Perplexity Computer Just KILLED Claude Code (Side-by-Side Test)

So the AI world is never still and perplexity introduced perplexity computer today >> which seems to be really interesting right from just by looking at the promotional material uh it's essentially you just ask it's kind of like open claw a lot of people are saying it's like open claw um but uh the ability is that they can put let's say for example stock analysis that that they're doing this here it can orchestrate the um entire procedure to um to a state that you don't really need to know any code. >> Yeah, >> I would say to be able to produce reports, produce dashboards, produce >> Oh my god. >> reports like this. >> Wait, so you're telling me I can ditch my open claw and just go for perplexity computer? >> Yes, kind of. Kind of. And what's kind of interesting, I guess the biggest difference from just building this on your uh cloud code is that you don't need to use you don't need to know as much knowledge it seems like. And also what's kind of important what they're showing you later on, right, is that you can actually time it as well. You can set up u essentially cron jobs which is what um open claw is really good at. So you can say hey yo you know if any video changes every like every month week give me a new report or do a follow-up right? So, um, you know, do a followup and monitor this for you. So, I feel like that's kind of their main advantage. Also, I would say that they're probably building this for a long ass time, too. And this was before OpenClaw. So, they kind of like came out with the OpenClaw concept, but like >> the Thunder got stolen by OpenClaw. >> Yeah, they're pressured. They got to put it out now. >> Yeah, they got to put it out now. Computer works, right? So, they they've been they've been very quiet because we don't um we actually bought >> the pro plan. Yeah. But we don't use it, right? Okay, it's a situation where we don't use it. So, right now it's available for you to try. Um, and um, it's actually quite interesting. I want to do this for this video. So, they gave a few examples and I actually want to run um, I actually wanted to run cloud side by side to see if you know what's the difference between running perplexity computer and uh, running it on something like cloud code. Yeah, right. That's kind of interesting. Or maybe ask your agent to do so. Right. That's but um, I don't think my agent is smart enough. I I think it needs cloud code to be honest because cloud code has optimizations for code and understanding code base right so let's let's let's let let it work through and see if it can build something of a similar nature >> but what's kind of interesting with perplexity computer is that it can run task in parallel so it's very good at this this orchestration where >> yeah these sort of optics are really what I would like we don't I don't personally see that for open claw and it's really tempting me to kind of switch to this but I heard heard it's kind of expensive. No. >> Yes, it's kind of it's not just kind of expensive. It's very expensive. So, let me log into this one. This is this this is my frustration. This is what what I got perplex perplexity computer. Like the the the the scammy part I would say about them is that we already have Perplexity. And what's kind of preventing me from, you know, throwing endless money at this is that we already have one of these plans, right? We already um I already have the pro plan. It's popular. Hey, this is good. But you need to max plan to go for it. >> Okay. >> All right. So, I don't want to I don't want to pay them annually. Okay. I do not want to pay these guys annually because I feel like >> um I'm already in a good position with claude code. Yeah. Right. Where I'm already paying these guys annually. I don't want to pay infinitely infinite guys uh infinite different services. That being said, of course, in terms of what people actually built and the the hype around that, what I saw, uh, yet again, this was released six hours ago and people have already started building something very good. They essentially built the entire Bloomberg um, terminal with this, which is wow, that's that's that's that's insane, right? If if you really think about it, a perplexity computer uh, P computer Bloomberg can't spell right now. I always have problems spelling on live streams, but yeah. So, um this is Hamptonism. She built what looks essentially like blue and blue terminal. >> Um and uh they use what's interesting is I guess why it locks you in their ecosystem is that they use Perplexity Finance to generate the information for you too. >> So, it's just like very integrated within their ecosystem. M >> but ironically speaking, of course, if you actually look at the tasks that they're they're building, they run uh Opus 4.6. So there >> they're still calling Opus. >> Yeah. >> Right. Is that surprising to you? >> Yeah, it is. >> Yeah. >> But to a point where it's kind of uh reinforcing me more like like Sorry. It's sort of making me want to get this more, right? kind of packaging and that's what perplexity was good for because perplexity back in the day they were very good at search right people were like oh perplexity is the best at search and I agree with that >> and then they went quiet for a while because >> you know I guess they I guess they realized that every other company is starting to do search as well and that the other companies are being better at it so the introduction of a pressy computer allows them to orchestrate the entire >> kind of process a bit better >> but I feel like it's also like a high IQ play like here in Hong Kong We don't have access to Opus 1.6, right? So why don't I just get this one? >> Correct. Correct. Correct. >> Use that. >> Correct. So actually Opus is already starting to build here. >> Okay. >> So I think the only the difference of Opus is that right now at this current setup. Okay. Um it requires you to I will say uh you know I just copy and pasted the command inside right um from the exact same prompt and you can already tell the difference with opus in a sense that opus >> didn't really understand the task at hand and it didn't do all the research that it did. So I think that's the difference it with perplexity where um it it knew it had a better orchestrator I would say it had a better team lead >> which broke down the tasks into subtasks and then send the agents out. >> You can manually do this in prop uh in cloud code. All right. So you can say hey use team agents to do this and we can make a tutorial video on that later as well where uh you can open a team uh you can open up the uh terminal and then it shows you multiple teams doing so but at the same time you can see that >> I say cloud just tried to brute force it whilst >> perplexity went for a smarter approach to solve this problem. >> Does that change your opinion on still sticking with cloud code? >> I I'm a control freak. >> Yeah, >> right. This is my problem. I'm a control freak. I'm like I like to see what the processes are along the way. >> So anyways, I I'll open this up to you guys. I mean, obviously this is very very early on. We'll do some experiments and we'll do some full tests on it, but we want to just get you out our initial thoughts on this uh on this, but it's actually very clear just if you ran the same prompt with Claude, >> Claude doesn't break it down as well. I mean, I'm so tempted to say, "Hey, use agent teams." Why didn't why did didn't didn't you use agent teams to do this? Oh, or plan it out properly. I'm so angry right now. You can sense my anger. I'm like, yo, I I want to show you that I I wanted this video to be, yo, maybe Claude is just as good. You know, maybe our conclusion is just that Claude is just as good. But I'm just like I'm so angry. And then and then Claud is like, you're let me show you this more clearly here. You're right. I I should have planned it out first, right? >> Yeah. >> So, Claude tried to I I maybe it's my fault for not entering plan mode, but yet again, as an average >> new user, I don't think you want to be entering plan mode, exiting plan mode. All right. Uh tell tell it to do um different uh sub agents. And I'm like, why don't you use agent teams? And he's like, yo, you're absolutely right. So I think like we can I I'm pretty sure like if I spent three hours on cloud code I can get a similar result right but that the issue is that three hours. Yeah. Right. I think that's the that's the issue there. And more importantly if you don't really know how to use cloud code >> um you know you're not going to be able to build your Bloomberg terminal and disrupt a billion dollar company or trillion dollar company. Right? So, I think that's a key difference there. In terms of spending tokens, obviously, I'm still spending a lot of tokens. It's actually quite con >> um it's actually not too expensive considering they're giving you credit. So, yet again, maybe if you're I guess if you're not already on the cloud ecosystem, I think it's just switching over to this might be a good idea. >> Yeah. >> So, I think like it just depends on what type of user you are. Um but I think it's a such a shame they just kind of like gated it. Like at least give us pro players that already pay for you guys. you know, we've been supporting you guys for such a long time. Why don't you just like give us one try at it, you know, >> but you know, that's life for you. So, anyways guys, that kind of wraps up our quick take on um perfecting computer. I really think that this yet again, this is really progressing AI. I think this is yet again something that will be endlessly attacked by Chinese uh companies to steal the secret sauce. So everything automates and flows it as well as how um it's doing things right now. This kind of gives us back the the madness days. All right. Everyone used Madness for a while. It's your personal computer. It can write stuff code. But but it kind of fell off because Madness didn't really have any special tech for it, right? It didn't have a large language model. >> Same thing with Plexity. Actually, Plexity is not using their own large language model. They're still using ones outside. They're just using they're just plugging it in. They realized that they didn't have enough money to beat, you know, Elon Musk or Anthropic. They didn't have enough money to beat on producing a better large language model. So, they just made a better agent orchestrator, I would say. Um, >> that sounds like like perfect for people like me then. >> Yes. Ironically, >> and in terms of cost, like you know, if I were to start from scratch, like if I don't use AI and then I'm coming into this game, >> the max plan for cloud is like 200 per month. This is 176. So you're saying, you know, >> perplexity computer is basically open cloud plus with open focus 4.6. >> Yes. >> So I save >> money >> $30. I don't know. I don't know. >> I don't know. I don't know. You guys decide. I mean, we're we're trying to make it as affordable as possible. You know, our on on our content here, yet again, I I want to see how popular this goes because like, you know, on our one, we're like, "Yeah, let's let's get your open cloud up and running for like $3." you know, like this is a game changer for like literally $3, right? This is um it's like, yo, um but yet again, Perplexity is asking, you know, >> the bank here a little, you know, pretty penny, you know, >> and it's actually $200 on a on a plan. So, we'll see. Uh my only disappointment was my my last plan was very useless. So, I don't know. Uh we we'll have to see about this one. We'll have to I I I'll run the I'll run this Claude thing. I'll try. You know, CL Claude's congratulating himself on a good plan already. Oh, wait, wait, wait. He's congratulating himself on a good plan and he's he's doing simultaneous. He's he's he's trying as hard as he can to beat Perplexi Computer. All right. Um but, you know, Yeah. He's running two agents at the same time. He he he's doing it. He's doing it. He's he's I'm just as cool. Yeah. you know, >> but Proper perplexity probably trained the orchestration of this a little bit better in a sense that >> um it's probably a a background uh understander for what the topic is and then Perplexi orchestrates it itself rather than just directly telling the >> clock code to orchestrate it. So I think that's the only difference they have there. They probably have a few more templates inside that work really well for them. I would say that's kind of the key difference here. >> So yeah, that's a realistic look on perplexity computer guys. Thank you guys so much for watching this channel. We read every single comment. Leave comments down below on what you like to see. If you are willing to pay, we are also willing to pay. Don't worry about it. It's just that my cloud bill is really expensive right now. So yeah, with that guys, I want to try every tool out there and subscribe more so we have more money >> to to make videos for you. Yes. >> All right. Uh with that guys, thank you guys so much for watching. I'll see you guys next time. Peace.