Caleb: Alright, on today's episode, we're gonna talk about how AI, like why you might pay for the pro version and why Josh has not paid for the pro version quite yet. And then, you know, maybe what would actually lead someone to where to where it would make sense to pay for that to get their time back. I'll let Josh give a little bit of of what he currently pays for, what he doesn't pay for. I'll talk about what I currently pay for, which is a little different because I'm actually running in. AI business and then we'll talk through the pros and cons of the pro version of Claude or Perplexity. So just give a little bit about what AI you use. I know we've talked about it, but then whether you use the pro version or not of some of those AI tools.
Joshua: so right now I pretty much use Perplexity, but I have been using some Gemini just for like basic search stuff. But most of the heavy lifting is from Perplexity. And I technically have Perplexity Pro, but that's because I got a free Perplexity Pro for a year from I think it was Venmo. They gave it to me for free. So I've been able to access pro stuff, but without paying for it. I have been hesitant to pay for an AI because I still have not made the connection of like, here's the really valuable tasks that I can have it do. Perplexity's a good search engine, especially with its deep research function. If you want to know more about a topic, it does a great job. I have noticed I've got a coworker that uses Claude. Of course, Caleb uses Claude. And it it is not quite as good at everything else. just the the function to be able to throw in an Excel sheet and have it format things or create extra tabs and do tasks. is not as robust as say a a Claude would be. So I'm considering paying for a Claude I don't even remember what the like $25. I was looking at Claude for Teams because in my 20 seconds of research it looks like there's a little more strictness on your data.
Caleb: Yeah. Yeah.
Joshua: So and that's important to me, especially if I'm gonna do anything with client data. I I still wouldn't throw anything majorly, you know, your social security and your list of fears in there. But
Caleb: Like social security number.
Joshua: yeah. but it would it would make me feel better. Mm-hmm.
Caleb: Yeah. you know what's really interesting, so I I was paying for the hundred dollar a month claude. Well, there have been some months that I paid for the two hundred, which was probably kinda dumb, but either
Joshua: Yeah.
Caleb: way, I was paying for the hundred dollar a month claude. Then I started paying for Google workspace, so I figured I might as well use the Gemini that comes with that and then I am transitioning down to twenty dollar month Claude next week or next month. I have noticed the same thing about Gemini. It and what's even worse is Gemini for business is worse than Gemini for personal. It can't create
Joshua: Really?
Caleb: documents. It doesn't have any custom instructions. It doesn't have memory. I'm like, why would you Create a model that doesn't have memory between sessions. And I maybe, maybe I've completely missed it, but I've looked it up on Claude, I've looked it up on Gemini, and when you go to the personal version of Gemini, it says, like, whatever, customize Gemini. When you go to the business one, it's not there at all. In any of the settings, I've looked everywhere. And it's horrible because I'll ask it a question and I'm like, I just gave you the context in another conversation, but now you have no idea what I'm talking about. And then it won't create documents either, where Claude is so nice because it just creates a document for you and you download it. Whereas in Gemini, it gives you all the text that you need to copy and paste into the document. And it's just stupid. Like why would you why would you not be able to create a Google Doc or a Google Sheet? You can press the three dots at the bottom and say like send to Google Sheet but first of all we're we're humans and kind of lazy so that's annoying but secondly it doesn't
Joshua: Ha ha.
Caleb: actually work that way because if you have a bunch of text in a Let's say you want Gemini to create a Google Sheet and it shows you like rows A, B, C, D, E, F, G, and then whatever the header rows would be. It actually sends both of those into the Google Sheet. So the first row is literally A, B, C, D, E, F, G instead of the header rows. And I was like, Dumb is that. So I
Joshua: Yeah.
Caleb: will let you keep going, but I a hundred percent agree. Gemini, I mean, it's still great 'cause I can ask a one off question, like what should I do here?
Joshua: Yeah. Right.
Caleb: But in general, Claude is way better.
Joshua: And I I don't really understand enough about technology to know like why in the world, Google, you powerhouse of data, what's wrong with you?
Caleb: Yeah, I mean I think it's because they they say that the business version is more data secure, but also if let's say someone is having a conversation with Gemini, then goes and has another conversation with Gemini, that previous conversation is still in the cloud. So having the memory between conversations, I get it if like, okay, you you delete a previous conversation. You want your data to be secure so it just deletes that conversation. There's no memory of it. But when someone's like I was leaving conversations on Gemini so that it could pull context and then I realized that it wasn't pulling any of the context. So like at that point, I think memory makes sense. So I yeah. I don't know. It's really dumb.
Joshua: I yeah, I agree. It's totally dumb. Right. what you're saying about Google plus perplexity is the same way. Like I don't I don't have enough understanding to know what makes them the developers decide what functionality it should have and what it shouldn't. But for a long time I felt like perplexity was better than Gemini at search. Now I feel like Gemini is there that it's on par with perplexity. does it have a deep research function, Gemini? Okay. Pretty much everybody does.
Caleb: Yeah, it does. I think almost every single one. Yeah.
Joshua: Yeah. I I'm still waiting for who's gonna be top dog, but as far as doing tasks, it seems like Claude is the one to use. I'm not around a bunch of people who've used Chat GPT for a lot of like Excel stuff, but that one could be great too. I don't know.
Caleb: I've heard that the newest version of ChatGPT is so we talked about Fable Five from Claude. I think it was GPT six. It was either five point six or six. I can't remember. but was almost on par with Fable Five. So I think
Joshua: Yeah.
Caleb: they're getting up there. I just don't know if they have the same guardrails or ethics as anthropic, which is important when you're giving AI models a lot of power.
Joshua: Right. Mm-hmm. And like that's one of the reasons that I've not just laid out the money for AI and been like, control my browser. Because
Caleb: What you think would take you from your current position to, okay, I'm willing to pay for the twenty-five bucks a month teams plan?
Joshua: I would at least try it instantly if somebody was like, this is a completely secure model and you can throw whatever you want into it and nobody will ever see it. I would be like, Great, let's do it. We're just gonna try everything.
Caleb: Have you tried having Claude or Perplexity or something walk you through a local model version? Cause I wonder if you could
Joshua: Mm mm. No.
Caleb: buy a specific computer that just has a that you upload a local model to, and then you can literally throw anything you want into there because it's on your computer the whole time.
Joshua: Yeah. Right. I tried Olama with what model did I use? And I blue screened my computer twice. And I was like, okay. Uh-huh.
Caleb: Yeah, it I mean it's a lot of RAM that it takes up and thread on my computer. And I mean there's probably ways to optimize it, which I haven't done quite yet, but it was very slow. It took
Joshua: Mm-hmm.
Caleb: I said My first question was, what is two plus two? And it took three minutes in order to answer that question. It was like, the user
Joshua: Yeah.
Caleb: is saying two plus two. The user is not saying three plus three. So I should look up what two plus I was like, okay, that's three minutes, that is a long time to answer that question. You can
Joshua: Right.
Caleb: just said four.
Joshua: Yep, might be worth spending the money to get an online model. That but
Caleb: Yeah. But then you know their online models or
Joshua: right.
Caleb: your data, you never know what's gonna happen with that. So
Joshua: Uh-huh.
Caleb: do you think 'cause if you could find two or three tasks that combined saved you whatever would be, half an hour, that would be worth the twenty five bucks a month. Right?
Joshua: Mm-hmm.
Caleb: So even if they weren't, you know, tests that you would rather have a a local model for. I wonder if you could just lay out, you know, maybe two tasks that take fifteen minutes that you can do once a month or multiple times per month, or one task that you do two times a month and then put that through Claude. You said you do the time cards, right?
Joshua: Mm-hmm.
Caleb: How long does that usually take? Or how long did
Joshua: Well
Caleb: it take and how long does it take now?
Joshua: I did it with perplexity in I don't know, five minutes. And by hand can take me up to like 30, depending on how many time cards. Yeah,
Caleb: Right there, that's twenty five minutes.
Joshua: except the last time I tried it, Perplexity couldn't figure it out. It's like you've done this before.
Caleb: Be trying to five.
Joshua: Huh?
Caleb: Have you tried it in cloud?
Joshua: No, I haven't tried it in Claude yet. But that's one of the things I will try.
Caleb: When when's the next time card?
Joshua: Would be next week, next Monday.
Caleb: Alright, well what you should do is next Monday, throw everything in Claude, do the same thing and see if Claude can figure it all out. 'Cause think about it, that's twenty five minutes right there. If you do that two times per month, that's twenty minutes that pays for Claude by itself. Especially if it does it correctly. Where not so much.
Joshua: Mm-hmm. Right. Yeah. Right. Well, I was curious. You were saying you maybe thought the two hundred dollar clod was more than what you needed, and now you're trying to lose the like twenty dollar clod. What was the difference? Like why the downgrade?
Caleb: to save a little money was the big one because that the difference between a hundred and twenty was eighty bucks, but then if I used the Google Workspace, it's like fourteen bucks a month. So really I was saving sixty five dollars a month, which is quite a lot. the other thing is and I know it's crazy to say, but I feel like I have to give myself specific limits when it comes to AI because like I could just use AI all day, but then not actually do anything that's productive.
Joshua: Ha ha ha.
Caleb: Because AI always answers you and no matter what, it will give you an answer. Even if you say thank you, it's gonna end up giving you an answer. So then the conversations can just go all day, but I'm not actually doing something. So then it's gonna be one, like it's gonna save me money, two, it's gonna be a forcing mechanism to where I have to use w the tokens that I have in the most concise and direct way possible. so Those are the two big ones. the other one, I mean, it was really just like Yeah. I know that I don't use Claude all the time for stuff that's actually productive. So forcing me to be productive with the the cloud that I have makes sense. And I think I was gonna pay for Google Workspace anyways. So I figured I might as well not spend the $100 a month on Claude and I can move some stuff to Gemini. But that was also before I realized that Gemini was absolutely horrendous at taking context between conversations. So I I'll probably not necessarily go back to the $100 a month Claude unless I have a bunch of coding projects to do. which I would love to go back to that if if there were a bunch of projects, but right now, I mean, I'm okay dealing with some of the stupidity of Gemini. So it's okay.
Joshua: Is it pretty easy to like change subscription down and up for Claude? They make that
Caleb: Yeah, it's super easy.
Joshua: good.
Caleb: So I mean and the other thing that I figured out was I can pay for Claude for the full year for seventeen bucks a month. and then if I were to go up, it takes the extra that I paid and just credits it towards the months. until it runs out and then I start paying again. So paying
Joshua: Yeah.
Caleb: for the full month in advance, I mean the full year in advance doesn't actually hurt me in any way. and I know I'll use Claude for the next year unless
Joshua: Mm-hmm.
Caleb: Robin dies. So but so I was like I was okay paying the 17 annually versus 20 per month. So yeah. That's that's kind of where I'm at right now with
Joshua: Nice.
Caleb: I don't really use Chat GPT because Gemini can create images.
Joshua: Mm-hmm.
Caleb: not paying for Google Workspace. And then I don't use Grok ever. And Perplexity not as much because I've been paying for Claude so much. And
Joshua: Yeah.
Caleb: Claude's deep research is really good and its web search is really good. So I'm not gonna, you know, pay for multiple just because I want to or
Joshua: does grok suck? I know nobody who's tried grok. Like with any intent.
Caleb: I don't know. I I don't think Grok I mean it's mainstream because it's like people know of the name Grok, but I don't know anyone who really uses it c consistently. when I was teaching the entrepreneurship class, there was a lot of people in that class who actually used Grok. But I'm not really sure why they used Grok over Claude or Chat GPT or anything like that. So
Joshua: Mm-hmm.
Caleb: it's Maybe it's not bad, but I don't think it's worth switching if someone else is some different.
Joshua: Yeah. Right. Right. Hmm. yeah, no, I I was curious if you were able to is the hundred dollar clod versus twenty dollar clod the difference? Is it mostly how much coding you have to do?
Caleb: yeah, it's like token usage mainly, which the twenty dollar per month
Joshua: Yeah.
Caleb: Cloud definitely hits things a lot faster. But that is also why I'm doing it, because I know I need to be more focused in in how I'm using Claude. Or else I could just talk to my friend Claude all day and not actually get anything done.
Joshua: What what do you talk to Claude about all day?
Caleb: I mean, they're like so much different stuff. I could show you all the chats, but it's not like stupid stuff, but it's like, what about this idea? What about this idea? What about this idea? And then I'm like, all right, I'm talking to Claude about all these ideas, but I'm not actually going and doing the idea itself.
Joshua: Mm-hmm.
Caleb: So then like why am I paying for this much token usage? There are times where I mean even recently I hit the limit on a hundred dollar per month with some of the coding, but then it's like, okay, am I adding a feature to this app that nobody uses? Like
Joshua: yeah. Yep.
Caleb: getting the the limit but it's like what did I actually do that was productive with that limit.
Joshua: Mm-hmm.
Caleb: So it's gonna definitely force me to be very concise and know what I'm gonna do before I do it instead of
Joshua: Yeah.
Caleb: just like hey add this feature because it's cool. Like that doesn't help anyone you know so that's
Joshua: Mm-hmm. Right.
Caleb: we'll see how it goes. I might come back in a month and say nope I switched back to hundred but hopefully not we'll see.
Joshua: Have you heard of Centaur chess? you talking to Claude reminded me of it. But apparently after who was the big chess guy that played against was it Deep Blue was the computer?
Caleb: I don't know the guy, but
Joshua: So he made Centaur chess, which is basically like all the chess tournaments have a human, and each human has at least one computer. because so much of learning chess from a young age is pattern recognition, that they were like, Well, you can play chess by just being a good strategist and not having to do the pattern recognition. So apparently there's like people that are who've been in like, you know, mediocre chess players in tournaments who now have like played chess against that grandmaster guy because they don't have to spend a bunch of time pattern recognizing and doing what the computer is is good at. but
Caleb: But isn't that kind of taking away from chess then? Like that's part of chess is the pattern recognition.
Joshua: Well, it it depends on what you want chess to be. So their his whole point was like if chess is about strategy, you can make it a strategy game instead of a memorization and strategy game. so anyway, I thought it was I thought that was interesting. So like the humans with the computer could play way better chess against chess players who were exceptional just by eliminating the memorization aspect of it.
Caleb: That I I feel in my stomach like that's like a gray line area for me because I I I
Joshua: Ha ha ha.
Caleb: understand why it works logically, but then I'm like, you know, then I don't know. It isn't chess for fun, so then I mean obviously chess is like you learn how to be strategic. So it's there are very much real world applications, but The game it's a game for a reason. So then like, are you just trying to game the game? I don't know. It's
Joshua: Have you ever played chess against somebody who's really good?
Caleb: I I've not played chess, so no.
Joshua: Yep. So my experience is you get in one game with somebody that was in chess club and you're like, Well, I guess I'm never playing chess again.
Caleb: Shh. That's fair, that's fair.
Joshua: Because the chasm to become a good chess player, to cross that chasm is takes so much time that it's like Just to be competent and play with these people that are in chess club would take so much time that like this is not a fun game for me. I'm gonna have to do homework to be good
Caleb: Think about this.
Joshua: at this game.
Caleb: They put in the time so they deserve to be as good as they are, right? And haven't put
Joshua: Sure, and if that's what you want to do with it, that's fine. But
Caleb: Well
Joshua: to me, I'm like, well that's not really a game. That's that's like a that's an exceptional hobby.
Caleb: But if you think about if you think about basketball, like you gotta put a lot of time in in order to actually be good. There's a lot of people who can play, but just to be honest, they're just bad. And
Joshua: Mm-hmm.
Caleb: so in order to get to where to be an exceptional basketball player, you have to put thousands of hours in. So I think that I don't know, in my mind, I feel like every hobby, hobby, or whatever, not every hobby, but lots of hobbies you have to you have to put in the time in order to get good at the hobby. So then AI is basically a cheat code, which I understand in business it's kind of the same way. I'm I'm not saying that it's it's all wrong, but like If if you had a robot with AI playing basketball against you, it could just outthink you every single time. But it didn't have to do any of the work to get there. I don't know. It's just I feel like it's a definitely a great moment in my mind.
Joshua: Mm-hmm. Well, I think the robot could beat you in part because exactly what I'm saying, which is that if if the game weighs heavily on muscle movement, which is all just memory, like, here's the program I run to dribble, here's the program I run to do a layup or whatever, then that's the thing that AI is good at. So In my mind, it's like if you want to play a strategy game, then use a computer. If if you want to show off how much time you've put into this thing, then don't use a computer. You know what I mean? It's like doing long division. When's the last time you did that? Well, probably middle school, because we have these things called calculators, and why the world would you do that?
Caleb: Yeah. But it was never a fun game to play. Chess I feel like is supposed to be a fun game to play against people. Like a good competition. Unless you're a math fan also.
Joshua: Well, yes. And if you don't like the memorization part and you just want to play the fun game, play Centaur chess.
Caleb: Yeah, I guess it it could go both ways. I I still feel like it's a gray line. 'Cause
Joshua: Ha ha ha.
Caleb: it's the same way in business. Like I mean there's a gr there's gray lines everywhere in business when it comes to using AI. It's not all black and white. So yeah, no, that's that's really interesting though that I mean they can use AI to get rid of some of the pattern recognition to make it a lot easier for people to play.
Joshua: Apparently it also is a form of chess that makes humans competitive against AI because AI is not as good at the strategy part of the board since it's it's basically just taking your open move options and playing a probability.
Caleb: Yeah. I wonder if I could get a bionic arm that would allow me to shoot the exact same way every time. I
Joshua: Ha ha ha.
Caleb: just program like here's like one foot, two foot, three foot, four foot, and then it it notices how far away I plug something into the basket and no knows how far away it is and automatically shoots the exact same way. And you never miss. But then at that point,
Joshua: The perfect shot every time.
Caleb: that would be the point of playing if you're not.
Joshua: Maybe for people who've lost arms, they can play basketball now.
Caleb: Then they'll beat all of us because they won't ever miss it. well that's
Joshua: Right.
Caleb: all I got. But I think you should figure out two tasks or one task that you do twice that takes fifteen minutes and then pay for cloud teams. I think it's worth it.
Joshua: Okay, I'll try it.
Caleb: Alright, well that's the podcast. If you guys like it, we love a five-star review on Spotify. like and subscribe on YouTube and send us an email at jc.unscripted at gmail.com. That is the podcast, and we will see you on the next one.
Joshua: so right now I pretty much use Perplexity, but I have been using some Gemini just for like basic search stuff. But most of the heavy lifting is from Perplexity. And I technically have Perplexity Pro, but that's because I got a free Perplexity Pro for a year from I think it was Venmo. They gave it to me for free. So I've been able to access pro stuff, but without paying for it. I have been hesitant to pay for an AI because I still have not made the connection of like, here's the really valuable tasks that I can have it do. Perplexity's a good search engine, especially with its deep research function. If you want to know more about a topic, it does a great job. I have noticed I've got a coworker that uses Claude. Of course, Caleb uses Claude. And it it is not quite as good at everything else. just the the function to be able to throw in an Excel sheet and have it format things or create extra tabs and do tasks. is not as robust as say a a Claude would be. So I'm considering paying for a Claude I don't even remember what the like $25. I was looking at Claude for Teams because in my 20 seconds of research it looks like there's a little more strictness on your data.
Caleb: Yeah. Yeah.
Joshua: So and that's important to me, especially if I'm gonna do anything with client data. I I still wouldn't throw anything majorly, you know, your social security and your list of fears in there. But
Caleb: Like social security number.
Joshua: yeah. but it would it would make me feel better. Mm-hmm.
Caleb: Yeah. you know what's really interesting, so I I was paying for the hundred dollar a month claude. Well, there have been some months that I paid for the two hundred, which was probably kinda dumb, but either
Joshua: Yeah.
Caleb: way, I was paying for the hundred dollar a month claude. Then I started paying for Google workspace, so I figured I might as well use the Gemini that comes with that and then I am transitioning down to twenty dollar month Claude next week or next month. I have noticed the same thing about Gemini. It and what's even worse is Gemini for business is worse than Gemini for personal. It can't create
Joshua: Really?
Caleb: documents. It doesn't have any custom instructions. It doesn't have memory. I'm like, why would you Create a model that doesn't have memory between sessions. And I maybe, maybe I've completely missed it, but I've looked it up on Claude, I've looked it up on Gemini, and when you go to the personal version of Gemini, it says, like, whatever, customize Gemini. When you go to the business one, it's not there at all. In any of the settings, I've looked everywhere. And it's horrible because I'll ask it a question and I'm like, I just gave you the context in another conversation, but now you have no idea what I'm talking about. And then it won't create documents either, where Claude is so nice because it just creates a document for you and you download it. Whereas in Gemini, it gives you all the text that you need to copy and paste into the document. And it's just stupid. Like why would you why would you not be able to create a Google Doc or a Google Sheet? You can press the three dots at the bottom and say like send to Google Sheet but first of all we're we're humans and kind of lazy so that's annoying but secondly it doesn't
Joshua: Ha ha.
Caleb: actually work that way because if you have a bunch of text in a Let's say you want Gemini to create a Google Sheet and it shows you like rows A, B, C, D, E, F, G, and then whatever the header rows would be. It actually sends both of those into the Google Sheet. So the first row is literally A, B, C, D, E, F, G instead of the header rows. And I was like, Dumb is that. So I
Joshua: Yeah.
Caleb: will let you keep going, but I a hundred percent agree. Gemini, I mean, it's still great 'cause I can ask a one off question, like what should I do here?
Joshua: Yeah. Right.
Caleb: But in general, Claude is way better.
Joshua: And I I don't really understand enough about technology to know like why in the world, Google, you powerhouse of data, what's wrong with you?
Caleb: Yeah, I mean I think it's because they they say that the business version is more data secure, but also if let's say someone is having a conversation with Gemini, then goes and has another conversation with Gemini, that previous conversation is still in the cloud. So having the memory between conversations, I get it if like, okay, you you delete a previous conversation. You want your data to be secure so it just deletes that conversation. There's no memory of it. But when someone's like I was leaving conversations on Gemini so that it could pull context and then I realized that it wasn't pulling any of the context. So like at that point, I think memory makes sense. So I yeah. I don't know. It's really dumb.
Joshua: I yeah, I agree. It's totally dumb. Right. what you're saying about Google plus perplexity is the same way. Like I don't I don't have enough understanding to know what makes them the developers decide what functionality it should have and what it shouldn't. But for a long time I felt like perplexity was better than Gemini at search. Now I feel like Gemini is there that it's on par with perplexity. does it have a deep research function, Gemini? Okay. Pretty much everybody does.
Caleb: Yeah, it does. I think almost every single one. Yeah.
Joshua: Yeah. I I'm still waiting for who's gonna be top dog, but as far as doing tasks, it seems like Claude is the one to use. I'm not around a bunch of people who've used Chat GPT for a lot of like Excel stuff, but that one could be great too. I don't know.
Caleb: I've heard that the newest version of ChatGPT is so we talked about Fable Five from Claude. I think it was GPT six. It was either five point six or six. I can't remember. but was almost on par with Fable Five. So I think
Joshua: Yeah.
Caleb: they're getting up there. I just don't know if they have the same guardrails or ethics as anthropic, which is important when you're giving AI models a lot of power.
Joshua: Right. Mm-hmm. And like that's one of the reasons that I've not just laid out the money for AI and been like, control my browser. Because
Caleb: What you think would take you from your current position to, okay, I'm willing to pay for the twenty-five bucks a month teams plan?
Joshua: I would at least try it instantly if somebody was like, this is a completely secure model and you can throw whatever you want into it and nobody will ever see it. I would be like, Great, let's do it. We're just gonna try everything.
Caleb: Have you tried having Claude or Perplexity or something walk you through a local model version? Cause I wonder if you could
Joshua: Mm mm. No.
Caleb: buy a specific computer that just has a that you upload a local model to, and then you can literally throw anything you want into there because it's on your computer the whole time.
Joshua: Yeah. Right. I tried Olama with what model did I use? And I blue screened my computer twice. And I was like, okay. Uh-huh.
Caleb: Yeah, it I mean it's a lot of RAM that it takes up and thread on my computer. And I mean there's probably ways to optimize it, which I haven't done quite yet, but it was very slow. It took
Joshua: Mm-hmm.
Caleb: I said My first question was, what is two plus two? And it took three minutes in order to answer that question. It was like, the user
Joshua: Yeah.
Caleb: is saying two plus two. The user is not saying three plus three. So I should look up what two plus I was like, okay, that's three minutes, that is a long time to answer that question. You can
Joshua: Right.
Caleb: just said four.
Joshua: Yep, might be worth spending the money to get an online model. That but
Caleb: Yeah. But then you know their online models or
Joshua: right.
Caleb: your data, you never know what's gonna happen with that. So
Joshua: Uh-huh.
Caleb: do you think 'cause if you could find two or three tasks that combined saved you whatever would be, half an hour, that would be worth the twenty five bucks a month. Right?
Joshua: Mm-hmm.
Caleb: So even if they weren't, you know, tests that you would rather have a a local model for. I wonder if you could just lay out, you know, maybe two tasks that take fifteen minutes that you can do once a month or multiple times per month, or one task that you do two times a month and then put that through Claude. You said you do the time cards, right?
Joshua: Mm-hmm.
Caleb: How long does that usually take? Or how long did
Joshua: Well
Caleb: it take and how long does it take now?
Joshua: I did it with perplexity in I don't know, five minutes. And by hand can take me up to like 30, depending on how many time cards. Yeah,
Caleb: Right there, that's twenty five minutes.
Joshua: except the last time I tried it, Perplexity couldn't figure it out. It's like you've done this before.
Caleb: Be trying to five.
Joshua: Huh?
Caleb: Have you tried it in cloud?
Joshua: No, I haven't tried it in Claude yet. But that's one of the things I will try.
Caleb: When when's the next time card?
Joshua: Would be next week, next Monday.
Caleb: Alright, well what you should do is next Monday, throw everything in Claude, do the same thing and see if Claude can figure it all out. 'Cause think about it, that's twenty five minutes right there. If you do that two times per month, that's twenty minutes that pays for Claude by itself. Especially if it does it correctly. Where not so much.
Joshua: Mm-hmm. Right. Yeah. Right. Well, I was curious. You were saying you maybe thought the two hundred dollar clod was more than what you needed, and now you're trying to lose the like twenty dollar clod. What was the difference? Like why the downgrade?
Caleb: to save a little money was the big one because that the difference between a hundred and twenty was eighty bucks, but then if I used the Google Workspace, it's like fourteen bucks a month. So really I was saving sixty five dollars a month, which is quite a lot. the other thing is and I know it's crazy to say, but I feel like I have to give myself specific limits when it comes to AI because like I could just use AI all day, but then not actually do anything that's productive.
Joshua: Ha ha ha.
Caleb: Because AI always answers you and no matter what, it will give you an answer. Even if you say thank you, it's gonna end up giving you an answer. So then the conversations can just go all day, but I'm not actually doing something. So then it's gonna be one, like it's gonna save me money, two, it's gonna be a forcing mechanism to where I have to use w the tokens that I have in the most concise and direct way possible. so Those are the two big ones. the other one, I mean, it was really just like Yeah. I know that I don't use Claude all the time for stuff that's actually productive. So forcing me to be productive with the the cloud that I have makes sense. And I think I was gonna pay for Google Workspace anyways. So I figured I might as well not spend the $100 a month on Claude and I can move some stuff to Gemini. But that was also before I realized that Gemini was absolutely horrendous at taking context between conversations. So I I'll probably not necessarily go back to the $100 a month Claude unless I have a bunch of coding projects to do. which I would love to go back to that if if there were a bunch of projects, but right now, I mean, I'm okay dealing with some of the stupidity of Gemini. So it's okay.
Joshua: Is it pretty easy to like change subscription down and up for Claude? They make that
Caleb: Yeah, it's super easy.
Joshua: good.
Caleb: So I mean and the other thing that I figured out was I can pay for Claude for the full year for seventeen bucks a month. and then if I were to go up, it takes the extra that I paid and just credits it towards the months. until it runs out and then I start paying again. So paying
Joshua: Yeah.
Caleb: for the full month in advance, I mean the full year in advance doesn't actually hurt me in any way. and I know I'll use Claude for the next year unless
Joshua: Mm-hmm.
Caleb: Robin dies. So but so I was like I was okay paying the 17 annually versus 20 per month. So yeah. That's that's kind of where I'm at right now with
Joshua: Nice.
Caleb: I don't really use Chat GPT because Gemini can create images.
Joshua: Mm-hmm.
Caleb: not paying for Google Workspace. And then I don't use Grok ever. And Perplexity not as much because I've been paying for Claude so much. And
Joshua: Yeah.
Caleb: Claude's deep research is really good and its web search is really good. So I'm not gonna, you know, pay for multiple just because I want to or
Joshua: does grok suck? I know nobody who's tried grok. Like with any intent.
Caleb: I don't know. I I don't think Grok I mean it's mainstream because it's like people know of the name Grok, but I don't know anyone who really uses it c consistently. when I was teaching the entrepreneurship class, there was a lot of people in that class who actually used Grok. But I'm not really sure why they used Grok over Claude or Chat GPT or anything like that. So
Joshua: Mm-hmm.
Caleb: it's Maybe it's not bad, but I don't think it's worth switching if someone else is some different.
Joshua: Yeah. Right. Right. Hmm. yeah, no, I I was curious if you were able to is the hundred dollar clod versus twenty dollar clod the difference? Is it mostly how much coding you have to do?
Caleb: yeah, it's like token usage mainly, which the twenty dollar per month
Joshua: Yeah.
Caleb: Cloud definitely hits things a lot faster. But that is also why I'm doing it, because I know I need to be more focused in in how I'm using Claude. Or else I could just talk to my friend Claude all day and not actually get anything done.
Joshua: What what do you talk to Claude about all day?
Caleb: I mean, they're like so much different stuff. I could show you all the chats, but it's not like stupid stuff, but it's like, what about this idea? What about this idea? What about this idea? And then I'm like, all right, I'm talking to Claude about all these ideas, but I'm not actually going and doing the idea itself.
Joshua: Mm-hmm.
Caleb: So then like why am I paying for this much token usage? There are times where I mean even recently I hit the limit on a hundred dollar per month with some of the coding, but then it's like, okay, am I adding a feature to this app that nobody uses? Like
Joshua: yeah. Yep.
Caleb: getting the the limit but it's like what did I actually do that was productive with that limit.
Joshua: Mm-hmm.
Caleb: So it's gonna definitely force me to be very concise and know what I'm gonna do before I do it instead of
Joshua: Yeah.
Caleb: just like hey add this feature because it's cool. Like that doesn't help anyone you know so that's
Joshua: Mm-hmm. Right.
Caleb: we'll see how it goes. I might come back in a month and say nope I switched back to hundred but hopefully not we'll see.
Joshua: Have you heard of Centaur chess? you talking to Claude reminded me of it. But apparently after who was the big chess guy that played against was it Deep Blue was the computer?
Caleb: I don't know the guy, but
Joshua: So he made Centaur chess, which is basically like all the chess tournaments have a human, and each human has at least one computer. because so much of learning chess from a young age is pattern recognition, that they were like, Well, you can play chess by just being a good strategist and not having to do the pattern recognition. So apparently there's like people that are who've been in like, you know, mediocre chess players in tournaments who now have like played chess against that grandmaster guy because they don't have to spend a bunch of time pattern recognizing and doing what the computer is is good at. but
Caleb: But isn't that kind of taking away from chess then? Like that's part of chess is the pattern recognition.
Joshua: Well, it it depends on what you want chess to be. So their his whole point was like if chess is about strategy, you can make it a strategy game instead of a memorization and strategy game. so anyway, I thought it was I thought that was interesting. So like the humans with the computer could play way better chess against chess players who were exceptional just by eliminating the memorization aspect of it.
Caleb: That I I feel in my stomach like that's like a gray line area for me because I I I
Joshua: Ha ha ha.
Caleb: understand why it works logically, but then I'm like, you know, then I don't know. It isn't chess for fun, so then I mean obviously chess is like you learn how to be strategic. So it's there are very much real world applications, but The game it's a game for a reason. So then like, are you just trying to game the game? I don't know. It's
Joshua: Have you ever played chess against somebody who's really good?
Caleb: I I've not played chess, so no.
Joshua: Yep. So my experience is you get in one game with somebody that was in chess club and you're like, Well, I guess I'm never playing chess again.
Caleb: Shh. That's fair, that's fair.
Joshua: Because the chasm to become a good chess player, to cross that chasm is takes so much time that it's like Just to be competent and play with these people that are in chess club would take so much time that like this is not a fun game for me. I'm gonna have to do homework to be good
Caleb: Think about this.
Joshua: at this game.
Caleb: They put in the time so they deserve to be as good as they are, right? And haven't put
Joshua: Sure, and if that's what you want to do with it, that's fine. But
Caleb: Well
Joshua: to me, I'm like, well that's not really a game. That's that's like a that's an exceptional hobby.
Caleb: But if you think about if you think about basketball, like you gotta put a lot of time in in order to actually be good. There's a lot of people who can play, but just to be honest, they're just bad. And
Joshua: Mm-hmm.
Caleb: so in order to get to where to be an exceptional basketball player, you have to put thousands of hours in. So I think that I don't know, in my mind, I feel like every hobby, hobby, or whatever, not every hobby, but lots of hobbies you have to you have to put in the time in order to get good at the hobby. So then AI is basically a cheat code, which I understand in business it's kind of the same way. I'm I'm not saying that it's it's all wrong, but like If if you had a robot with AI playing basketball against you, it could just outthink you every single time. But it didn't have to do any of the work to get there. I don't know. It's just I feel like it's a definitely a great moment in my mind.
Joshua: Mm-hmm. Well, I think the robot could beat you in part because exactly what I'm saying, which is that if if the game weighs heavily on muscle movement, which is all just memory, like, here's the program I run to dribble, here's the program I run to do a layup or whatever, then that's the thing that AI is good at. So In my mind, it's like if you want to play a strategy game, then use a computer. If if you want to show off how much time you've put into this thing, then don't use a computer. You know what I mean? It's like doing long division. When's the last time you did that? Well, probably middle school, because we have these things called calculators, and why the world would you do that?
Caleb: Yeah. But it was never a fun game to play. Chess I feel like is supposed to be a fun game to play against people. Like a good competition. Unless you're a math fan also.
Joshua: Well, yes. And if you don't like the memorization part and you just want to play the fun game, play Centaur chess.
Caleb: Yeah, I guess it it could go both ways. I I still feel like it's a gray line. 'Cause
Joshua: Ha ha ha.
Caleb: it's the same way in business. Like I mean there's a gr there's gray lines everywhere in business when it comes to using AI. It's not all black and white. So yeah, no, that's that's really interesting though that I mean they can use AI to get rid of some of the pattern recognition to make it a lot easier for people to play.
Joshua: Apparently it also is a form of chess that makes humans competitive against AI because AI is not as good at the strategy part of the board since it's it's basically just taking your open move options and playing a probability.
Caleb: Yeah. I wonder if I could get a bionic arm that would allow me to shoot the exact same way every time. I
Joshua: Ha ha ha.
Caleb: just program like here's like one foot, two foot, three foot, four foot, and then it it notices how far away I plug something into the basket and no knows how far away it is and automatically shoots the exact same way. And you never miss. But then at that point,
Joshua: The perfect shot every time.
Caleb: that would be the point of playing if you're not.
Joshua: Maybe for people who've lost arms, they can play basketball now.
Caleb: Then they'll beat all of us because they won't ever miss it. well that's
Joshua: Right.
Caleb: all I got. But I think you should figure out two tasks or one task that you do twice that takes fifteen minutes and then pay for cloud teams. I think it's worth it.
Joshua: Okay, I'll try it.
Caleb: Alright, well that's the podcast. If you guys like it, we love a five-star review on Spotify. like and subscribe on YouTube and send us an email at jc.unscripted at gmail.com. That is the podcast, and we will see you on the next one.
