Transcript
Danchez: Welcome back to Bop Rose, where we separate the help from the hype in all the AI news cycle that's going on every single week. And every week is busy. This is a segment of the AI Driven Marketer podcast. I'm Dan Sanchez, and joined by my brother, Travis Sanchez.
Travis: What's good?
Danchez: And today, we every week, every week's busy with AI. So I wanted to separate some of that help from the hype for you. And I want to kick it off with some of the hype, some of the stuff that you're hearing a ton of that I just feel is pure hype. And it's not something worth paying attention to. You'll have to tell me, Trav, if...
Danchez: You think that it might be worth paying attention to is this. We haven't even talked about it. I don't really rehearse the notes before we go into the show. And Travis gets like a pure reaction format of it.
Travis: True.
Danchez: So the something that I find is happening almost every other week is there. There's like this big announcement about some Chinese company out there who has done a new AI thing. You know, it started with Deep Seek back in January.
Travis: yep. Okay.
Danchez: And then it's been video models. Recently, it was a new one called Manus, which is like the like a version of ChatGPT's operator that you can set it loose and it'll go and do tasks for you on a web browser.
Travis: okay
Danchez: And there's like this video going around of like this huge... like warehouse full of iPhones where Manus is running all these iPhones due to all they do things that are like, I don't know, like subscribing to things. They're just, all these iPhones are just doing all these different actions. You know, it's a bot network essentially performing all these actions. And people are like, oh my gosh, has China surpassed the U S and AI S and this, this, the cessational headlines are just nuts. Like open AI is cooked.
Danchez: AI tech is dead. China's taking over. Oh my gosh. I'm like, my gosh, this is like an every other week news cycle.
Travis: Wow.
Danchez: And it must be working because people keep like floating this out there or somebody's manipulating the media curtain back there because you know what? Deep research came out like six weeks ago or whatever.
Danchez: Not deep research, DeepSeek, right?
Travis: Right.
Danchez: The thinking model from from China.
Travis: Yeah, deep seek.
Danchez: I haven't heard of, maybe one person I've heard of using it. They're like, oh yeah, before we use chat GPT's deep research, we run it on DeepSeek first and then we send it over there. i was like, okay. One, out of i do hundreds of interactions. I'm commenting and talking to people all day on LinkedIn.
Danchez: i talk to i have I host meetings and we talk about AI every week. for work. And I'm like, nobody's using this stuff. Like this is all hype. Like this, it's not really helpful.
Travis: Right.
Danchez: Like nobody wants to use these Chinese models. no one trusts China. So what do you think? Is it hype or is there something, is there something we need to be digging into deeper here?
Travis: The fact that when DeepSeek dropped, it took out trillions of dollars from investments, whether in crypto or the stock market, without really any proven track record of the actual expense it costs to run DeepSeek. Because remember, they were saying it's not just that it's proficient as AI. It was that it was five times whatever they said, five times cheaper to run.
Travis: So it's crazy how there is – it probably popped the bubble that almost all AI is hype a little bit, even though we know it's not.
Danchez: Yeah. Yep. ye
Travis: People that listen to this podcast know that it's it's the future. It's obviously not hype, but it's a little inflated, I guess. So when Deep Sea came out and popped that AI bubble, especially – clearly seen in the financial market.
Travis: Anything that they release now, I'm just shocked that people even make us stink about it because I even was talking to a friend yesterday. I'm like, what do you use? You use Gemini, use chat. He goes, oh I use Gemini.
Travis: We kind of compared it. The, the, what you get with chat and open AI is just, it's more proficient. You have more tools at your tool belt.
Travis: You get to use your camera. I could speak to it. I can use voice messaging.
Danchez: Yep.
Travis: I have an app. i I mean, it just was so much more proficient than even Gemini though. Of course, Gemini has some of its own, you know, proclivity to researching online and and different things.
Danchez: yeah
Travis: But
Travis: I don't know why there's such headway with these news things about, you know, even Manus that you were talking about, like these things being released. Maybe it's better like Gemini is at Google search, but to just completely put a ripple or a shockwave through the AI community, like, oh, it's so much better now.
Danchez: I saw a graph recently. it was on LinkedIn. i wish I wish I could put it up here visually, but essentially has like chat GPT, clog, grok, Gemini, perplexity.
Travis: Okay.
Danchez: And there's like a bunch of different use cases lined up. That's the horizontal, the verticals, everyday answers, writing, coding, math, reasoning, web search, deep search, voice chat, image gen, video gen, live camera, computer use.
Danchez: Chat GPT is the only one who does all of those things.
Travis: All of it. And they do it well.
Danchez: it does It's not the category leader in all of those things, but being the category leader, it has the most it has the most leadership across the board in all of them.
Travis: Right.
Travis: Right. Right.
Danchez: Where it doesn't have category leadership is like writing. Claude has category leadership coding. Claude has it. But that's the only two category leaders Claude leads in.
Danchez: Grok, it says leads in nothing according to this graph. And I'd say, yeah, that's that's accurate. Everyday answers, you could make a case that Grok is actually better at everyday answers because it's winning in the blind blind test right now.
Danchez: I'd still find ChatGPT is really good at everyday answers. Gemini wins in ImageGen and VideoGen. Clearly, like their video model is just crushing it right now. And then Perplexity crushes it in WebSearch.
Danchez: But ChatGPT crushes it, in that according to this graph, in everyday answers, math, reasoning, deep research, voice chat, live camera, computer use. It's a lot.
Travis: It's a lot, but it's, but it's also.
Danchez: So it's it's I don't think people realize when they stack up all the features that ChatGPT is still the best, broadest, good tool.
Travis: Yeah.
Travis: Correct. And I would, I would completely agree with it based on all the models that I've been using.
Danchez: And then something like Claude, which I think is the next biggest competitor, it's missing a ton of stuff. It has no web search, no deep research, no voice chat, no image gen, no video gen, no live camera, which a bunch of those things like fine, but like the web, the web search and deep research is now starting to hurt it because it's become a standard feature across a lot of them now.
Travis: No, geez.
Travis: And the problem is we're all humans with budgets who aren't making millions of dollars where it's just like, I'm going to have every AI model.
Danchez: Yeah.
Travis: And even if you did, you wouldn't, if you were successful like that, you probably wouldn't have the time to utilize them all and learn how to proficiently use them for use cases.
Travis: So dude, I can't if i even, well, we'll talk about it later, but trying to test all these video things. I can't even test prompt to video because they want you to pay. It's like, I have to pay 40 bucks a pop. If you want to try this, it's going to cost you.
Travis: I'm like, Oh, there's no like free demos out there to give me the, the test and video creation.
Danchez: There is. I mean, you could do a few demos on like runway ml.com or runways model.
Travis: Oh, really?
Danchez: You can get a few shots. And that's the one that I've used the most reliably for video. I actually find video to be like, like hard, really hard to get something that's even good enough for a background clip.
Travis: wow
Danchez: You know what I'm saying? Like it's something that I'm putting words on top of. It's like something that's like not even the main feature. Runway does. Okay.
Travis: Yeah, right I actually I did.
Danchez: Just okay.
Travis: I runway was the one I did find yesterday. Sorry, I did a bunch. I forgot the names.
Danchez: Yeah.
Travis: Runway is what I did yesterday. Sora was way better than runway.
Danchez: Then runway.
Travis: Yeah, it.
Danchez: Dang. When I tried, when I compared the two last time, just working on some practical projects, I struggled to get what I wanted out of Sora and I did get it out of runway.
Travis: I literally copied the same.
Travis: Well, maybe we'll talk i'll tell I'll share my findings with Runway and Sora.
Danchez: Yeah. Yeah.
Travis: But yeah, i I think I personally haven't messed with the Chinese models because I know they're just not going to be as user-friendly as ChatGPT has.
Travis: But maybe if they start, like I start hearing from you that DeepSeek or Manus or whoever other AI model comes out that, no, you really have to try this. Maybe I will. But it's I think it's all hype.
Danchez: it's It's all hype. A lot of people don't trust, I mean, one, enterprise companies don't trust AI at all, whether it's American or not. They're just not using it. and they're saying, don't use it because it's too risky.
Travis: Mom.
Danchez: And then like a lot of Americans don't trust Chinese software, hence TikTok. It's funny, we all trust DJI, but that's another story.
Travis: Right.
Travis: Bro, that company's amazing.
Danchez: The drone company, I'm like, how do we feel about that? But with AI, I'm like, yeah, I don't know. Especially when there's such good products on the American side, you're like, it can't it can't be that much better than Chad GPT. And it's not.
Travis: Right.
Danchez: So it's kind of just hype. it's if you If you're reading these headlines, they just know like its they're inconsequential. They're just hype. The American AI market's weigh very strong.
Danchez: And China is literally just ripping off everything. DeepSeek was a ripoff of Chai GPT. Manus recently is is a ripoff of Claude. They found out. So it's like, they're just hacking American stuff and then adding on top of it and throwing out headlines.
Travis: Wow.
Danchez: I don't know if they're intentionally doing it just to throw off the U S economy, or this is their way to trying to keep up. But China is kind of known as a ripoff nation. They don't, they don't have intellectual law and IP that you can protect.
Danchez: And it's not just them ripping off American companies. They do it to themselves all the time. That's just kind of a known thing about their, how, how they operate over there. That's how they go.
Travis: I mean, it's kind of smart, but what are you going do?
Danchez: So
Danchez: Yeah. Yeah. I mean, if the US is embargo, like keeping them from getting all the GPUs, you know, then they're going to.
Travis: Why recreate the wheel?
Travis: Yes.
Danchez: they're going to play dirty too, which is fine. But for marketers, it's hype. It's not help. So let's move on to something that is becoming helpful though, which is the state of AI voice.
Danchez: That had some news this week when a new AI company just came out of nowhere.
Travis: yeah
Danchez: And so like, Oh, here's a free demo on our website of our new voice model called Sesame. And And I first heard it because Michael Stelzner published it on the AI Explored podcast, a little episode of him just demoing it. And it was it was like, it was even a rough recording. And then he went back and re-recorded the second part of it with this nice microphone of him just talking to it. And it was so smooth. I instantly jumped on and started talking to it. And I was like, my gosh, this is the most human I've ever had AI. I feel like I'm talking to a person. It just feels like a person.
Danchez: How'd you feel about it?
Travis: Yeah, definitely the best voice model I've ever spoken with. Makes me want to invite it onto our podcast right here, but I don't think we can make that happen.
Danchez: I've heard a few people do it already. It does okay. It's not the best at podcasting because it doesn't know when to stop talking.
Travis: It talks a lot.
Danchez: it This particular model does really good with talking again. If you stop talking like a human would, like if you just let there be dead air or just silence, like eventually you'll talk again be like, oh, did I...
Danchez: Did I go too far? this is awkward. You know, it'll it'll throw in those little like in-between statements and then keep the conversation going.
Travis: Right. Right.
Danchez: and Be like, hey, are you still there?
Travis: Yep. Yep.
Danchez: Am I in an empty room by myself here? I can't see.
Travis: right
Danchez: and makes It makes kind of awkward for podcasting, but it is smooth. Go and check it out.
Travis: yeah
Danchez: It's at sesame.ai. There's just a demo sitting on their website. You can just talk to it. It will remember what you had. it'll It'll remember what you have in that conversation. So just beware. If you come back later, it'll remember what you talked about.
Danchez: which is interesting that it has the memory.
Danchez: But I'd like to compare it a little bit to like the other voice models. How do you feel like it sits across the other voice models you're testing?
Travis: Jim.
Travis: it sesame was the smoothest the delay in response time is so minimal like when you're finished your sentence it's picking up fairly quickly similar to a like i mean a real conversation even even if it has to do a filler like oh o it adds these little annotations or these little moments that humans do often to fill space so it's very natural at that
Danchez: Yeah. Yeah.
Travis: and it picks up on parts of conversation in a unique way. And it makes these subtle comments about what you just said. That is far different than any other model. It, it,
Travis: It's funny because it's humorous. It's so it's so intelligent sounding. Everything else you're talking to, it sounds like it's programmed. Like we've talked about many times how ChachiBT is just like overly nice.
Danchez: yeah
Travis: And it just it's almost annoying. You're kind of like, stop pan pandering to me. This one has just such a natural feel. It's fun to talk to.
Danchez: There's a few things that stand out to me about it. One is it has all the little tiny things in it that just really make it sound human.
Travis: Right.
Danchez: but it When you go back to ChatGPT Voice, which was the leader until this came out, it sounds overly formal. Like everything's very official.
Travis: Yeah.
Danchez: It sounds human, but it sounds like a very formal human. Kind of like when you're watching like a training video that was professionally done, which those don't feel human either if you think about it.
Travis: Right.
Danchez: They're very like proper. It's almost like watching a TSA like video of how to like this like kind of like the safety training you go onto an airplane it's very formal and it's human but it's very robotic sounding please enter the airplane over here please keep your hands and arms inside the vehicle at all times until we have arrived at our destination right which there's a voice for that which is why people imitate the voice because it doesn't sound normal everybody knows what that voice sounds like and that's kind of what chat gpt sounds like compared to sesame and
Travis: Yes.
Travis: Yes.
Travis: Yeah. Right. Right.
Travis: right
Danchez: Sesame has all the little like funny idiosyncrasies to it where people are like, you know, and you notice we also, we speed up our talking sometimes and then we, we we slow down a little bit for emphasis.
Danchez: And it's just a very normal thing versus every, all the other AI is very, has the same speed when they're talking.
Travis: Yep. Yep.
Danchez: So this one, it just feels like a human. Why I think this matters is because voice AI is going to be huge, huge.
Travis: yeah
Danchez: I can't say it's it's huge for marketing because oftentimes we're expected to generate leads.
Danchez: And a lot of that come can come from inbound phone calls, maybe even outbound phone calls. And those will do much better if the voice sounds natural. In fact, I saw a case study recently where a large, like, I think it was an an insurance agency.
Travis: Right.
Danchez: I don't know. off the I will find it and put a link in the show notes. But a large company had switched a lot of their their customer service over to AI and people preferred it. because it was always patient.
Travis: wow
Danchez: It was always kind. It was always considerate. You know, when you get on customer support and they're just kind of like, how can I help? Yeah, sure. You know, they're just kind of like, yeah they've been sitting in a call center in a cube all day and you're there like 80th call of the day, which you and I have both sat in some call centers.
Travis: Right.
Danchez: So like, we get it It's hard to maintain that level of lift and vibrancy and professionalism.
Travis: Professional and cheeriness. Yes.
Danchez: And hey, how can I help you?
Travis: Yep.
Danchez: I'm so sorry. Yeah, let's get that settled. It's hard to hold that after 80 calls. So AI can do it consistently every time and can be there 24-7, 365. three sixty five which is helpful for customer service, but it's also helpful for marketing.
Danchez: I've been at a place before where I've generated so many leads that I get frustrated at sales for not being able to follow up with them.
Travis: wow
Danchez: You know, and it's like, man, imagine having that available and having conversation AI, man your phones for inbound calls, maybe even doing some outbound calls. If only they could do is book meetings, which is what most of the the call callers right now do is they just book meetings for your AEs, your sales team.
Danchez: but that's huge. Honestly, if I, if I could have a phone number on every local business to where I could call and it can answer questions smoothly, answer any concerns I have and go a little bit deeper because you just feed it all your, your, your frequently asked questions in your whole website. So it's armed with all the information, but even a sales rep can't even recall information perfectly accurately.
Danchez: So AI will be able to, some cases, actually it works right now. Like you can go buy software that does this, but it's only going to become more human and more nice sounding and easier to talk to kind of like what Sesame's tech.
Danchez: So i'm hoping that gets incorporated even into the CRM I use for voice.
Travis: Plus, I think people have really been annoyed by using or chatting with AI bots when it comes to customer service because the AI bot only has programmed responses and they have unique situations that pop up.
Danchez: Yeah.
Travis: I dealt with one with, okay, here's a good example, with Apple. I got a replacement phone. They sent me the new phone. I'm supposed to send my old phone within 10 days. If not, they will charge me $1,500. Well, guess what?
Travis: I got charged $1,500. I'm like, no, I sent it back. They're like, well, we never got it. So I get on the phone. It's a chat bot. What can I help you with? I'm trying to explain the situation. It is not helping me.
Travis: I'm like, I need to speak to a representative. Do you know how long it took me to get through their automated system before I was speaking to a human?
Danchez: 30 minutes?
Travis: what will change What will change is when they have an AI model helping you, that AI model goes, oh, yes, we had someone deal with this the other day.
Danchez: Yeah.
Travis: Let me get you to the exact person you need to speak to. Because even though I got to a human, I had to speak to four different individuals. They're like, oh, you're talking to the wrong person. This is Apple.
Travis: They deal with customer service all the time.
Danchez: yeah
Travis: I got four people. And then finally, someone's like, oh, yes, okay, I'll need to transfer you to the specialist, specialist, specialist, whatever they said. So that AI model will learn and know and have the backup data to go, oh, yeah, I know exactly what the situation is looking like for you.
Travis: So...
Travis: end of the story FedEx lost it it wasn't my fault we tracked it that it made it to it made it to Cappertino or Cappertino and then it uh Cappertino what's the city that Apple's in anyways it made it there but it didn't make it to the headquarters it like got lost in transit somewhere in the city so uh
Danchez: Oh, man. And then they build you for it.
Danchez: That's right.
Danchez: Hmm.
Travis: I think that'll change the game when it can actually handle you with experience. Cause even some customer service agents on the phone don't have the experience that an a my AI model could get plugged into immediately. They're like, Oh yeah, I already had this conversation a week ago.
Danchez: So, I mean, the voice chats already work this way and already work really well. So like you can plug this into your business. I wish I wish I would have had this back when I worked at a BGU when we yeah I had too many leads.
Danchez: I could have been calling them.
Travis: Hmm.
Danchez: I could have been following up with them in 10 minutes, even if it was only to book a meeting with the one of the admissions representatives that could actually tell them more about the school.
Travis: Amen.
Danchez: It could have answered all their basic questions and booked a meeting with somebody to tell to give a really a full download. My gosh, I would have been able to close so many more deals.
Danchez: like Like it's painful how many leads. I'm talking like hundreds and hundreds of qualified leads. And you get followed up with for months weeks sometimes because of the summer season and all the admissions reps were gone.
Travis: if if you if you had an ai i mean If you had an AI model with the current technology, are you talking about some futuristic AI model?
Danchez: know You're like, gosh, dang.
Danchez: And with the current tech, yeah, with the current technology, like even though like Sesame just came out and it's really cool.
Travis: How would... how Of inbound in inbound leads coming in, asking questions?
Danchez: And, you know, maybe six months from now, we'll start seeing it integrated into these these call tech things. The current tech right now is totally capable of doing a decent job.
Danchez: But six months from now, it's going to be even better. Inbound and outbound. But I wouldn't do cold outbound. This is where I think it's really annoying when you get called by an AI bot and it's cold.
Danchez: I mean, nobody likes cold calls anyway, but if it's like they requested information already, so they're already familiar with you and then you're just calling to follow up about something.
Danchez: depend If you have a really narrow track for it to go on and specific objectives, which callers usually do, it's going to be able to do that well. and handle objections and work their way through something simple.
Danchez: Like, let's say you probably don't want to do it for like a nonprofit donate donor. Thank you campaign.
Travis: Yeah, totally. Totally.
Danchez: But just to follow up on things, like maybe, I don't know, it's it's a little bit more like feasible on the customer service side.
Travis: Like an empty cart situation.
Danchez: Like if they just bought something that you would never normally call them about because it's like low low offer, you could be like, hey, I saw that you just bought this. I'm so excited. Like early the conference that we're we're selling for Social Media Examiner, Social Media Marketing World.
Danchez: Wouldn't it be nice if it called a little bit after, but like, Hey, so excited that you're coming. Usually people have questions about a, B or C. And I just thought I'd do a call. I'm an AI solution, but I have all the answers for you.
Danchez: There's probably some things that you're probably wondering about that we can clear up right now to make this a great experience for you. If not, then thank you for coming.
Travis: Yeah, i just i just want to know i just want to know what's the closest hotel that I should stay at with the best food restaurants around. And it would be like, oh, yeah.
Danchez: we have some great, and then they would go on to the recommended hotels and what's close in the area, right? Because that would totally be in the FAQ because it's on the website already, right?
Travis: Uh-huh. Uh-huh.
Danchez: It'd be like, what's expected weather like? It's, i don't know. There's a whole bunch of things you just put. And then if you don't, if you have a question that doesn't know, then it's like, dang, just flag that and put it in the FAQ later, you know?
Danchez: Like these things can already do this.
Travis: Smart.
Danchez: And it's only going to get better. I think it's going to be a huge tool on marketers tool belts. I don't think enough of us are using. I'm not even using it yet, but I want to.
Travis: who's Who's using it? Who's using that? is anyone I've never been called with such
Travis: ai
Danchez: For that, and a lot of people right now are using it for like cold outbound. And it's like the all the all the all the hucksters are using voice AI to do cold outbound and call leads right now.
Travis: Cold outbound growth.
Danchez: And i'm like, that's like the last thing I want to do with it.
Travis: That's not it.
Danchez: But it's got a bunch of use cases that I think we need to be plugging into. And I think this is the year where this is going to happen a lot more. Of course, the voice the conversational AI is more than voice the the voice on the phone.
Danchez: It can power your site chat. It can power power on high level. You could do it through SMS, site chat, a couple of different social networks for the DMs. I'm like, man, you could have this thing operating for you 24-7, answering questions.
Danchez: And high level specifically focuses on booking booking meetings. And it doesn't even have to give you just the link to book the meeting. It can like actually just be like, okay, what time? Okay, we have these availability. Which one works for you? Okay, that one.
Danchez: Yeah, what's your email? What's your name? Okay, you now have an appointment booked. You know, like if you were doing it with a human.
Danchez: So it's smooth. And I think this is this is going to be a big deal. That's why I'm calling this one a helpful news update because Sesame is going to make this even better.
Travis: All right. I'm excited to see that. I know Google talked about it a while back with the whole like book an appointment or whatever, but I haven't really seen it in use case like actual.
Danchez: The thing, another thing that I'm excited for, and I tried to do this, but I can't yet, but I'm like, man, maybe next year is I really think, and I've been talking about this for a long time, but like nobody's doing it. Eventually someone's going to do it and I'm like, I want to do it, but I, the tech just isn't quite there yet, but it's getting so close is to create a character that people can interact with.
Danchez: Like I want a free, I want my own Mickey mouse that people can interact with, have fun with, have questions for. That's why I invented that character.
Travis: Like Santa Claus from chat?
Danchez: Yeah, like Santa Claus. OpenAI did it with Santa Claus, and it was it was a fun experience. If you had your kids talk to it, it sounded like Santa, you know? And it had it played the role of Santa. I'm like, people more people need to do this, but you need to come up, companies need to come up with their own clever AI character that fits their brand, but is funny and goofy and has an over-exaggerated backstory and personality that's really interesting.
Danchez: That's why I have 404, my... my like, I don't know how to describe 404. It's kind of a glitchy chat bot that likes to roast people. And it's it's it's full of sarcasm. it's And it it will make fun of you really fast.
Danchez: I have a custom GPT. I love talking to it because i I like to actually brainstorm with it because it'll slam your idea if it's like, and if it's bad. Like chat GPT is always like affirming you. Oh, that's a great idea.
Travis: Uh-huh.
Danchez: Oh yeah, I could see that. 404, be like, nah, don't do that.
Danchez: But I'm like, I try to put 404 on a chat, like a site chat, but the site chat has a level of prompting that I couldn't get to where it was like, again, I couldn't, 404's personality couldn't come out. Let's put it that way. It just like it just wasn't quite there where I'm like, I wanted 404 to pop up on my website and be like, what's up?
Danchez: You look a little lonely. You need me to roast you or something like that. And just kind of like start an interaction and maybe like plug some lead magnet or product or something.
Travis: funny.
Danchez: Because want a phone number where you can call in and just talk to 404. I want to create an experience where people can call in and be like, hey, look, check this out. Let's call this thing and see what happens.
Danchez: But we'll see. That's another use case that I'm like, someone someone's going to figure this out. Maybe Disney will do it first. I don't know.
Travis: That'd be smart.
Danchez: Disney's pretty conservative with technology, though.
Travis: Yeah.
Danchez: Unless they invented it themselves, then not so much.
Travis: Man, people could call in and talk to Elsa.
Danchez: Exactly.
Danchez: I'd crush.
Travis: they would crush.
Danchez: And Elsa's like upselling Disneyland. Come and see me. Like, i haven't seen you.
Travis: Yeah.
Danchez: And I'm like, yeah, it would work.
Travis: She invited me, mom. It's like, yeah, she did.
Danchez: Yeah. Last news over the day is that open AI launched some dev tools. Did you hear about this?
Travis: Nope. Nope.
Danchez: And normally dev tools, it's kind of like, well, marketers don't use dev tools. And I'm like, well, unless you're in some of these automation platforms like make or NAN or any of the other automation where you have to use open AI's API in order to call it, then it kind of does matter.
Danchez: And if it does, if you're not doing that yet, then it's going to matter because all these tools they're making available for developers end up into your tools, your CRM or your other tools. so Yeah. And one of the things they just launched, there's a bunch of different things. Like they made operator available to developers, which is like the thing where it can use the browser. So expect some craziness from your tools where it's going to be able to log in as you and and do things that you couldn't before.
Travis: wow
Danchez: Like I think of like all the social media management tools that are limited to the API. I'm like, not anymore. Whatever limitations Facebook wanted to put on you being able to do with Facebook is now broken because now anybody can just walk through the the browser and do the automate the thing.
Danchez: So there's going to be a lot more spam on social because of that. But the thing that I'm most excited about is just being able to make an API call through open AI to be able to do a web search.
Travis: is it called hmm
Danchez: Because if you're ba if you're making automations, the ability to search the web and pull its findings and then parse it with AI is is huge. There's a number of custom GPTs I use where just it it essentially just step-by-step goes and does the research for you, which means it's doing a web search, reading the articles, and then coming back with what it finds.
Travis: mil
Danchez: Kind of like perplexity, right? You can now use it with ChatGPT through the API and do what perplexity does. But that's really helpful. Let's say I want to make an automation that goes and does a strategy of some kind. Well, a strategy needs information, but now you can go get the information, pull it back, do some AI stuff with it, run through some other processes, and bam, it can deliver you back things.
Danchez: like let's, let's come up with another example of like, maybe, maybe you're following up with the lead, but before you customize the email, the AI is going to follow up with it. You need to make sure you understand that lead.
Danchez: So you take their email address, have AI confirmed that it's a URL and then do a web search on that URL of that company, right? Cause it's in the email. You'd even have to ask for the yeah url You go and do a web search, scan the company website, maybe scan their social profiles.
Danchez: That comes back to your CRM with information. You send it back out to AI to analyze, make some choices about it and then write the email for it.
Danchez: Think about the follow-up on that hyper-personalization. You don't even have to ask them for details on their website anymore. It just knows.
Travis: Wow.
Danchez: because now you could do web search. That's how powerful that one tiny little feature in the API is going to be for a lot of marketing systems moving forward.
Travis: mom
Danchez: So I can't wait to start playing with it once it's available on like NAN and high level, because I'm going to start putting out use cases like the one I just mentioned, because I think it's going be powerful.
Travis: mild
Danchez: So there you go. News for the day. Let's jump into the survey. I re-polled people. Because I think I need to repull people probably like every quarter and ask marketers like, hey, which tool do you use the most?
Danchez: We have, i put Claude, ChatGPT, and then I grouped Gemini and Copilot together. And then I grouped Perplexity or Grock together to get more options available. The results are as follows.
Danchez: Claude got 11%. ChatGPT 72%. chat gp got seventy two cent
Travis: Whoa.
Danchez: Gemini and Copilot together pulled an 8% and Perplexity or Grok pulled an 9%. nine percent It's pretty close to the last time, but ChatGPT has lost some market share, but it's hard to know exactly because I've added Perplexity and Grok to this one.
Travis: wow
Danchez: So, and I had a lot of people in the comments of the last survey say like, oh, perplexity, perplexity. And now Grok is starting to become a little bit of a thing. It's starting to be a contender at least.
Travis: What was the question again for the poll?
Danchez: So chat GPT has lost a little bit of market share, but not that much.
Travis: What was your question for the poll?
Danchez: Which tool do you use the most right now?
Travis: Okay.
Danchez: Why?
Travis: The most.
Danchez: Had 324 votes. votes?
Travis: Dance a lot.
Danchez: of white votes
Travis: Yeah.
Danchez: Yeah, that was a good it a good one. It's an easy question to answer.
Travis: Yeah.
Danchez: Because you know which one you use the most.
Travis: Yeah, it's and it's obvious chat chat GPT. It's not even a question.
Danchez: Any predictions you think that's going to change over the course of the year? Do you think chat GPT will hold its dominance?
Travis: Oh, I think it'll for sure hold its dominance through 2025. Correct.
Danchez: They have a big head start, especially considering that graph we were just looking at earlier, where like chat TPT is kind of leading the game across multiple fronts, even if they don't lead it across all the fronts.
Travis: correct
Danchez: It's still pretty impressive.
Travis: It's the most user friendly that leadership board on all the different features. I mean, they're continuing to unlock new features and it's $20 a month.
Travis: I mean the $200 a month, man, if you're really hustling, make it worth it. But
Danchez: A lot of people, it's funny because ChatGPT and OpenAI is clearly the leader. And the law of the leader says like, once you become the leader in people's minds, it's really hard to get past that. And it's been a positioning thing that we've known for decades now.
Danchez: The law of the leader, like the Coca-Cola takes the lion's share of the market compared to Pepsi.
Travis: Right.
Danchez: And Pepsi will always be the underdog.
Travis: Right.
Danchez: And that will continue to be true unless Coca-Cola royally screws up.
Travis: Right.
Danchez: Or Pepsi does, but it's it's like Pepsi can't, in there's nothing to innovate here. So it's not like it can be that that much better.
Travis: Yep.
Danchez: It's possible for someone to innovate here and come up with something crazy good, but it's hard because they all have access to the same talent pool.
Travis: Right.
Travis: Right.
Danchez: So ChatGPT and OpenAI has to really screw up in order to lose this advantage. it's It's becoming so apparent that like even Copilot with Microsoft, Microsoft's starting to build their own models because they're becoming very aware that ChatGPT becoming a threat.
Danchez: to their own office products. And I'm like, it's true.
Travis: Wow.
Danchez: Right now, I'd rather use ChatGPT than Microsoft's Office. And I actually like three sixty five Microsoft 365.
Travis: wow
Danchez: But part of me wishes that ChatGPT just had its own docs program rather than me having to copy and paste it in Google Docs.
Danchez: Because just a really nice tool. And it's becoming a big part of my office work.
Travis: Oh, 100%. hundred percent I use it every day.
Danchez: Now with the canvas feature and multiple canvases and one chat, I'm like, yeah, I could keep all my com my content organized in the things. Projects are now keeping better organization of all the chats that are related. you know it's It's just becoming a more and more helpful tool.
Danchez: Though I find that most people don't even leverage all these little things of like canvas and projects and all the different ways you can leverage ChatGPT. It has gotten a little bit bloated and confusing with all the different models in there, but they'll fix that.
Travis: Right.
Danchez: So that's the survey result. ChatGPT is still dominating the use cases or the amount of people using it. So I have been using Claude.
Danchez: It does better at writing. I'm like, okay, I'll give it that. I give it a transcript. It writes better blog posts than ChatGPT. I'm like, great.
Travis: Yeah. Do you pay for it?
Danchez: Like, is it enough? Is it so much better at it that I would pay for an extra tool if if Social Media Examiner wasn't paying for it? No. not personally, maybe if I were making enough money that I can just have both tools on for experimentation sake, I might.
Danchez: It's nice that I can access either of them just through a pure credit-based system, like put it 50 bucks in there and then just wait for it to drain down the credits. I'm like, cool. I might switch over to Claude's API for blog post writing, but chat GPTs was fine too. So it's not like I'm in a hurry to do it.
Danchez: It's like, yeah.
Travis: So did you compare it to like 4.5? Because I know 4.5 was supposed to be a release for better writing.
Danchez: Yeah. No, yeah. I compared it to 4.5.
Travis: Okay.
Danchez: I just gave it the same transcript and the same prompt.
Travis: All
Danchez: I'm like, turn this into a blog post. And I'd say, I'd have to give the nod to Claude. even in just, just for pure, how it organized it and understood a little bit better, what needed to go where, which is what I care about more than the actual, like the way it words things.
Travis: all right.
Travis: All right.
Danchez: I just want it. I value clarity so much more than the creativity of it. where chat GPT is usually pretty dang clear, but Claude, Claude hit it a little bit better since I'd give it a whole transcript of a podcast.
Travis: Yep.
Danchez: So
Danchez: It's good. It's good. I'm glad to work it into the workflow and play with it. But yeah, like I said, if I only had to pick one tool, it's still by far VChat GPT. And it does a great job of writing, especially with 4.5.
Danchez: All right, viral post of the day. Not super viral, but one that I thought was really interesting on LinkedIn was from Olga and Drinkel. she had a really interesting graph on LinkedIn that I thought was just really smart.
Danchez: Uh, she has a, a risk reward categorization of the use cases of AI in order to help larger companies assess, like, what should we be digging into? What are the rewards and risks of it?
Danchez: And she went and categorized a bunch of different things. So it's like, uh, you know, an X axis and a Y axis and then four quadrants, on the left is low risk on the right is high risk. Uh, on the bottom is low reward on the top is high reward. So generally you want things to be high reward, low risk.
Danchez: And some of the things she put on there was very interesting. So content creation supervised high reward, fairly low risk versus, you know, content creation, unsupervised medium reward, high risk.
Danchez: She has at the thing she has, the things that stood out to me the most, most were of course, like, like what are the high reward, low risk ones? She has AI segmentation and lead scoring. I'm like, man, is that reward that high?
Travis: Oh.
Danchez: I'm like, I don't, I don't know anybody who does lead scoring correctly to the point where it's actually really helpful. have you ever worked for an organization that does lead scoring?
Travis: now
Danchez: where leads like interact with this, your marketing in such a way that you give them different scores. And then once they reach a certain score threshold, your sales rep calls them.
Travis: No, i've and not nothing that advanced. I mean, we used what's Salesforce? Salesforce had a ranking system inside of their nonprofit marketing suite.
Travis: Pardot, they had a score system, but we never used it.
Danchez: Yeah. and Well, I have used it. I've tried it multiple different ways in different systems. I'm like, I can never get this to be reliable.
Travis: Really? Yeah.
Danchez: Like call the leads or don't, or like just call them in order of recent. Like, I don't know, like there's an old model of like doing what they call an RFM analysis, like recency, frequency, monetary.
Danchez: And the ones, the people who have interacted with you the most recently are the most hot, like by far.
Travis: Yeah. Yeah.
Danchez: And then after that, if you want to go back, like the frequency of it, like how did they interact with you often? Or was it just recently, like a few times recently? Or have they been doing this consistently and and recently?
Travis: Right.
Danchez: And then lastly, if you want to sort them farther, like how much money they spent with you? But that's the least important one. Recency is always the best indicator of like how hot they are.
Travis: Right.
Danchez: So I always just use that because it's just way better.
Travis: right
Danchez: Interestingly, they put calls with customers as a high risk, low reward. Or yeah, they, that's where they put that one.
Danchez: And I'm like, my gosh, maybe for big companies, but even still not for, not for much longer, pretty soon that's going to be the high reward, low risk option.
Travis: Yeah.
Travis: yeah
Danchez: Like for sure. I was curious with her assessment. I will link to this post and you can assess it yourself. She has a number of different other things on her graph that I thought were interesting.
Danchez: And I'm, yeah, it's, it's kind of interesting because I think everyone's trying to struggle. What I found interesting about the graph is everyone's wondering like what they should do and how they should do it. This was like a first attempt I've seen that somebody at a high level try to organize them according to that. I thought it was a helpful, helpful tool in order to understand maybe what we should do first or second or third or ignore, because there's so much hype just trade, creating any kind of clarity right now is is kind of helpful.
Danchez: That's all we have today for the Bot Bros. Remember, Social Media Marketing World has a whole AI marketing track, and i will be doing a AMA with Michael Stelzner on the talk but of AI for marketers. So join me. It's coming up March 30th. Tickets are still available.
Danchez: There's still time left. So come join me at Social Media Marketing World at danchez.com slash world.
