TEA 251: ChatGPT vs Claude vs Gemini (Which AI Is Best For Business?)

ChatGPT, Claude, and Gemini all cost about $20 a month, but they are not built for the same jobs, and being loyal to just one of them is quietly costing you money. We pay for all three, and each one has a completely different role inside our business.

In this episode, Josh and Dylan break down exactly what each AI model is actually best at, how they use all three together every day, and which one deserves to be your business workhorse.

Inside this episode:

• The marble race theory of AI: why the leader keeps changing and why model loyalty is a costly mistake
• What ChatGPT is genuinely the best in the world at right now (and the brainstorm-then-handoff workflow we run between it and Claude)
• Why we still call Claude "the business AI" for content, automations, skills, and interactive dashboards
• How Isaac, our client experience director, used Claude Code to build in one single day what would have cost $25,000 in dev work
• Claude Fable 5 writing four emails in Josh's exact brand voice with zero edits (and why it costs $7.70 per million tokens, the most expensive model in the world)
• The nerd stats: the fastest model responds in 0.66 seconds, the slowest takes nearly two minutes, and the smartest two models by a long shot
• The one thing Gemini does that ChatGPT and Claude literally cannot (upload up to 2GB of video and have it analyzed), and why we call it the Peggy of the AI race
• The expensive mistake ecom brands keep making by letting AI make decisions on their Meta ads (we covered the full story in episode 247)

It's not about which AI is the best. It's about knowing which AI is best for the job in front of you, and for about $60 a month you get three specialists that would cost a fortune to hire.

Loved this episode? Drop us a rating because we're going for #1 ecommerce podcast in the world and every single rating moves the needle.

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

► Visit Our Website For Training and Resources

► Leave Us An Honest Rating, Email An Image Of Your Rating To team@theecommercealley.com, We'll Send You A $10 Amazon Gift Card As An Appreciation Gift!

► Learn About Our Mentorship Program For Ecom Brands Making Over $10k/month

► Checkout Our Software, Breezeway - Never Second-Guess Your Meta Ads Again

►  Follow Josh on social media:
YouTube | Instagram | Facebook | TikTok | 

Transcript

SPEAKER_01 0:00

What is the difference between Chat GPT, Claude, and Gemini? What should you be using inside your business? Now I would say that most of us, if you're like us, you picked up Chat GPT years ago and you never really re-evaluated other options. Uh, or maybe recently you have, you've seen a lot of things coming out about Claude, and maybe you've evaluated them and maybe you've entirely jumped shipped to one or you're only using one. The question is, what is the best and what should you actually be using? And so we're gonna we're gonna do a walkthrough of this on all three of these large language models, or LLM is the short acronym uh for that, because they change a lot. And if you look at it from a price standpoint, they're all pretty cheap. I mean, they're $20 a month. I think that's the average. Is that right, Dylan? Something like that? $20 a month? Yeah, I believe so. 20 to 25 of them. If you're on team plans, it's a little bit more usually. But we actually use all three of them, and we have used all three of them. So we're gonna give you our take on the benefits and what you might want to start using because uh my my perspective, and I told Dylan this yesterday, is I am I am I am in an open relationship and LLM curious uh for for for all three of these models that we that we use. And I think that it's important because I think that loyalty of one uh can actually be very costly because they all kind of work a little bit different. And so we want to talk through some practical, real-life business use cases of these and how we're kind of using them. And then hopefully you could just take these and maybe apply them to your business as well.

SPEAKER_00 1:36

You have to think about it from like a business perspective for these for these LLMs, they're trying to differentiate themselves from the other models. So they will kind of find the holes or find the things that they think they can do really well and put a lot of time and attention into optimizing their model to do that thing. So in this episode, we're gonna go through basically what each LLM has decided to put their time and attention into optimizing for, because you can tell the difference. Like you really can. Each one has has strengths and weaknesses. So you really have to be you a lot of using AI is knowing which AI to use for a specific task, which I hopefully this episode will make a little bit easier for you.

SPEAKER_01 2:20

Yeah, and and here's how I think of it. I was actually so at the gym, I don't know why, but I'm on this stair stepper, and there are all these TVs around us at the gym. And I stair step for 15 to 20 minutes. If you've never have you ever done a stair stepper before, yeah. Oh, it's like you will pour sweat, your heart rate will get up like 140, 150 for me within about 10 minutes. And so it's really good for getting your heart rate up versus running takes forever, it feels like. So I stair step almost every day that I'm at the gym before I lift. And for some reason, I just time it perfectly that at 5 45 a.m. almost every time, there's on this screen in front of me, there's the they have racing marbles. No idea why this is so like it's like literally these complex marble racetracks where people will get at the top and they drop their colored marbles. I'm pretty sure that's totally brain rot. But I'm just like stair-stepping and I'm watching these marbles, and I'm like, oh, which one's gonna take the lead? Which one's gonna win? And what happens is throughout these, like these, you know, little courses or whatever that they're like, they drop them and it's all downhill the whole time. I'm just watching red marble takes the lead, blue marble takes a lead, and then it goes around a corner, and then the yellow one swings out and takes the lead. And that's literally how I feel about the LLMs right now. Literally, because it's like uh Claude at the beginning of the year comes in like a freaking wrecking ball in the business world at least, and it just d punts Gemini and Chat GPT like to the side and just takes a lead by a landslide. And so we moved to Claude. We actually did an episode. Do you know what episode number that was? I could go look it up while I while I talk through it here. So we did an episode on how we abandoned ChatGPT and we moved to Claude. Now that was a really good headline. Uh we didn't entirely abandon ChatGPT, but we moved like 99% of our attention to Claude.

SPEAKER_00 4:10

Episode 237 from April 6th.

SPEAKER_01 4:13

Go listen to that. We dive really deep into why we did it and the benefits of it. But we moved to Claude because we, and we still believe this. Claude is like it is the business AI tool. It is for automation for so much more. It's going to save you so much time and so much money. And so we still believe Claude is really, really good. But we still had ChatGPT because we pay annually for the team. So it's kind of sitting over here. And then what we learned in Claude was, oh, this takes a lot of, you know, a lot of token, it's usage, it's way more expensive than ChatGPT. And so we started to learn the different, we so we kept that, and then we we we also have Gemini as well, and we started to really seek understanding of how they all work. But anyway, we moved to Claude, and then ChatGPT really starts to come back with with codecs that they're now trying to compete with, Claude Code, and then all of the different agents that they roll out, and then like ChatGPT is now making great strides, and then chat image two or whatever it is for the image generation comes out. So it's kind of like one pools ahead and then one pulls ahead in an in the same area, and then one launches this brand new thing that didn't exist, that that now is like in the lead, but they all kind of go back and forth. And so we're we try to stay like agnostic to it, but try to understand what are the strengths of all of them, and then let's use them for the strengths, and if that ever changes, go to the thing that is the strongest in that area. And that's kind of how I look at all three of these.

SPEAKER_00 5:29

Yeah, I think a big thing that happened at the beginning of the year, I don't know what they were doing over at OpenAI. I think they were all like taking a nap or something for the first like few last months of last year and the first few months of this year, because they just got so slow at iterating and optimizing and releasing new things where Claude came in and just like wrecked everyone. Like Claude was rolling out. I think when we when we f officially decided to switch to Claude, Claude was talking about how they were rolling out like 30 new features a month. So that's like a new feature a day, versus ChatGPT hadn't released a new model in the city. You can even keep a lot.

SPEAKER_01 6:07

I mean, still Claude is releasing a lot.

SPEAKER_00 6:09

They are, they are, which like actually, funny enough, I've been following, I don't know if you've been following the SpaceX IPO, which is at literally happening today as we're recording this. First actual AI IPO. And then chat open AI and Claude both are having an IPO later this year, and Claude is looking like they're going to be valued higher than ChatGPT, which is crazy to me because ChatGPT was first on the scene, but Claude has just done better and gotten better.

SPEAKER_01 6:36

They're also really targeting business owners that have the most disposable income to spend $1,000 a month on a tool.

SPEAKER_00 6:41

Yes versus most users are like, oh, use a free version of ChatGPT and use all your and I think I think that's really where where the differentiation that we're about to talk about is going to start to fall. Business models versus personal models. Because I think ChatGPT and maybe even Gemini a little bit are kind of focused on the the end consumer, the the general use person for AI, where Claude is is not focused on that at all, which I think is why well, I don't want to jump the gun, but let's start talking about ChatGPT.

SPEAKER_01 7:11

Yes, yeah. So here's how I view it, and I'd be curious your thoughts. I view uh ChatGPT as the best like for brainstorming, for like even a lot of research. I'll use it as this uh this like preliminary ground before I take things into Claude. Claude, by the way, for me, is usually the end zone because I work a lot in content, and content Claude is just gonna crush. Other than Image Gen, which is the other thing that ChatGPT is fantastic for. Uh its image creation beats everybody right now. We use Higgs field, which is another, it's a wrapper tool that could pull in all kinds of AI. And in Higgsfield, I'll even use ChatGPT or Chat Image2 or whatever the heck it's called. But ChatGPT's image creation is subpart of none. Like it is, it is the it is the best. Um, so I kind of use it for quick brainstorming, image generation, quick answers on things because it's really, really fast. It even the thinking mode is really quick at getting outputs. Claude is a little slower, so it's really fast at that. And then it even has a nice voice mode that I find really valuable. Now I use WhisperFlow to talk to text in everything, but with with ChatGPT, it's so good that you don't really even need to, and it's a lot faster to just use that. So I look at it from that standpoint that it's great for brainstorming and figuring out the preliminary things, and then I will take that and I will move it into the you if because I write a lot of content, I then move any research that I need, and I just say, hey, I'm gonna move this into Claude. Can you give me a rundown of this so I could just copy and paste it over? And it's like boom, and it gives me an like and then I move it over.

SPEAKER_00 8:40

Yeah, it's funny that you mentioned that ChatGPT is fast that you feel like it's faster. I pulled some stats for for the nerds out there. I pulled some stats and GPT 5.4 nano and 5.4 mini are the fastest AI models that we're going to talk about today. There are some faster ones, but the the ones we're gonna talk about today, those are the fastest at the the way it's rated, once again for the nerds out there, the way it's rated is time to first response. The time to where you first start to see text showing up on the screen. 5.4 nano is 0.66 seconds, and five 5.4 mini is 0.68 seconds. So like a little over half a second, and you start to see responses from GPT 5.4 nano and mini. However, GPT 5.5 is the third slowest of all the ones we're gonna talk about at 87 seconds to first response.

SPEAKER_01 9:33

That's shocking to me because I don't I don't see that delay in stuff. I put stuff in there and it's happening within 20 seconds. Uh 5.5. Yeah. Okay. Maybe the the test they're running with was with like some long prompt that they used to cross home. That is probably some complex thing to measure the sophistication and speed, answering a more complex thing.

SPEAKER_00 9:51

Yeah, because this was a like independent research that I found of people like researching every LLM, including like NVIDIA and Grok and Lama and all the like the ones that no one talks about. You don't hear people talk about as often. And some of those actually ranked a little faster and stuff, but we're sticking to the main three today. And then what was also interesting, GPT 5.5 in my research, I found is the second most expensive AI model that we're going to be talking about at $4.35 per 1 million tokens, which is which is really nerdy, but it you can kind of just think of the the $4.35 number per million tokens, which you said writing an email going back and forth.

SPEAKER_01 10:33

It might only be 20,000 tokens.

SPEAKER_00 10:35

It's not that many. It's not a lot, which is why most LLMs you can just pay monthly for, but some of the more expensive ones you go over your usage because of how expensive they are.

SPEAKER_01 10:44

I have never hit the thing about Chat GPT, the reason I use it is because if you're unfamiliar, there are usage limits within the AI, the different like platforms. None of them, I've never hit a limit on any of them. Have you hit a limit on ChatGPT before? I've never hit a limit and had to pay overages or you for over usage in anything but Claude.

SPEAKER_00 11:04

I think I may have hit a limit one time back when I was pushing Chat GPT to the limits before it was really good, before AI in general was really good. I was trying to make an app, and I think I hit a limit at that point, but I was on a free plan, so I just stopped and waited until the next day.

SPEAKER_01 11:21

Yeah, so so like Claude has these Claude, we hit usage limits the first month we were doing it. I think that first month we ended up paying like $800 or $1,000 in overages in Claude that month because the team was like really heavy on it. Now it's dialed back and we don't pay much beyond the typical usage limits. Even though it doesn't really matter to me because the output is worth it. But I never hit that limit. I never hit limits inside of ChatGPT. And I think this is a good note. Like if you know the strengths of all of them, not being only in one is gonna actually save you money because you could use the tool that's best for the thing and then move it into the other thing when that's necessary. And then it save it spreads your usage out amongst them. So that's why I'll use ChatGPT for brainstorming and research because the it maybe it's the usage limits are higher, I don't know, but I've never had to pay for overage of the $20 or $25 a month seats that we have for the team. Yeah. And then I pull it into Claude so that I can use the most complex, most expensive models in there so I get the final output that I really want.

SPEAKER_00 12:19

Yep. For me, ChatGPT's image generation is like top top tier. For me, as of right now, I think that's probably the only thing I use ChatGPT for is the image generation, just because it it really is the best. And it's also pretty fast. ChatGPT was the first to come out with image generation, and I think they kind of have a little bit of an edge on that because of that. I remember when it first came out, it was like it was it's it's that initial response to AI where it's like, wow, this is amazing. And then you try to do anything further than like the initial make this photo look dumb. And it doesn't turn out. Now with with ChatGPT's image generation, it's actually genuinely impressive what it can do and what it outputs.

SPEAKER_01 13:03

Yeah, and I would say that like just a really practical use case for this is if you need to build new creative for like your ads, you can go grab an inspirate an inspirational ad that you really like, and then you can take all of the message that you need to deliver inside an ad and then say, Hey, I'm gonna drive traffic to this page, and you can upload the image inspiration and then say, I want to build an uh an ad similar to this one, but with our company branding go for this type particular product, and it will literally create insanely good creative. So it's like, and you can say, I want a 1080 by 1080 and then 9 by 16, and it will literally do it. So uh from a practical sense, that's something that you could do, have it build creative for you, and the outputs are gonna be they're just gonna crush pretty much anything else. Gemini is pretty, pretty bland, and then Claude is terrible for image gen. Claude, can Claude do image gen? I think that it yeah, I believe it can. Yeah, I believe that it can.

SPEAKER_00 13:58

I didn't know that.

SPEAKER_01 13:58

Yeah, yeah, yeah. It can. I just don't use it a whole lot. Yeah, speaking of Claude, let's move into the second one. I look at Claude as the like I look at it as so ChatGPT I would call more of like the ideas machine and like preliminary research and image gen stuff. I look at Claude as like for content, design, and automation. And that's super business stuff. So Claude is more of like the business tool that we use, and uh the writing outputs of Claude are so, so good. And it has given us the ability to streamline all of our team because of the the custom connectors that it has and the skills that you could build inside of it. You could share skills with other people. We're not gonna get into the weeds of that. That other episode that we talk, we go really deep on that stuff. But basically, we could share skills amongst our team. So our ideal customer profiles, our different offers, our programs. And so, if like Chris is designing a sales page or something like that, it will pull in our offers, it will pull in our ICP, our messaging frameworks, and it will literally write the page so so good. Same thing with emails. I hand write a lot of my emails, but there are some things that I'm like is more informational that I don't, and I could the output of it is really, really good. Especially Claude's new fable. Fable five is which just came out as the recording of this episode two days ago.

SPEAKER_00 15:23

Uh I think it was like two days ago.

SPEAKER_01 15:24

The hard part of writing content is I'm like, I've written so much content that I am so specific about the control of the quality that I die a little bit inside if it doesn't meet that quality. Just like for you, it's like their design and branding standards that Dylan will die inside if we if they're subpar. Yeah. That we hold ourselves to. And so that's me with content. And the hard part is it has been so difficult to replicate the brand voice and the quality of copy or content that I write. And so Fable is the first time that I've actually seen it take the Josh Coffee brand voice and be able to take our ICP and all this stuff and write something for me from me just talking to it saying, Hey, I want to write an email on this, boom, boom, boom, boom, boom. Here's a story, here's an example, here's some numbers. And it's the first time it's written this morning. It cranked out four emails of me just talking to it, and it wrote it in my voice 100%. I had zero edits. I'm like, whoa, this is like now for me, that is that is so fast because I I I could talk stories and tell in the value is in the content, not the writing of it, but it writes wrote it exactly how it write it, and that's what was so cool.

SPEAKER_00 16:31

I think one thing that Claude really has going for him, as dumb, as dumb as this is, is their marketing. It's not really marketing, but the way they communicate their their product. Oh, yeah. Like ChatGPT, when it first came out, it's like, we just came out with Chat GPT 4.5, 4.5 mini, 4.5 nano, 4.5 pro, 5.5, chat GPT 5. It's like, what is all like what is the point of all of this? I don't understand. And then it was always ranked super dumb where it's like you couldn't really tell. It looked like GPT 4 was better than 5, and like it just never made sense. Claude is really easy to approach, and I think it's easy approach. And I think there's two reasons. One reason it's easy to approach is because the models just make more sense. Like it they're the models are very easily laid out. There's four models. You have Fable, Opus, Sonnet, haiku. That's it. Like it's super simple, and you can tell the order that they that they go just in the drop down. And then the second thing is how they split out everything. I feel like with ChatGPT, it's all hidden in settings. It's all hidden. You have to deep dive, you have to find all the things that you need. Versus with Claude, right on the left-hand side, you have claw design, claude code, connectors, skills. Like everything is super high level. It's such a small thing, but I think it allows you to dive deep into what Claude can actually do, what the difference is between all of the different tools inside of Claude, connectors and skills, and all of that. I think the branding of Claude Design. Well, what do you think Claude Design is for? Probably design. Claude code. What do you think Claude Code is for? Probably code. I think it makes it a lot easier versus like with Chat GPT, you have to know what model is best for the thing you're trying to do versus just going into the place to do the thing. Like that's that's one thing that I think Claude really has going for it. Something else, Claude Fable and Claude Opus 4.8, based on the the research that I did, are independently rated as the highest intelligent models across all AI. Makes a lot of sense. It it does. And it's really by a long shot. Like the the intelligence level is you can see a steep drop-off. Like if you were to chart it, you could see a steep drop-off once you pass those two. They are the smartest. However, you pay for that. Fable is $7.70 per 1 million tokens, the most expensive model in the world. This includes all those crazy like NVIDIA grok meta models that no one talks about. Fable is the most expensive. So just to just to remind you, the second most expensive is $4.35 per million tokens. Almost twice as much, and less. So you're paying you pay for it, but I think the outputs are probably worth the cost.

SPEAKER_01 19:21

I love how they use tokens because it's so like it's vague and ambiguous. And people like, oh, a million tokens. It sounds like a lot. I don't know. Uh-huh. I don't know what that is. But but my my whole perspective is like the speed that you get is so great. And I use, I don't try to use the lowest value or the lowest ones. I I like, or I ascend to the highest if I need it. So if I'm just doing brainstorming, you could use some low. If it's like a super small, simple task, I'm like, hey, can you go into my daily tracker and add this to my tasks? I'll use a lower model. But if it's like if I'm doing writing or like high value things, I want the best outputs. And the cost is so small compared to what the alternative is. I look at, if you look at like if you look at multiplying, I would look at AI is a simple multiplier because it's simple because you just choose a higher model and and you can multiply, have a much you could create a lot more content with higher models that are more accurate and saves you a lot of time. People are a multiplier as well, but it's a complex multiplier. So AI will multiply your save you a bunch of time. Hiring more people will save you a bunch of time so you could scale, right? And not have to do all the work yourself, but it's really complex. You have personalities, you have skill, you have training, you have coaching, you have all this other stuff. And so if the alternative of using more expensive models is I just have to hire people to do the work, oh, I would rather just have our entire team use the highest models and then pay an extra $500 or $1,000 a month, whatever it is, in overages, because now our team can output something worth like three different people, is is is how I kind of think of it. But yeah, Fable, um, I've only been testing it, it's brand new. Yeah. And the content of it, it output is very, very stellar. I've used the the low version, the medium, the high. I haven't used like the max Fable Max. I'm sure that's real expensive. I haven't used it, but honestly, I'm gonna try it just because I want to see what's the output of this. Because medium and high got me 99 to 100% output of what I needed, and I had no tweaks.

SPEAKER_00 21:25

Something else that I've noticed with Claude, I think you've probably seen this as well, is Claude is really good at building visuals for you, like graphs and interactive dashboards. It has a whole, once again, goes back to the the marketing of their of their software, the artifacts. You can go to the artifacts on the left hand side and it'll build you essentially a website or a dashboard to look at the things that that you want to, where really no other tool will do it that well. GPT 5.5 can do it a little bit, but it's it's not nearly as good as Claude because it builds interactive dashboards where you can select. Time frames and then you can ask it questions about it and everything. So Cloud's really good for visuals and things like that. Like internal tools, dashboards, like you said, automations. And then I've been able to build like crazy coded Python scripts and apps that are running on servers and stuff through Cloud with no Python GitHub or server experience, which without without hiring a developer or without having to learn it, I was able to just take my idea, give it to the AI, and it was able to accomplish it.

SPEAKER_01 22:34

I even like uh earlier this week, Isaac, did you see Isaac's thing that he built? Isn't that insane? Yeah, that's crazy. Isaac used Cloud Code. So Isaac's like our client experience director, and so like he built something that if you're a client listening, it's gonna be releasing in July for you, and it is insane. It is a he used code to basically build an entire thing for all of our clients. It's like a platform, essentially. It's really cool. So I don't want to spill the beans, I guess. Anyway, I'm not gonna- I don't want to ruin Isaac's idea. It's really cool. It is really cool. We released it on July 20th, or July 2nd when we do all of the new release of content, all kinds of stuff. Um, but all that said, Isaac built that in one day. That would have taken $25,000 in devs in months to build what Isaac built in a single day using Claude code. And I mean it has login stuff. I mean, it's so cool what it has. But anyway, yeah, I think the Claude definitely highest value. I also think a lot of the automations and the things within it, too, like cowork and the ability to like have it doing multiple things at once is another really big one. So I'll use cowork. And if you want, if if if we'll get working on one thing and go to the next thing and then do the next thing and then do the next thing, and it will just start cranking and doing so much stuff. So your efficiency will skyrocket if you're using Claude versus Chat GPT. You have like single chats going, and like sometimes they don't they stop running if another one's going, and so that's been my observation in in that regard.

SPEAKER_00 23:56

And then lastly on Claude, just you know, for the nerds out there, the nerd stats, Claude is the slowest AI. So if you care about speed, which you know probably doesn't matter all that much, I know I'll sometimes find myself prompting Claude and then just, you know, twiddling my thumbs a little bit. But uh, it is rated to be the slowest at 114 seconds to first response for Sonnet and 107 seconds to first response for Fable. So something to keep in mind. It is gonna be slower, but obviously you're gonna get better outputs for it.

SPEAKER_01 24:29

Awesome. Now, before we hop into the last one, Gemini, and we'll talk how we're kind of using Gemini. If if you're using a lot of AI in your business, one thing that we've seen a lot of people make mistakes in is using AI to make decisions on their meta ads. And we actually did you did an episode on this a couple weeks ago. We'll drop the episode link to that, but a couple weeks ago. It is uh episode 247, May 28th. 247. Dylan talked about how we have clients that are like using the the connectors for like meta ads and trying to make decisions on their meta ads based on data that Claude is pulling in. And and we have so many countless clients that are literally trying to export CSVs and have Claude analyze their different campaigns and make recommendations on turn this ad off, turn this ad on, increase this budget, and then it'll give you data and performance on all of it. And you know what we found? A lot of it is actually inaccurate and it has absolutely wrecked campaigns. And so, little note of caution don't don't use it for for all of your thinking. We use AI right now for creation, not letting it do our thinking for us. Now we'll let it like brainstorm, but we're not gonna let it do all of our thinking. However, the tool that we use to help us with meta ads that is data driven and is based on what's actually working in meta is Breezeway. And if you haven't heard of Breezeway, go to breezeway.co. This is the tool that we've co-founded that has really helped us and hundreds of e-com brands make smart decisions since we released it at the beginning of this year. And it's for Shopify brands that are direct to consumer that are running meta ads. And this tool will basically help you make the decisions based on your average order values, your gross profit margins, your break-even return on ad spending cost per acquisition, your targets, and it gives you breeze scores, it gives you breakdowns and visuals and helps you make decisions confidently in your business so you're not wondering like, hey, is this move going to be the bad move? Here's the data-driven truth of what might happen if you do that. So go check out breezeway.co. You can get a free trial there. What I think it's like 10-day trial. Yep. And within 10 days, I'll tell you, you'll see the value in it. And I mean, it's only like 49 bucks a month. So it starts at 49, and it's trust me, it will pay for itself way faster than an agency that charges you thousands of dollars. Yes. Will do. All right, let's move on. Number three, and the final one is Gemini. How much do you use Gemini?

SPEAKER_00 26:47

Well, I can skip to my conclusion if you'd like. Uh I wrote, I wrote, so my my thoughts on Gemini are it's just meh. It really it's just meh. It's not great. Okay, it's great at one thing, and that's the multimodal. It can it can look at videos and images and tell you very very clearly what's happening. It's good at that. Everything else, though, is just meh. It's it's not necessarily great, and especially my research backed this up. That was my feeling, but then my research, it didn't fall as the fastest, it didn't fall as the slowest, it didn't fall as the smartest, it didn't fall as the dumbest, it didn't fall as the most tokens or the least tokens. It just was right in the middle. And I think it kind of goes back to what we were talking about with ChatGPT as well, where it is a the the Gemini is the most consumer-focused AI. Like GP Chat GPT is trying to be consumer focused, like they're rolling out ads and stuff now. But Gemini, honestly, the where I use Gemini the most is in Google searches. Because you you do a Google search, and the first thing that shows up is a Gemini response. And I'll often read it because it if it can give me a decent response, I'll take it. You know, I don't need to it's usually for like the dumb things, like, oh, who's this guy in this movie? You know, and it shows it shows an AI response, and they're usually pretty good. But I would say that's the only re that's really the only time that I use Gemini outside of video or image understanding, because that that's really where it shines, which is a really interesting place for an AI to shine, because that's really the only thing it does shine in.

SPEAKER_01 28:30

Yeah, I'll I'll share how I kind of use it right now, but before I do, you know, in Hamilton, there's in the musical Hamilton, there's this song called The Skylar Sisters, and it's like Angelica, Eliza, and Peggy. Gemini is Anne Peggy. Yes. Yeah. Like I'm here too, guys. Like, hey, don't forget about me. I'm the little step sibling or whatever. That's pretty much how I would view Gemini. I I agree. I agree. As you're talking, I'm just picturing that scene.

SPEAKER_00 28:59

And it's just crazy because it's like, it's freaking Google, man. Well, you have the most money of anyone to invest in this, and they're just like, and then you know, sorry, I have to go on a slight, very small tangent here. Apple just rolled out, WWDC was was on Monday where they rolled out all their new stuff. They just rolled out Siri AI, which will be coming in in the fall, that they partnered with Google for, which they say Gemini isn't powering it. I don't really know what all that necessarily means, but they picked Gemini to to power the new Siri, and I'm just like, I don't know. Like, you picked the most bland, safest option. Like, if only you could have picked Claude. Like, and they showed off some of like the Siri responses that it's gonna have. And it's like, this looks like Chat GPT responses from like three years ago. So that's that's my take.

SPEAKER_01 29:51

Here's what I think. I think that they were like, you know, Google is the cheapest to probably partner with probably because Google is just getting pummeled by AI right now, and they have all these this server space that they're probably like, man, we have so much server space for all this AI computing power stuff because all the stuff that we once were used trying to do in Google is no longer relevant and really SEO is not really a thing anymore. Yeah. So we could free up some server space for AI, and they're probably like the cheapest to work with. Yeah, anyway.

SPEAKER_00 30:20

That is probably true, especially because the next two are getting ready to to IPO for ridiculous valuations. Yeah.

SPEAKER_01 30:27

So here's how I I use it primarily. We will use it for, like Dylan said, we'll use it for analyzing video and images. And it is really good at this. It is good at that.

SPEAKER_00 30:38

It is the only thing it's got going for it, in my opinion.

SPEAKER_01 30:40

So one of the benefits is like when we're trying to build creative, sometimes we'll pull like uh winning creatives, or we'll pull like competitor winning creatives or just creatives that we really like for inspiration. And we could like take the videos of those and we can upload them in Google. You can upload, I think, up to two terabytes of video, or I'm sorry, two gigabytes of video. You can just upload it to Gemini and you could have it analyze the video for you and say, what's happening in this video? What are you seeing inside this video? What ICPs is it talking to? What types of angles might it be taking? What style of video would you categorize this as? And it will analyze what's actually happening inside the video. If you want to use Chat GPT or or or Claude, you're gonna have to upload a transcript because they won't analyze a video. You'll have to upload a transcript and it will just analyze the transcript, not what's actually happening. And so this is where Gemini has a really, really big advantage. And so how I will often strategize is I'll go to Gemini and I will have an analyze creative video or image and what's in it. And then based on what's in it, I can take that and go to Chat GPT or go to Claudia. I'm gonna go to ChatGPT if I need to build images, and I could say, hey, here's the inspirational thing. Here's a breakdown of what's what happened inside the here's a breakdown of what's happening inside a video, and I want to craft one of my own for this ICP, whatever, and it will literally help you like write a script, or you could build images based on that. And so I like to use Gemini. At this point, the only thing I use Gemini for is literally analyzing creative image and video, but that's it. Even its research ability, I know it says it could do good research. I don't think that's really good.

SPEAKER_00 32:16

Like I said, I use it in Google searches, and I think that goes to show it probably is good at research because when I do a Google search and you know, ask it anything from who's in this movie, or you know, I I'll even ask, like, is Meta down today? You know, and it'll it'll say, Oh yeah, I found something on status gator, which is like a crazy where people can report they feel like it's down. And you know, it pulls it up, finds it. It makes it easier because you don't have to rely on SEO, you don't have to like jump over the ads, because I feel like that was an issue with Google where you know the first thing that it showed was always an ad, so you had to scroll past the ads. It just goes to the the first thing that the AI thinks is the the most valuable for your for your search. So I think it is good in Google searches, but I wouldn't go to Gemini to search something. I would go to Google and ask it a simple question, but I wouldn't go to Gemini and ask it to do research for me. I would use Claude for that, probably.

SPEAKER_01 33:15

Yeah, or Chad GBT for that. Yeah, I would agree on that. Well, hey, to close up here, he I think that the the our thought is it's not about what AI is the best. It's just what is best for the job that I'm trying to get done. And if you ask that question, it might change over time because like the little marbles that I was talking about, sometimes one takes a lead in an area and another takes a lead. Like Google is pretty Gemini is pretty far behind and and ChatGPT is crushing with images, but we know that Google's really good with image and video analysis, and they now have the image and the video capabilities built into the side now, and they're developing that even further. Uh the outcomes are still pretty rough, in my opinion, right now, but it's gonna get better. And so maybe they will take, I would guess that at the end of the day, Google's gonna trail on image and video because they just have more to analyze and more data on that than probably anybody else. And so I think it's not what is the best, it's it's what's best for the job. And if you look at all three of them, you're talking $60 a month to have access to three different tools for three different specialties. And if you try to compare the alternative of hiring people that are really good at the different specialties that they might do, it's gonna cost so much more. And so that's how we're using it. That's how we kind of view each of these tools at this moment in time. Maybe that will change. But hopefully this was valuable. If you did like this, please do us a favor. We're trying to become the number one e-commerce podcast on the planet. And it would mean a lot to us if you just spent five seconds and you rated this podcast on whatever platform you're listening to. If you're on YouTube right now, uh go ahead and hit a thumbs up and actually drop in the comments. Let us know what uh what platforms are you actually using right now the most, and maybe how are you using? Are you using them the same way we are? Or maybe there's something that we're not doing that we maybe should do or should start to use them for. Uh, but we would appreciate it, and we'll see you in the next episode.