Reviewing Claude Opus 4.5

0:00Guess what? Claude Opus 4.5 is here. And people are saying it is the world's best coding agent. But I needed some proof. So I put Opus 4.5 head-to-head with Gemini 3 Pro. I built a SaaS app and I wanted to see who would win. And the results might shock you. I did it with my friend, The Boring Marketer. and in this episode we are going to show you how opus 4.5 stacks against gemini 3 pro and also how you can get the most out of opus 4.5 hint it has to do with setting up clod skills and we show you exactly how to do it enjoy the episode we got the boring marketer on the pod james by the end of this episode, what are people going to learn?

0:55Well, I'm just going to open the curtains and show everyone sort of how I'm using the latest models. I want to dig into Claude Opus 4.5. I want to jam a little bit on Gemini 3.0 Pro and want to show people how I'm building landing pages and websites as a non-designer, non-engineer that are driving great conversion. Can we live play with 4.5, Opus 4.5 and see what happens? Let's do it. Let's do it. And I also think something that would be kind of interesting is if we did sort of like a test. So let's use Opus 4.5 with the front end design skill. Let's see what we can come up with. And then let's do Gemini 3 Pro with the same prompt.

1:42And let's see what we can come up with there as well. Let's do it. All right. So here's a way we could do this, Greg. You want to get an idea from Idea Browser and run that and see what it comes up with? Sure. So do you want me to give you today's idea of the day? Yeah. I actually really like this idea. Estate dashboard for keeping families informed during probate. Estate executors face constant phone calls and texts from anxious relatives saying, what's happening with dad's estate? Estate Clear creates a real-time family dashboard where executives can post updates, upload documents, and track progress while family members get instant notifications instead of playing phone tag.

2:25I mean, it's one of those small ideas. How many estates are there? Not millions in the U.S., at least. I don't think. Actually, maybe there are. I don't know how many, but like I've, I've gone through that process before and you know, it's a black box and it takes over a year. You have no idea what's going on. You're getting these replies from lawyers and stuff like that. Like give me the, the door dash visual. That's like, just letting me see what's going on, you know, like that's totally, yeah. And you'd probably save a lot of like money on billable hours to attorneys going back and forth, asking them questions and stuff.

3:09And the retention is probably insane. Yeah. You're not, you're not going to cancel that. All right, let's get to it. All right, cool. So I'm just going to paste this into here. Kind of the full thing that you, that you sent me. So it's truncating it, but you get the idea. So let's see. I want to create a conversion optimized landing page for my new startup. Build a.html that I can view locally. You must use the front end design skill. to generate the page. So typically what I would do here is I would put this whole description and actually let me go ahead and paste this here just so people can can see it.

4:13Typically what I would do is I would do some foundational research you know like I would put this into there hey use the perplexity MCP tell me who is in this space what's a unique kind of differentiating angle that we can take, blah, blah, blah. But we really just want to see, hey, what do these models do with minimal direction if we just give it the concept? So you must use the front-end design skill to generate the page. Let's just start there, and we'll get a gauge on design and what it cooks up. All right. So meanwhile, I'm going to go over to AI Studio, and I'm going to give the exact same prompt.

4:59And could you use, like, do you have to go to AI studio? Could you not have used cursor and just done it within cursor? Sam Altman, the co-founder of OpenAI just said that it is the era of the idea guy and he is not wrong. I think that right now is an incredible time to be building a startup. And if you listen to this podcast, chances are you think so too. Now I I think that you can look at trends to basically figure out what are the startup ideas you should be building. So that's exactly why I built idea browser dot com. Every single day, you're going to get a free startup idea in your inbox, and it's all backed by high quality data trends.

5:41How we do it, people always ask. We use AI agents to go and search. What are people looking for and what are they screaming for in terms of products that you should be building? and then we hand it on a silver platter for you to go check out. We do have a few paid plans that take it to the next level, give you more ideas, give you more AI agents and more almost like a chat GVT for ideas with it. But you can start for free, ideabrowser.com. And if you're listening to this, I highly recommend it. You can, you could use the cursor agent there. um however i just don't want to create some like confusion of it working on like the same things i want to try to keep this like a i don't know a unit test or something like that where we're kind of isolating variables um so let's uh we'll do it this way and we'll see okay so right now uh Claude is working on our design already.

6:44Let's show AI Studio and we'll get it cooking there. All right. So let's go back to the start. Let's just paste this in. And let's get my prompt that we used for the actual page. and we said, I want to create a conversion optimized landing page for my new startup, build an HTML that I can view locally. And I'm going to remove the piece about the actual like front end design skill, because we know that Gemini already has those skills. So let's let her rip and let's see what happens. what do you think's going to happen like who do you think's going to win i think that if i were to create a prediction here i think that gemini is going to come up with something more outside of the box i think claude is going to create something a little more of like what you'd expect.

7:54I think both will be good, but that's just my gut vibe. What do you got? What do you think? I wouldn't be surprised if Gemini put in AI features in the MVP. I've noticed it's been doing that. I think that's interesting. I think that my guess is Opus is going to come up come and do like a very solid fully featured sass app but it's going to be like more kind of similar what you're saying it's going to be it's going to be similar to you know what you'd expect but i think it's going to be really really solid yeah um whereas the gemini might have like a couple bugs and it's you know explored edges that we probably didn't even think of yeah Yeah.

8:41After we look at these initial ones too, I've had a lot of fun lately also just prompting the model and saying, all right, this is cool, but like come up with something completely outside the box and seeing how creative each model can get. So after we see the initial ones, we'll give that a spin too and see what happens. Cool. Okay, let's see where we're at. So Claude is currently doing its thing. And here we go. So this is Opus 4.5 with the front-end design skill. One shot. We have the description from Idea Browser and literally like a one to two sentence prompt. Stop answering the same questions from everyone in the family.

9:23So good. Yep. Estate executors face constant calls and texts from anxious relatives. Estate Clear gives you a private family dashboard where everyone stays informed without overwhelming you. So I like the overall like aesthetic of this. I mean, going through probate and doing all that and dealing with this in the family, like isn't exciting. Right. So like it's got like a nice sort of like subtle tone to it. It's not like screaming in my face with a bunch of like wild colors and stuff like that. clearly it is not uh just using the same kind of like purples and typical ai stuff so let's scroll down and see what else it did so this is your skills that you created hard at work right like from a copy perspective yeah this is so i only told it to reference the design skill actually so okay i didn't explicitly ask it to reference any any of the the copy skills um so i don't think that it's actually using those i think this is just what opus came up with okay cool yeah so uh we can we can go through it though and and i'll try to optimize the copy with those skills and see what it does um so i think this is really awesome like this is definitely like all you need for like an mvp or getting something like this live.

10:54And what happens if you click star free trial? Let's see. So probably nothing yet because we only built the landing page. Oh, okay. Yeah, we only built the landing page. Yeah, if we were actually wanting to build out this app, it would take a little bit more time for Claude to do. Anecdotally, it does feel like we've, with Opus 4.5 and Gemini 3, it does feel like we've entered a new stratosphere of vibe coding. Would you agree? I would definitely agree. Like, I think that the lines are just blurring across the board. Like, you know, the line between a non-technical person being able to code blurred.

11:41And I would always feel very frustrated with like the design portion of this whole process, You know, yeah, you can build stuff, but everyone can tell it looks vibe coded or just doesn't look right. Now that's that line is getting blurred. So all these lines are just blurring. And, you know, it's not necessarily like vibe coding or vibe marketing or vibe designing or whatever. It's just like vibe building in a way. And the entire stack is just kind of coming together into one place, which is nuts. All right. So there's Claude. Let's check in on Gemini and see what we've got there. All right. Here's Gemini's version.

12:27Keep the family united goes hard. That does go hard. Yep. Manage the estate. Keep the family united. Stop playing phone tag. The private dashboard for executors to share updates, documents, and milestones with the whole family at once. Very clear. Clear, but I don't know. It doesn't have a... I liked some of the animations on 4.5. I don't know if you noticed that. Yes. Like the box on the right can add a little movement. I also like the layout of the 4.5 one. I can get the copy and I can get that initial visual like right away without needing to scroll or scan down the page. All right. So we've got this here.

13:15Social proof. This is very similar between both models, like this section. Here it is, the AI. Don't know what to say? Let AI write the update. You called that for sure. Yep. But I think that's really interesting. To me, I probably wouldn't have thought of this. And it's also, you know, if you're trying to create a startup right now that, you know, gets traction, I think having some of these AI first ideas really help. Yep. This is cool. It already built a little kind of like preview of some functionality here into the landing page. So I like that idea a lot. I like the imagery here. Feels more human, I guess.

14:10which is also something that coding models have missed in the past. Everything feels very robotic. You've got the emojis. You've got the same style of illustrations and stuff like that. And I think with Nano Banana 3 Pro coming into the fold too, Gemini has an incredible asset to leverage with its design capabilities for, you know, things like this. We've got a very similar kind of like pricing card between both designs. So, man, it's tough. I think both are good. I think from a pure aesthetic standpoint, I think I like the Claude version better. That's just me. No, I think you're right there.

15:04Like, I think objectively it was nicer. I think from the feature perspective like I said, the AI feature is really interesting I'm impressed that it worked too I mean my real question with 4.5 is okay great, now it's made a landing page but can it create a SaaS app?

15:27Ultimately I'm interested in pushing 4.5 to the limits and I would be curious if it actually could build an app that I can sell? It absolutely could. Um, I mean, you know, if you wanted to build this entire product and Claude code with 4.5, you, you could definitely do that. Um, you know, like I, like with, with Claude 4.5 Sonnet, like I've taken, um, you know, uh, idea for a SAS from, from zero to production ready. So you can certainly do it with either of these models. Now, what I have personally found, and to be clear, I haven't tried to build a full on application with Gemini 3 yet. But I think that Claude's earlier models, like I would always go back to them.

16:24Like when I was, I was trying to use Codex and GPT-5 before 5.1 came out and, you know, Gemini 2.5 Pro and stuff like that. And while the logic is good and the thinking and the debugging is good, getting from zero to fully ready was easier with the Claude models. That's just been my experience. So I I always kind of try to lean on Claude as the workhorse and then have these other models doing what they're good at as sort of advisory model, if you will. So here they are like side by side. Now, I mean, if we wanted to, we could we could see how each of these builds out beyond the landing page and just say, hey, create a clickable prototype for for the concept.

17:19And we can kind of get a gauge on how it's thinking there. Let's do that. Okay. Because I'm curious. Let's do it. Let's do it. All right. Let's go back to cursor, and we will get started there. So I want to build the full clickable prototype of the app locally so that I can open up a.html and see how it looks and feels. And if we wanted to build, I know we're not going to do it today because we don't have two and a half hours. Yeah. But if we wanted to build the back end, we could. Oh, absolutely. I mean, we could set up the entire back end. So some of the things that I've used in my most recent builds are like Neon, which is kind of the database, plugs in nicely and plays well with Vercel.

18:22clerk for auth and login, easy to spin up OAuth or anything like that. I would probably deploy onto Vercel or Railway or something along those lines. You could integrate Stripe for the payments and the subscriptions relatively easily as well. And you could build the entire logic and application here for sure. so I'll give the same prompt, uh, while we're waiting there to, uh, to Gemini, you know, one of the things that I, that I really like about Gemini and, and just Google generally is how like, and we've talked about this a lot is just how vertically integrated it is. Um, now developers and stuff have their preferences for what tools and what things they want to use like at different levels of their stack.

19:27But if you think about like 90 % of people, like someone who's just trying to build their app or get something out there or whatever, having everything like integrated into AI studio from, you know, auth to your storage or database to easily integrating AI models into your product to hosting, et cetera. Google has the full range of capabilities there that they own that they can integrate directly into like AI studio, which I think from just a convenience perspective for most people is a really powerful value proposition. And from a cost perspective, from a speed perspective, from just keeping things organized perspective, that's a huge advantage that they have as they push this forward.

20:22Totally. I mean, I don't think I'm going to switch from my iPhone, but I've never been closer to switch to an Android. It's pretty crazy. Something else that we've talked about is, you know, not only do they own the foundational model, you know, the Gemini model and all these other tools and stuff, the data that Google has is insane. Obviously, we all know that, you know, a lot of people use Google workspace, like to basically run their life, like their email, their calendar, their documents, their spreadsheets, whatever. And so you have that whole layer, that application layer and the tools layer.

21:04And then, you know, you've got the devices, like you mentioned, but beyond that, like the TPUs and everything. Now I'm not like a semi conductor expert by any means, but what I do know is that, you know, they work with like Broadcom or whatever to manufacture like their own chips. And it's a lot cheaper, which is something that Google loves is being able to offer, you know, undercut the market. And like owning this entire stack is just crazy to think about where it's all headed, you know? And they also own, I think, 14 % of Anthropic. True, true. So they're both kind of Google products at the end of the day, right?

21:47Yeah. 4.5 and Gemini. Yep. Have you been playing with Nano Banana at all, Greg? i have um it's you're talking about nano banana pro right yeah the latest release yeah it's significantly better than the first version then the the number one thing i noticed is just text is better yeah so it was so frustrating with nano banana that you know i'd create a youtube thumbnail and it would be like so close uh but it would add like an extra o and i would be like Take out the O and it would be like, no. It just wouldn't be able to do it. And there were some tricks around that, creating a new chat using that image as a reference image.

22:32But with Nano Banana Pro, I'm just finding 50 % better outputs. I'm also using glyph.app to help get the most out of Nano Banana. by the way no affiliation with Glyph no affiliation with Google no affiliation with Anthropic just tools that I'm using that helps me prompt Nana Banana Pro a little bit better it's kind of crazy to think about obviously the use cases is crazy to think about there's a bajillion startup ideas that I'm thinking about right now specifically around Nano Banana Pro. For example, I saw someone created infographics. Did you see this? This guy who created the infographics startup?

23:30No, I haven't seen that. I don't remember his MRR from it, but basically so many people want to create infographics. Would you pay$20 for an incredible infographics? Like maybe if it helps you get 50 ,000 impressions on X. Right. So he just optimized Nano Banana Pro for infographics. Got a good domain name, created a good brand around it. And like what are other categories for that? Yeah, yeah. Something I was using it for, you know, Black Friday is coming up or whatever. I want to do some like retargeting campaigns. I want to run some ad campaigns on Meta. And I saw this format, maybe this is a little sauce for people, like similar format to this like infographic approach.

24:23But I got these ads on Instagram and it was kind of like the one page map to accomplish this. Let me show you actually real quick. I'll just pull up some of the little ads that I created because I actually thought they were pretty good. Um, so my idea for the ad, uh, is kind of leaning into that infographic style. Um, how can I like give people value like in the ad itself, you know, uh, before I ask them to do something. So we've been doing a lot of these like free workshops or whatever. Uh, they've been doing well. My, my concept was, Hey, let's take some like short clips or replays from the workshops and some key takeaways people should know.

25:09let's run an ad, drive people to each of those, see which one, you know, converts, helps us build our list and then, you know, convert people into the community or whatever. So this is like one of the ads that I, that I created. So I've got one of these for each of the workshops. And then I've got sort of like a transformational ad that's like before vibe marketing, after vibe marketing as well. Like I try to see what's going on in like the direct to consumer e-com world and figure out how I can apply it to, uh, to my world. Um, they're testing way more ads than any other industry. They're learning really fast, borrowing some ideas from there and then translating it into like, you know, sort of the creator world is I think there's some alpha there.

25:53Um, but yeah, what do you think of this ad on a stake? Do you like it? Do you not? Is it too, too much text? To me, it's too much text and it's too busy on the negative side. But what I do like about it is I find the recordings in, the AI finds the moments, the saved layouts snap into place. The titles look really great. The call to action looks really great. The visual style I think looks pretty good, like that handwritten style. So I think this is for me one prompt away from being a really, really high-performing ad. Yeah, I agree. Yeah, and I think just those barriers, you know, like, oh man, I'd love to run a campaign, but you know, I've got to go get a designer to make some good ads for me.

26:42They have to know how to design good ads and all this stuff. Like those just aren't blockers anymore. Like you can open up Gemini three, for example, and you can say, Hey, go and research the top performing ad formats for me across these industries, find the common patterns, look at my business and give me the hook, the angle and the visual style that I can go and paste into Nano Banana Pro. Like you've got the thinking and the testing and the data from all kinds of people that have gone and done it and scaled it and everything else. And you can take that knowledge in a snap, go create a good image and then launch a campaign that could potentially, you know, generate fresh sales or revenue for you.

27:27So just the speed you can do it and the barriers dissolving is just, I don't know, pretty mind blowing. So let's check on our progress. So this is the family view. First impressions here, like it maintained the aesthetic perfectly, like down to the font. You know, it's not your typical build a SaaS dashboard AI look, you know, at all. Like a lot of those always look the same. Like, I don't know if you've seen like clawed artifacts and stuff like that, like that sort of vibe emojis and everything like none of that. Like this looks really good. Here's kind of your progress sort of thing that we talked about.

28:15We've got recent activity, the different family members here as well. It went ahead and built out a lot of these sub views in this prototype. We've got the updates keeping you informed with the latest estate news. This is a really nice touch. We've got document storage, which is cool. Milestones. Just as you're clicking around this, it's like, why doesn't this exist? Yeah, it's a really good idea. A couple pages aren't built, but yeah, you get the idea. You can post an update, sharing the news with your family. This is a great idea. It kind of reminds me of like, what's that platform like Carta or whatever, like for your company, like fundraising and financials and stuff like that.

29:11There's also been a couple like family office as a product startups that have come out as well, kind of keeping all of that information safe and stuff. So there's some, there's some good sort of examples out there to pull from for this. The only thing is like for this, I would say, what's the recurring kind of use case, I guess. Like one way to think about it would be, okay, the estate clears, what's next in the journey, right? Do you need financial advisory or financial help? Do you have some sort of like estate law play or integration that comes next? So like this is one step in the journey.

30:01I would say to really improve the concept, it's like, what's the recurring use case after that? That's the next phase of life after this. I'm going to do a whole episode on vertical apps like this and how to structure them. But the hardest part of building a vertical app is figuring out what is the workflow, figuring out what is the social aspect, and then obviously figuring out what is the vertical and once you have these people in the network then the workflow might change but you know yeah there's some there is a workflow there that you can that you know you just have to figure out what that is and if you don't know what that is like you can show this to a family right you now have a prototype show it to them get feedback you could you know on idea browser, we have pro features that give you agents to help you figure that out.

31:01You could, you know, just take it one step at a time to suffer on retention initially. And then, you know, once you have customers just like build, build as you go. Right. Right. I think also, you know, maybe the first customer isn't necessarily the family, the first customers, maybe the probate attorney. right yeah you know like they've got the clients coming all the time and uh you know they can it's kind of like the b2b to c model if you will yeah so all right let's see what uh ai studio is doing real quick and uh let's get a gauge on so something that i've noticed is there's there's a couple errors here.

31:51So what do you do when you see errors like that? Fix error. How can you make this work? No. You know, for me, it's, yeah, just kind of a process of like trying to dig deeper into what the root cause is. So I use the terminal for all of the work that I'm doing. So if we think back to that cursor setup, I may say, hey, Claude, ultra think about this problem. You know, if you type in ultra think, it'll change colors and it'll use more, you know, power to kind of help to dig in or solve the problem. Other tactics that I've used with Claude, because that's like my daily workhorse, is I'll say, hey, if I'm really stuck, I'll say, hey, spin up specialized sub agents that are going to go and find the root cause of why this thing is breaking.

32:46I need like a Q &A tester. I need a senior sort of like engineer focus on this or whatever. And then it'll spin up these specialized agents that'll identify different areas that they're going to go and research in the code base. And then it'll come together with like a comprehensive sort of analysis for me. And then I can say, hey, all right, cool. Plan this in terms of like a step-by-step process that we can go through to fix it. If it's still stuck, that's where I'll use cursor agent with, you know, GPT or Gemini or something like that for kind of an outside perspective. Now, one of the things that Anthropic is kind of talking about with Opus 4.5 is that its ability to kind of stay on track is really powerful.

33:32Sometimes when the code base grows, it's harder to kind of manage the context to find what that single problem or whatever is causing that's breaking the app. So I haven't built something full with Opus 4.5 yet. I'm thinking that its ability to work through the bugs without sort of getting lost is going to be a lot stronger. And I wouldn't maybe need UltraThink or sub-agents or maybe even those additional agents on the side. So I'm really curious to dig into that and see. which is a big deal it's a big deal for your uh the health of your heart yeah for sure so here's our prototype um within uh gemini here so little bit of a different vibe um more like less dashboard more sort of like digest or updates or a space for the sorry it updated the updated the landing page a little oh no it's just a sizing thing sorry okay so um a little more of like a social vibe here uh like sharing an update or connecting with the family uh it's got the ai features like built in like polished with ai like if i want to update the family.