're all terminal guys running it through there and I'm just like frick I don't do this. And I so I'm using the app, trying to like sync it up with things, trying to use it. I don't know what it is, bro. I feel like it's like the difference between using a PC and using an Apple computer. I get confused using Claude all the time and I get stuck. And then I go try to do it in Codex and it just happens faster. across all these little tiny things that I'm trying to do. I don't know what it is. Maybe it's just my personality, but every time I use open AI stuff, it gets done. If I try to use Claude, it takes me two or three times longer.

Ken Freire: Yeah. I you know, I am going to at some point test out codecs on certain X certain projects that I'm working on. Right now I'm working on a big project in on Cloud Code that makes it like I don't want to switch at this moment. Let me finish my project and then I will switch to codex and be like, ⁓ my gosh, it might be ten times easier to do this. ⁓ but but the key is the MD files, right? What?

Dan Sanchez: If the cool thing is it's e it's easy to switch now. Yeah, the MD files are all the same. So you just switch you just give cloudecks or now chat GPT work. I don't know. They if they have access to the same file bases and you can transfer your skills over easily from one to the other. So it's like whatever. If it it someday like Claude might come out in three months with the thing that's just really better and then I switch over, but it won't nowadays because of the way things are organized, it's it doesn't really matter. You can switch harnesses pretty quick.

Ken Freire: Yeah. I I think the harness is definitely gonna be one of those things. Now, before we go on too many tangents, I will say for people who are trying the mobile app, ⁓ it might take a little bit. From my understanding, they're still rolling it out. So if you don't get it, just be patient with that. And then also make sure you update your app for the Chat GPT. Yeah.

Dan Sanchez: For chat. Yeah. I th ⁓ thought they were finished. They started that was a week ago they rolled that out. So I don't know. If you don't see the like little toggle at the very top of the chat GPT app, it says chat, and then if you toggle it, it should say work. Or it'll give you those two options. So test it out. Tell me what you think. Find us on LinkedIn, DM us. ⁓ moving on to the next thing is chat chat that we're all ta it's all chat GPT today. Okay, because I'm obviously that's my ecosystem. ⁓ the other thing they launched, of course, was new models. But I thought the work thing was actually more important than the models, which is crazy. It used to be the other way around, but the harness is actually becoming the bigger part of the conversation, not the raw power of the model. But the models are still important. So ChatGPT released th three new models and each of them have like three or four different weights. Remember last year before they released ChatGPT five, there was a problem because you're like, ⁓ when should I use O three? Or what was it? 4.5 was a thing. And then there was 4.0, and there was like all these models and different ways of using it. And it was confusing because only pros knew kind of knew how to toggle between all these different things, like a stick shift. You're like, gee, I'm gonna switch down to this one, I'm gonna switch this one. Yeah, we're we're in that problem all over again. And it's actually more confusing because we have like the power model of Soul, like Sun, and we have Terra, which is like the middleweight model, and then we have Luna, like the smallest fast model. And then each of them have light, medium, heavy, sometimes super heavy or whatever. Like there's like three or four models between them. And then you're sitting here as a marketer, the devs, the devs like it, but even the devs probably only go down to like bottom gear and top gear a lot. They're like trying to go with the cheapest one or they're gonna throw everything at it and go at the most expensive one. But for us marketers, I'm like frick, I don't know. I'm usually just like just give me the middle one until I need more power and then I'm going the The biggest one.

Ken Freire: Yeah, I never use the lowest tier. Like I don't think I've ever used the lowest tier on Claude or Chat GPT. It's just like middle or high.

Dan Sanchez: Lowest tier is really important when you're doing things programmatically. So if you are designing something for your website and you want your website to like every time someone submits this form and you want AI to analyze it and then send you some results or the results to the user, then thinking about like these things matter because they cost. in the API credits. So like you usually want to go as low as you can to keep the cost down. That which that's what matters. ⁓ that's why devs love it and think about it all the time. But like normal users, like we're just usually going middle or high. And so that's kind of what I do. I'm I'm I'm in Terra middle because I don't want to burn through my credits too fast. And honestly, like the middle ground is so freaking good now that like I I don't have I'm not trying to solve world math. I'm just trying to get a landing page up. That's most marketers. We're just trying to get freaking crap done. So like like the middle way They're so good now that like it doesn't like the models have gotten beyond what most marketers need done most of the time. So it doesn't like the models are going to keep getting better this year, and the only thing marketers will notice is that they just get smoother and can go for longer periods of time without checking in. That's what we'll notice. The devet the devs actually have to calculate things and like it's much a bigger architecture and all that kind of stuff. But for most marketing, it's like I don't think it matters that much. I'm not throwing complicated enough stuff to even need fable. I don't need to program video games.

Ken Freire: that's what like I had a friend of mine who's like, have you tried Fable? I'm like, I don't I'm not working on a project that requires Fable right now. That's worth it. Like why am I gonna spend time doing it? You know, I'm re-hauling a website that I own, but like I I don't need Fable for that. ⁓ but I think the one thing I will say is Seoul is supposed to be the competitor to Fable but I heard it's not really there yet. ⁓ it's

Dan Sanchez: ⁓ that's even worth it. Hmm. It's close. It's close. It's comparable. It's in the same range. And in some places I think does a little higher, but in general, the consensus that I've kind of seen is that Fable's still like a half step above it. You're like, Okay, cool. So Anthropic's definitely been taken off there. And then of course everybody likes using them for all kinds of for the the code stuff. So I'm like, great. I'm not coding anything nearly complex enough and I'm just recreating WordPress sites. I'm recreating WordPress sites. And the other thing that I've figured out recently, Ken, is that ⁓ I'll I'll say I'll save that. I'll save that for a little later. We're gonna talk about high level again, of course. Because one one more thing I want to talk about before we talk before we move on is chat GPT voice.

Ken Freire: Yeah.

Dan Sanchez: They did that, they upgraded that last week too, and that's that's been about a week and a half now. And I freaking love it. It's amazing. It's been long overdue because the last voice model was two years old based on the old four O model, like the OG four ⁓ model. The one that was super sycophantic, hallucinated all the time, would just lie to your freaking face. Of course it didn't lie, it's just predicting incorrectly, because To lies to have the intention to deceive, and it's not intentionally deceiving you most of time. It's just guessing incorrectly. ⁓ and it was just bad. It after t it was great when it came out, it was b mind blowing, but now it was getting behind. And nobody had really pushed the ball for it, but now ChatGPT has its voice app and it's so much more fluid. You could be like, Hold on, I'm gonna tell you a story and then when I ask for feedback I'll I'll Cut cut in. And then you can go and talk for a while and be like, okay, so now that I've given you the backstory on this this situation that I'm in, help me think through the what I should be considering. And then it'll start talking. It's so much more fluid. And it's smarter. It can look up stuff in the background while you're talking and actually do the research and actually think more critically. Because before it wasn't a reasoning model, now it has some lightweight reasoning to it, which has actually caught it up. It's not as smart as like your these new models they just released, of course, but we didn't really need

Ken Freire: I have not used it yet. It's on my like, here's my weekend drive. I'm gonna work on, I'm gonna play around with it.

Dan Sanchez: Yeah. I'm using it the most while commuting, which I commute four times a week now. So it's amazing. And it doesn't like it used to have a problem with car noise. It would like think you're talking, but it was actually just road noise or some siren going by. No, it doesn't get confused by that anymore. So you can literally just I turn it on and I'm like, hey, ⁓ processing this situation, or hey, I'm gonna be jumping into this meeting soon. Here's the situation. I need some help thinking through considerations or different ways this could go in this meeting I'm about to jump into. And it will just process it with you. It's it's like having your friend on superdial to like be a buffer anytime.

Ken Freire: Yeah. That's fascinating because like I I didn't like the four point ⁓ voice model just because of all the things you've already referenced. ⁓ but my wife loves it. My wife loves the voice model. Like she's always talking to it and I'm like, it this is too slow. I need something faster. But now I'll try it on my next drive.

Dan Sanchez: Yeah. Yeah. It's really fast. It's still not perfect. It's still like you have like a when we speak, we often have these like little moments when we pause, just like I did a second ago. And it will because it's trying to be faster. it will interject too much, but it's really quick to interrupt. It also has all the little verbal cues like you just did, like yep, you know, like it has all those in it too, which makes it more fluid too. It has it's very conversational, much more fluid. It's not perfect, but it's getting there. I honestly think the future of AI's voice. Like again, Jarvis is the best picture I've seen of where we're going in the future. And we're all just gonna be talking to our agents. Unless we're in a situation where we can talk and then we could do it via text. And it'll be like texting us. Kind of like if you remember having those meetings in the past where you have like everybody around a big table and your your coworker, BFF, ⁓ is like texting you the thing that everybody's thinking. ⁓ That'll be your AI agent because it'll be listening, be like, Hey, like your boss just said this. Do you want me to get started on this so I can have it done before you before the meeting? It'll be like texting you and you'll be like, peeking under the phone. Yes.

Ken Freire: Yes.

Dan Sanchez: That's how it's gonna be. Yeah. Yeah. Or it'll just anticipate it and do it, but you know, like so we'll always have that agent kind of listening on and being able to be like to talk to while we're working on things. Like we'll be working in apps and the the agent will be like collaborating with us vo via voice mode the whole way through. That's kind of how I see the future being and we get closer and closer to that every single year. But I think voice is the main mode we will be talking to a lot of agents.

Ken Freire: All meniotas, just automatic yes, just do it. ⁓ Yeah, one hundred percent. All right, let's move on to our sponsor.

Dan Sanchez: Sponsor, which of course is high level. The reason why I love talking about high level is because actually a lot of people have predicted this, and I think it's true that software, the software industry is gonna take a hit. But it's not the big software piece, it's gonna be all the middleware. It's gonna be all the extensions, the plugins, the things that you need that are special to you. I don't think like there's some people saying like they're gonna rebuild all their software and they're gonna pay for no SaaS. I actually think that's a mistake because you don't wanna be responsible for all the biggest pieces and the compliance and the things that we don't know, we don't know. That's why I high. High level because ⁓ of course they I've been plugging them on the show forever, but you really should look into it if you haven't checked it out yet. It's kind of like an all-in-one CRM system. Think Hubspot, but actually more powerful and way less expensive. Like a hundred bucks a month to start, and it gets you all the features. Like then you hold back. You get all the users, all the features, all the text, all the emails. There's some variable costs for every text, but an email, but it's like dirt cheap compared to everybody else. ⁓ that's why I love it. And it has all the tools in it, but it's nice then you can. Have one central tool, and then you can have AI just code all the that's what I'm doing now. I'm essentially having AI code all the tools and then it just I'm like, ⁓ and AI integrate this into high level, and it taps into its API and just builds its own integration. So my new website I'll be launching will just be automatically integrated into high level. It'll code its own forms, and then instead of using high levels forms, I'll just pass it through the API to update contact records and kick off workflows and all this stuff. And it's got an impressive amount of AI tools. I still think their AI tools in high level are up and coming. There's some that are really good, like the phone system, being able to answer phone calls with AI, fantastic. ⁓ it's website builder. I'm just gonna use ChatGPT because I'm already paying for it. ⁓ and it's easier, faster. but it's become one of the central pillars that I use and I recommend to everybody. I think we will all have some combination of like our own tools that we're vibe coding and a central hub like high level. You might have a few central hub like like Google Workspace and high level or like two central hubs and maybe a few others, but like that's it. And then ever all the other tools you have are just your own that you can get to do whatever you want, whenever you want. But it's good to not have to think about the security of my whole database because high level takes care of that and thinks through all the things that I don't have to think through. Of course I still have to think about the security of my own little stack so my EPI doesn't get out there, but that's Where everybody that's probably like why big players won't do this yet. But all you small players or small businesses, this is this is where it's at right now. Of course you have AI trouble triage all the security issues, but I think this is gonna be the future. And I think high level's by far one of the best tools to do it with. Ken, you've been using high level. You haven't probably started tapping into the API yet though. You started? There you go. You know what's up.

Ken Freire: ⁓ yeah. ⁓ I started with my new project that I'm working on. Yeah, yeah. It it's been a lot of fun. ⁓ I keep trying to push it. I'm like, hey, can you do this? They're like, sorry, we don't have act full access to do that. I'm like, come on, figure it out.

Dan Sanchez: I know. I'm like I'm like ChatGPT, can you just build the whole workflow and high level for me? It's like ugh you can't ha you can't do that via the API. I'm like crap. And there's an agent high level to build workflows, but it's not very good.

Ken Freire: Yeah. I've been trying that multiple times. It's it's not. Yeah, so yes, that happens all the time. I I use that their agent on and Go High low, it doesn't work. I've tried ⁓ Cloud Code to do the workflow or the automation. It doesn't, but it gives us a it what I've really enjoyed is that it's starting to give me really clean and clear data of like how to build the workflow and thinking through all the other workflows that I'm building, I'm starting to let Cloud ⁓ know like here's all these major workflows that I have.

Dan Sanchez: Yeah. Yeah.

Ken Freire: Make sure that I'm not messing anything up. ⁓ and it's been keeping everything pretty clean, which I've thoroughly enjoyed.

Dan Sanchez: Yeah. Yeah. So if you want to test out high level and just give it f take it for a spin You they generally have a two-week free trial, but you can get an extended ⁓ 30-day trial if you go to dance.com slash high level and take it for your spin yourself and find out. It's a powerful tool for the least amount of money that I've seen, unless you're using some freebie tool out there, which if you're brand new and don't have any money, then definitely start off with a freebie tool like SendFox. I like kit until you have to pay for kit, and then ⁓ high level's way better. I'm like, use kit.com until you run out of the freebie tool, and then once you're ready. to actually invest in something, just migrate to high level. It would way better. Sorry, kit. I've tried I high level's better. Moving on to everyday AI, I like to give some examples of what's actually like how you're using AI in life. So I'll I'll kick it off while you you think of something. ⁓ the voice mode, I I gotta say, is like so good for processing crucial conversations or hard conversations you want to have. Because you know, conversations can take different paths. When you go into voice mode and you kind of say, Hey, wait a minute, I'm gonna give you a long backstory. And then when I say to give me feedback, give me feedback, you can give the whole backstory of the situation while you're driving or doing laundry or something.

Ken Freire: I got a few ready.

Dan Sanchez: Then you could just talk it through and have it ask you questions, talk through different scenarios and be like, okay, let's role play and give me some feedback. It's great for that. It's so good. ⁓ what do you got, Ken?

Ken Freire: The biggest thing I'm working on is overhauling ⁓ my blog. So you you know I have a blog that ⁓ it was just outdated and I didn't have all the meta tag, like all the tags set up, everything. So I just one day decided to ask Claude and I'm like, hey, can you like update all these things? And it's been doing it and now I'm able to ⁓ auto like upload my my blogs. I'm just giving it all my ideas, it's reframing it, it's grading it. So now I'm posting way more.

Dan Sanchez: Yep. Mm.

Ken Freire: And it's just been a whole lot more fun. Like I'm getting excited to to get this blog up and running again.

Dan Sanchez: Next thing I have is again using voice mode. Something I love doing now is I love to when I'm commuting, love to listen to podcast, audiobooks. ⁓ love to learn. But now I can s that like I have a whole new way of learning because the voice mode is actually good now and can research things for you while you drive. And so like I'll have a question in my mind while I'm listening to a podcast. I'll pause the podcast, flip it open to voice mode, have it on like the what is it, the programmable key on my iPhone. I just click on this key and then voice mode pops up. So I don't have to like look at it. I just push the key. And then I just I ask it questions and learn through that way. So like right now I'm on like ⁓ I finally gotten into weightlift ⁓ weightlifting and I'm really excited about it. I love it. I don't know how I've waited so long to like get into weightlifting. but I'm I'm like on a I've been on a fitness kick for about a year and it's probably gonna hopefully be more than a kick. It's gonna be a lifetime. But now I'm into weight lifting weights and I'm trying to build muscle and gain weight. And it's a new new thing for me. So I have a lot to learn. There's a lot to learn in this topic too, but it's pretty well solidified like what Need to do. So every time I get stuck with the nuance of it, I just turn on voice mode. I'm like, well, what if this? What if that? What if you do this? That it's all about you know eating this many calories. And I I need to hit that within 100 calories, which is hard. But it's like if your macros are right, what what happens if the quality of the food changes? Like you went from whole foods to not so whole foods, but you're still hitting your macros in that. Like, what's the effect of that? And so we're talking through, and I'm actually learning all the tiny little nuances, like as I'm going. through it as I'm working out as I'm trying to you know count all the calories and get all the macros right ⁓ what to do and learn all the nuances along the way. Normally it would take a lot of research in order to find out to find just the right video that answers that exact question but now you can take your exact situation and actually talk it through with an expert that knows all the stuff. And if it doesn't then it goes and researches for you while you're talking to it. It's amazing.

Ken Freire: That's awesome. All right, my last one is my my phone. ⁓ I don't know, you kind of said you clicked the you did like a voice key, but I put Chat GPT as a widget on my home screen. Did you know you could do that?

Dan Sanchez: ⁓ no. I didn't know you could.

Ken Freire: So like as soon as I c I click on if you guys are watching the video, it's right there, right? It's like the weather and the the widget. And I literally click on the widget and it opens up ChatGPT right away. And it starts a new chat. And the reason I I love this is because you know, during the day you're just like, ⁓ I have a question about this, but you don't want to keep clicking or you I get ⁓ distracted and all of a sudden I'm on Instagram. But I click the widget, I ask Chat GPT my question and then I'm out of my phone. So it saves me a ton of time. The second thing too is With my f my kids, they're starting to use Chat GPT to ask questions. And sometimes I hand them my phone, but I also don't want them to like have control over everything on my phone, you know? ⁓ but like they know, just click the widget and they ask ChatGPT and then they're out. So it's been really helpful to just add that little widget on your phone. ⁓ it saves a whole bunch of little steps and the distractions that I get.

Dan Sanchez: Yeah, yeah. Yeah, yeah, that's cool. ⁓ for our next segment, we're gonna talk about a viral post. I actually saw this on LinkedIn. I forgot to bookmark it and ⁓ had to go find it a video because Jeep made an ad with AI. And it's probably one of the best AI ads that I've seen so far. But I'll play a little clip of it. I'm gonna share my screen and just open up this YouTube video. It's funny, I couldn't find it on YouTube, but like I found another podcast reacting to it and they opened up with it. So I'm like, is this weird that I'm taking a React pot doing a React Podcast on a React podcast? Even though I'm just showing the ad because they had the video clip on YouTube, but Whatever. We're just gonna do it anyway. So let me open it up and play. I'm just gonna kind of scrub through it, but everybody who's watching the video will get the sense and if you're listening on audio, I'll just kind of explain what's happening. But this is a fun I don't know, they kind of made a story out of an inanimate object here. So if you're w not watching, there is essentially a Jeep running down these rugged trails and this tire that was on the side of the road sees it and is now trying to chase it because it's like, ⁓ I want to be on the back of that jeep. It's like a normal tire chasing a Jeep. And it's going through all these obstacles and it keeps getting close, but missing the Jeep. It's going through mountains, it's traveling through freeways. It's just trying to catch up to get on the back of this Jeep that has a empty spot on the back. ⁓ you know where you put the spare tire and it keeps getting sidelined, it keeps getting sidelined until it finds the jeep at the end and it finally the jeep out and haws a tire in the place and it missed out. It falls over exhausted. Almost like it's on its last leg of its life, this random tire trying to roll through. Until the next morning comes, the dad of the Jeep finds the tire. And then turns it into a tire swing right next to the Jeep. And then it finishes with a shot of the tire swing and the Jeep. And of course, like this emotional music is playing, kind of like folk country song playing, and it's tugging at your heart. And it's kinda like I think the tagline of it is like just keep rolling for the Jeep. And I'm like, man. It's almost like Pixar meets Jeep commercial. ⁓ What'd you think, Ken?

Ken Freire: ⁓ you know, it's fascinating that at a quick glance you're like, ⁓ they probably filmed this and then you can start seeing some small little idiosequences, but most people won't notice that.

Dan Sanchez: No. I w I might have guessed, but I think I if I would have saw this out in the wild, I would assume that this wasn't AI. I might have wondered if it was AI, but I don't think I would have caught it. But it w it didn't really matter because you didn't i s essentially it's good enough, the AI is gone enough, and the storytelling is compelling enough that it doesn't really matter if it was AI, because it would have been CGI anyway, of a tire chasing it down and they put like personality into it with the way they shot it and the music overlay.

Ken Freire: Yeah, no, for sure.

Dan Sanchez: And ⁓ people are getting creative.

Ken Freire: Yeah. Well, and I think that that leads to our part of our conversation, but like a major point here is making sure that what you're doing is good enough. Like sometimes with AI you can find like you're trying to get perfect on something and you're running through this rabbit hole of busyness, but not actually getting the work done that you need to get done.

Dan Sanchez: Right. And in fact, this is gonna be the last segment of this podcast. Generally, I've been kind of finishing with the viral post, but now I'm gonna actually get into the meat of what marketing leaders really need to know and where most marketing leaders and their whole teams are getting off and sidetracked by AI. And I think this is gonna be a bigger conversation in the workforce as AI becomes adopted more and more. Because now it's like AI's well past ⁓ the the tipping point, right? It's crossed the chasm, it's into the middle, everybody's using it, and now people are using it more and more. crossed the chasm last year for marketers especially. ⁓ So, well, I'd say 70-80% of marketers that are actually doing marketing are using AI daily now, meaningfully. ⁓ So now we're getting stuck in the next trap. And this actually happened a long time ago in the manufacturing space because manufacturing went through an overhaul on automation in the 80s and it created a new problem. The problem that happened in manufacturing is now happening in in ⁓ office work and ⁓ what do we call it? Thinking work. There's a word for it. I'm blinking on it. You know what I'm talking about. ⁓ information work? No. Knowledge work. That's what it's called. Come on, Ken. You're supposed to help me out. In knowledge work, we're running into the same problem they ran into in the 80s. ⁓ and when they did automation in the 80s and all these manufacturing plants, they started having all these new machines automate things. It's funny, they were very efficient, yet they l had less output. They had less output.

Ken Freire: Knowledge work. Okay. I was like, Where are you trying to get at? The white collar job?

Dan Sanchez: And in marketing terms, that means we're actually not growing revenue. We're feeling like we're flying, yet it's not working quite as well. We're like, what is going wrong? So we're running into the same thing in marketing departments everywhere. And I see it all the time. They're flying with AI. They feel more productive, yet they're working as harder than ever before, and they're getting more done, yet the line's not moving. This is the new problem. And this is my theory, Ken. This is what's happening. Is that we're we're it's just like the manufacturing in the eighties and then they finally came out with a book that's become a freaking amazing business book called The Goal. That addresses it called and he launched this idea. If you haven't read this book, if you're listening to this, you haven't read this book, go and download it now, especially the audio version. Screw the paperback, get the audio version. It's like it's it's almost it in fact it's if you like Patrick Lenciona, it's kind of like fable like that, but it's longer, but it's all story. And none, it's all story, 100%. It's just like one long story, but the lessons are embedded in the story and they're obvious. But he talks about the theory of constraints. Essentially, manufacturing floors have bottlenecks. There are places where all the work of the manufacturing line where things are getting piled up. And if you don't fix that one particular bottleneck, everything else, it doesn't matter how fast or efficient it is. It doesn't matter how much time or attention you throw at it. Unless you fix this one bottleneck where all the parts are lying up and it can't go through this one machine or this one skilled person fast enough, then your your throughput's not gonna go through, which means your resume doesn't go up. That's happening in marketing now. That's happening in a lost a lot of knowledge work now. We're able to be more efficient, but we're not actually fixing Fixing the main problems holding back the flow of revenue. Every marketing department, every business has constraints. But if you don't tack if you don't tackle the constraint and do throw all your effort and energy at fixing the one thing holding things back, then it doesn't matter how efficient you are. It's about fixing one important bottleneck at a time. You know, if you fix one bottleneck, what happens, Ken? There's always another bottleneck.

Ken Freire: There's another one that you find.

Dan Sanchez: So it really becomes a hunt for bottlenecks because you and I, we've had enough work doing sales and marketing together that we always were finding bottlenecks. And that's on I think one of the reasons why we were able to grow the university that we both worked at. is because we were good at finding bottlenecks and talking about the bottlenecks every single week. And we would try different things to put friction at one part to slow down, ease up the bottleneck at one part of the sales funnel, because we're working in higher ed, so it's a freaking long sales funnel. Very complicated process of getting through all the higher ed loopholes and administr like admissions and applications and financial aid and all the all the gov all the government loopholes you have to jump through. So it was fun whack ⁓ all project of finding the bottleneck. ⁓ but I think we got pretty good at it because we found when we can start destroying the bottlenecks, at first it was lead flow, but then when we got too many leads, it became applications. And once we got too many applications, it became, ⁓ we can't complete any of them. People aren't finishing them, you know? And so we started figuring out ways to fix the bottlenecks in different all kinds of different ways. But That's what actually moved the needle. And I think that's what I'm seeing now across the board is that everybody, including myself, this was me. I like April, March, I was chasing all these fancy things we could do with AI, especially when I started getting a hold Codecs and it could do all this work with all these agents. And yet somehow the revenue didn't move. And now that's at Trim Healthy Mama, that's why I'm spending most of my time. It's like what's the three or actually what's the one bottleneck I need to fix right now? And it takes time. Honestly, and sometimes AI is not best suited to fix that. AI is best suited to fix some other problem, but not the bottleneck.

Ken Freire: Yeah. And and the thing is that it can't dis sometimes it can't discern the bottleneck because it doesn't have access to all the information and knows the exact problem, especially when it comes to revenue. It's like I I keep going back to the people I coach constantly, like, what is exactly the problem that you're struggling with right now? And I'll give you an example. Yesterday I was talking to someone and they're like, ⁓ I'm doing this. Like they gave me like five businesses that they're trying to build and run. And I'm like, Okay, well, what's your marketing lead flow engine? And they're like, ⁓ well, I'm trying seven different things. And I'm like, this is why you're not like you're not being successful, is because you need to pick one thing. What's the bottleneck? What's the major, what's your major issue right now? They're like, I'm not ha getting enough leads. Okay, let's talk about that. Let's focus on that. Then we'll talk about how do you how do you convert them into clients? And literally within 20 minutes of the conversation, a light bulb hit and they're like, I'm doing way too much. I just need to focus on on this. I'm like, yep, just do that for the next month. And just like the freedom they felt. All of a sudden they didn't need AI, they didn't need a bunch of stuff. They just needed to go figure out what that bottleneck was. And I think for most marketers, we get confused or just distracted by the shiny object syndrome of what's going on and they don't forget like the key things that they're responsible for.

Dan Sanchez: I actually don't even think it's something AI can help you with right now. Maybe someday when it has enough context, it can help you figure out a bottleneck. But I I think it's one of those things that's uniquely, at least for now, is one of those things that only a human can do, especially a human expert that's been through it before, that can actually go through and analyze. Or like even yourself, like like if you're a marketing leader in the department, you probably know what the bottleneck is. The hard part is it's it might it might be your systems, it might be your team or a team member that's the biggest bottleneck for the whole thing right now. It might be you as a leader, probably one of the hardest bottlenecks to figure out. But oftentimes it's as simple as like you're just trying to approve everything and everything's stacked on your desk and your team's trying to get approval and you're the bottleneck. So we have to fix that. The cool thing is is now we have this amazing tool that once we have the bottleneck identified, AI can at least help in ideation and trying to fix it. And it might be part of the solution to actually on the execution phase. Probably it can't, it probably probably can at least help. But But might not be the entire thing. You might not be able to automate the whole thing with AI just because AI can do all this crazy cool stuff now. It might be having conver that hard conversation. You know, that might be like that one hard conversation you're holding off with your team might be the actual thing holding you back. And AI can help you prep for that conversation like we've talked about. It can probably be an assist to helping you fix the bottleneck. And I think that's the cool thing is b now we have a ton more and a ton of more intelligence to kind of break apart the problem if we truly understand the problem and we have an as an assistant to help us tackle the problem and do it faster. But we actually have to focus that creative energy for the human energy. Energy and creativity and the AI intelligence at the right problem. And I think if we do that systematically as teams, like focus on the bottleneck and give it time, because it always takes freaking time, otherwise it wouldn't be a problem. ⁓ we will actually go farther much faster instead of getting distracted by all this AI crap. I think that's the new thing.

Ken Freire: Yeah, 100%. And that's why that's why like earlier I said, like, I haven't tried Fable. I'm not trying all these things because I'm just trying to focus on can I get this project ⁓ to start generating revenue? Like that's my end goal. Like right now with the blog, I'm like, can I get it to start making some passive income where I'm selling some low ticket items and I'm just looking at the data? That's all. Like that's the goal. I don't need a magic whistle. I don't need an awesome rebrand where things are f coming out, you know, of of the screen. Like, I don't need any of that. I just need it to generate revenue. So all my decisions are based off of that.

Dan Sanchez: I think one I've listened to the audio version. It's the best version of the book, The Goal. I'm gonna buy the paperback just as a prop for the show because I'm gonna be talking about this for a long time, I'm sure. ⁓ and I think I I like I need to print a banner that just says the goal or something like it. Because I'm like, this book is literally the answer to the problem people are running into right now. It's it's a different industry and a totally different time, but it's the exact problem we're running into now, which is bottlenecks. AI has made it so easy to get so much work done and now we're just bottlenecking and not growing because of this one problem. So you'll be hearing a lot about this on the show moving forward because this is the one massive problem all marketers are running into in this age of AI. We have to fix the bottlenecks. We might all grow this at some point, but I'm like, ⁓ might not. This has been even even though it's been around for a long time, manufacturing force still run into this problem all the time. The theory of constraints has lived on. It's one of those theories that I'm like, man, this is this is still relevant in almost every single part of the business. And of course, there are full business bottlenecks that might be bigger than marketing, but generally for a lot of businesses, the number one bottleneck is marketing. And within marketing, there's a specific bottleneck that's probably your biggest problem that you don't even you're trying to tackle a bunch of different things, and there's like one main thing that you should be focusing on. What is the thing? That's where you should be putting your time and energy. So until next time, thanks for listening to the AI-driven marketer.