Why Claude 3.5 Sonnet Is Better Than ChatGPT

I'm going to log in and show you how this can look. And the reason I also want to show you this is because I think quite frankly the app on the computer and the user interface and everything like that and the snappiness on the web and desktop app here is better. This feels smoother. Okay, truth be told, it feels smoother. You can go here and access settings inside of the actual application, which unlike with Chad GPT, you actually can access your billing and stuff from the application. Crazy. And as you probably saw for those of you that are Windows users, uh it actually works on Windows, too, which is crazy cuz usually they I don't know how to say this nicely, but JBT doesn't have that. Hi, my name is Demetri Bonichi and I'm a content creator, agency owner, and AI enthusiast. You're listening to

the AI agents podcast brought to you by Jot Form and featuring our very own CEO and founder Idkin Tank. This is the show where artificial intelligence meets innovation, productivity, and the tools shaping the future of work. Enjoy the show. I have to be honest, I'm a little frustrated primarily for the fact that Claude has actually been pretty good for a while and I haven't used it. I really haven't. I've used it to some small extents, but I haven't used it on a consistent basis, and I've been writing with it recently. And truth be told, I actually think it might be better than Chatty PT. And the reason I'm making this episode is because for those of you that are in the same boat as me, I think you should give it an on a shake. To go right off the bat, for anyone wondering, yes,

there's a free plan. It is okay. You can use Claude on the web, iOS, and Android. You have the ability to ask about images and docs and access some of the latest models or more specifically one of our latest models. It says now with the pro plan it's $18 per month which is a little bit cheaper than the chat GPT plus plan if you go on the annual side but on a monthly side cost pretty much it is the same cost in the team plan too. They're trying to stay competitive and enterprise you always got to speak to sales. So here what you get is Cloud 3.7 Sonnet with extended thinking mode, all of the models, and a bunch of really cool tools and integrations that I'm going to walk through in this episode that you honestly need to try out. So first of all,

when it comes to Claude, what it has above chatbt that you're probably unaware of is it has a style maker. So first we have normal, concise, explanatory, and formal. This is really important for a couple reasons. one which GPT you usually have to prompt saying in a concise way do XYZ thing or please explain to the fullest of your capabilities. Yeah. No, you can create a style and in that custom style what you can do is add writing examples and these are similar to how you can interact with GPT. However, it's pretty much just PDFs or pasted text documents that this works well. And to those of you asking cuz I'm working in the browser, the answer is yes, it does have a Mac OS app. I try to keep things off of my computer, but I'm going to use the Mac OS app to

kind of give it a fair shake versus chatbt as it does something that I like, which is have an installable app. I'm an app guy. I always enjoy having an installable app over just being on the browser. Even if I like keeping my computer a little bit clean, for the apps I use all the time, give me an app. I don't want it clouded in my browser. Haha, clouded. Get it? In all seriousness, I'm going to log in and show you how this can look. And the reason I also want to show you this is because I think quite frankly the app on the computer and the user interface and everything like that and the snappiness on the web and desktop app here is better. This feels smoother. Okay, truth be told, it feels smoother. You can go here and access settings inside of the

actual application, which unlike with ChatgPT, you actually can access your billing and stuff from the application. Crazy. And as you probably saw for those of you that are Windows users, uh it actually works on Windows, too, which is crazy cuz usually they I don't know how to say this nicely, but PBT doesn't have that. Or well, I apologize, it does have that, but it didn't for a while. So, they are competitive now, which is good. But it is on the Microsoft Store. And for some reason, it's rated teen for T1 for teen, which I don't really get how. But in all seriousness, I nobody really for a while has liked the fact that it's on the Microsoft Store, but that's besides the point. As you can see, from a profile perspective, a couple things to point out about Claude is the fact that it has

preferences you can set at the account level. So, I wanted to keep explanations brief and to the point or a bunch of other things like that with an entire guideline set that you can set here. But that kind of is a basic level you can have where this title exists or not. So I just want to point out pretty cool in this regard. Now some interesting notes about the models here. There's a lot less models. Uh 3.7 Sonnet is definitely the most intelligent one. And the difference between these models is essentially unlike Chachi PT where you have to pick between like 12 of them. There's a lot of them in all seriousness. There's let's count 1 2 3 4 5 6 7 eight options, right? No one uses these two. So there's really like five versus like they really need to combine the schedule tasks

into one thing here. But you got like four to five choices versus essentially two with sub options. We have Sonnet which is essentially your I don't know 4.5 I guess you could compare it to a little bit and then 3.5 which is kind of your fast model that's good as like 40. Now with deep thinking this is interesting. You have great math and coding challenges that you can use here that I find intriguing. So between explanatory and extended you can go kind of crazy here. So, what I find interesting in a couple different ways is simply how well this app functions. Snappiness, writing, all of the above. I think 3.7 sonnet is incredible. So, let's say for example, I want to take I made an AI terms and phrases blacklist. That's actually intriguing. So, if I add this here, I want to ask what about

this blacklist of terms makes it so much better at writing? So I gave it a hair. And a small thing to note, I have noticed less errors when it comes to working with the highest level of model in claude. A lot of times in chatbt, the prompts will just stop, right? I don't see a real-time ticker on how long it's taking in chatbt. It's kind of a different experience overall. A lot of times it'll just stop working. Not my favorite situation when I'm working in the app and it just stops working. But as you can see here, it says, "The document you've shared isn't exactly a blacklist, but more a catalog of patterns and phrases that AI tends to overuse. One of these terms can improve quality in several ways. Reduces formulaic structure." Okay. Eliminates cliches, encourages more natural flow, forces more precise language. That's

true. This does in fact make it the case. So based on this information, could you now write me a Twitter thread about the benefits of using Claude for your AI writing needs? So now it's going to make this thread and I'm going to do one more thing and just say please take this blacklist and make sure that none of the terms are being used to optimize the post. It's going to tell it to do better essentially. Hashtagbe better. Now, something to note, by the way, it says it's reached the maximum length. What is this? This essentially goes to the maximum length of its context window, which I'm glad that it told me cuz JPT use usually doesn't. All right, so let's take a look at the writing. Overall, I feel pretty good about it. I think that it sounds more human than Chetchip. This is

one of the big markers as to why do I think it's better. I I genuinely think it sounds better when I'm reading it. I haven't seen any of the magical or weird tuning that OpenAI did for 40 and the other models. I don't see any of the weird sort of unhuman terminology that you see in chatbt's writing which I appreciate. So after I'm done with that, I can go out and as you can see here, there are projects which I can add. And this is similar to Chat GBT. It does have projects as well. But a couple of features that I really like is they have cool integrations. They got GitHub integrations that are in beta. So, I can sign into my account there. And I actually have a Google Docs integration. So, say for example, I were to add my account, right? I could

have it interact with different Google Docs. Okay, that's intriguing. So, let's go here and I could search for recent or paste URLs. So, say I have a knowledge base of different Google Docs, which I do. I want to have it analyze my pitch deck and its talking points. So, what can be improved in this pitch deck and the talking points? Right. So, by the way, if I click on this, it does go to the Google doc. And it does seem to only work on docs that you own, not docs that you're in, cuz this is a more general account I use and it doesn't really have that perfectly. So, I'll analyze your pitch deck for improvements. Start with value prop. The deck opens with a linky introduction about myself. Fair enough. Interesting. I like the recommendations. Refine the problem statement. more concise and specific with

statistics or case studies is a good point. Can you replace the text by writing it yourself? See if he can do that. He's going to completely rewrite my pitch deck. Dang, man. If I click on this, by the way, this opens it up in like a canvas experience similar to what Chip has. Okay. Okay. Okay. Okay. That's pretty good. It even can recognize logos and stuff that's in there. It can even recognize what's in there. Okay. This is great. Look at that. It says what it did. It opened it up. Um, I could easily download this by pressing download a file. And what does it make it into? Is it make it a MD? Okay, so markdown file. All right, cool. Interesting. I like this a lot. I also can just copy it or I can press publish as well. I can't share it though,

which is weird. I can publish it, right? But I can't press share uh because it's an integration. So that's that's intriguing. And I can star the chat, which I like. Not something you can do in chat GPT to my knowledge. You cannot star chats. So, that's an interesting adjustment, but 3.7 sonnets's impressive to me. It's got some nice stuff going for it, and I really I really appreciate what it's doing. So, we're talking snappy, we're talking Google Drive integration, we're talking GitHub integration. I think it's got a lot going for it. So, overall, I am curious to ask, what are your thoughts on cloud? Do you think it's better than chatbt? Do you think it's worse? I think it's snappier. I think it's got a better functionality. I think it looks a little bit nicer. One thing I would note to point out is that

there's a bit of a visual difference to these. And I think this is more simple, right? And when you're interacting with different tools, I like simple. Sure, Chad GPT has a lot of stuff going for it, but between code and graphs and writing, I overall think that this is a really good competitor that a lot of people aren't talking about. The slight more human look to it as well is good. I think Chad GPT's got that corporate Microsoft look to it from when they got injected with that Microsoft money. Overall, good stuff. Let me know what your thoughts are in the comments down below and in the reviews on Apple Podcast. Thank you so much for listening to this episode and we'll see you in the next one. Bye.