Hello, everyone! We're on ToTheMoon. Technology news and sites from the Silicon Valley.
They said the agent would cost $20,000.
No, well, I think it's stupid. I don't think it's $20,000, well, it's the future. It's just gonna be very strong as differentiation, yeah.
If that system comes up, it's some kind of top pilot.
It's funny, therefore, that this is the point and context behind.
You only take off the release, take a few days and new information, new update. Two big topics for the review today. It's the appearance of a new player and a new name. And the big topic is how AI agent can cost $20,000. It seems to me that this is a very serious perception of what's going on. Because two years tell me that... They're telling me two things all the time. Or all people lose their jobs and they have nothing to live on, or all people will stop working because AI will help them all and they'll have money. Here. $20,000 is making some adjustments, I think.
But start with, but start with Manus first. Is Manus correct?
I don't know.
Sasha?
Yeah, I, too, when I saw, I read Manus, but I didn't see the transcribing, to be honest. Yeah, Manus is a total breakup, of course. I mean, there's still no access. But all the cryographers are very different, very different, not only from the AI world, but from the medical world, they say it's a complete break. I mean, it's basically a system that brought together all the best products. So, a very complex system, she and deep research can, she and the entrance, the entrance to the generation can, and the computer operation can, and can be in a very long memory, and can be reasoned within hours. It's clearly the result that she's giving... Well, it felt like after looking at how she works, it felt like MBA wouldn't be necessary, for example. Yeah, well, I mean, if someone goes to MBA to get MBA to be, it's not his calling, yeah, but the man wants to be there at 30 years, and, like, at 35, he wants to be. To become a consalter and so on, I think it's almost impossible to have these systems. Because another step-step of the output, sequence and logic, it's getting staggered. So no concsalter I know has the same good as a related, related toeps. And this system, it can go very deep. So it can, for example, when assessing company technical reporting, it can be between a strictly complex technical analysis, it can do everything in parallel to a market overview, i.e., understand the competitors and I'm not sure what the product is. I mean, this system can't. The same is true in medicine, which is medical, when the problem needs to be diagnosed, there is often a tendency for a doctor to make a huge contribution to the wrong decision. I mean, a doctor didn't have any disease or study badly at the institution or something. And it's like an initial diagnostic that's very strong. I think these systems are definitely the next level.
And they say they didn't train new foundational models, but that's a good result. And this. And the second news, I don't know if she's connected to them or not, but it was clear that China is already completely auto--- well, there are fully automated bot-ferms for mobile phones, yes. I mean, there's a lot of bot-ferm skies. It's like creating a scout, for example, on the same TikTok, loading the spectros, cross-promoting these species between the acquunts, there, building the AStore, AppStore, Google Play Store, pumping the installs.
It's just a banal QA. I mean, it was clear that these systems were fully operational. And that's me, of course, which makes me feel like there's some reason why in the Valley over the past ten years, there's no successful concue in America for the last 10 years. product. And China has done a lot like TikTok, all these short-drums I think I mentioned last time, also Temu, Shein. And I think we're seeing the same thing. It just happens that it looks like there's a very, very little chance in the consumeria-requester business. But in principle, it was very interesting that Satya Nadell had called the OpenAI product company. I mean, he's probably putting them in the same competition with Manus and DeepSeek Com.
I'm curious to know what you think, but I think the OpenAI lost the leadership they had.
What's the price of Manus? Look, what's the cost? They made Chinese to just explain a little bit of the audience to the end. We can go in there now, on the invitian code. The invisible codes were sold there for tens of thousands of dollars or a thousand dollars. But I understand there's an adequate cost to the system, right? I mean, it's not that you're supposed to pay $50,000 in front of the house in the first place.
No, no, I think the system is adequate. I think, yes, it's not worth a lot, yes, but it's just not set up yet, and we're all gonna be waiting very long. I mean, of course, deep research is already actively used. I mean, without deep research, I don't use it. Grok, in my view, is very much overwhelmed, in my view, by the Grok, and by the GPT, and Claude, respectively. I mean, in my opinion, Grok is a big leader, better than Perplexity, too. And, well, this connection is, but with the operator, and who can write the code, run the analysis, because, for example, to analyze the runways, well, there's a market flow, without the code, it's really hard, actually, on some kind of thing. It's already complicated. So these levels are like-- well, these levels, there and how, angles, what? Well, it's not serious, it's just a hand-writing place, right? So, plus there, there are many statistics that can be turned to the reliability of the findings. Well, I think it's a great use of the case. Well, which is very impressive that they've done more work, more work than anyone in the name of the case. I mean, I'm on their website, the number of cases they've written, well, at least they've written.
That's, uh, incredible. So no American--
Yeah, yeah. I'm coming back. So Grok won his category, in my view, in America. Ilon Mac, how good is he? But I look at Manus, their use of the case they covered, it's impressive. I mean, well, they at least gathered it in the same place, they tested it. That's very, very impressive.
Well, they have fun, yeah, use case, and they're gonna be doing this case separately. So they say, for example, we're analysing the YC start-up data. Does that mean they're putting in a separate analysis and characteristics?
I think they're, I don't think I know. I mean, the biggest question, of course, is the interactive system or not, I don't know.
I mean, it's interactive or not, I don't know.
Yeah. I mean, here. So, it's a interesting story, how much these systems, what they showed, they say we have a case, and this use of the case, they're still based on a left-handed application. We often saw them use a case in OpenAI ChatGPT, but this use of case is not a function that is personalized. I mean, Manus, he's still, this story, where, for example, they say we analyze the YC, Y Combinator, and they're going to get some data back there, they're training the model, some conclusions. show and add up some materials feeding that it has more accurately drawn conclusions or, in general, the data that are available within the system is sufficient to analyse not only YC but rather analyse, For example, any business incubator in the world that produces data as much as YC. That's not clear yet.
Yeah. I mean, the most interesting thing is how, on my part, the most interesting thing is how far they're actually going to go in the real way of a proactive, proactive question. Because when people ask to analyze the flow or make recommendations there, or they or people ask to make recommendations on the start-up,
Step, uh, well, I think that, uh, that's what people really think, and, uh, you know, you should ask a lot of clarifications.
Well, they have interesting examples. Even we'll look at the moment of exit, how open this system is or isn't. But there are interesting examples on the website that can be seen from the perspective of the skies. Like when you come and say, "I'm a teacher of physics in high school. Give me, please, and I'll cook there, sco-- give me a description, there, the law of impulse. Can you set up different services and so on? So who's interested in seeing how it works? If someone's already using the system, we're using the people who look or get access here, it's gonna be very interesting to know what you think. And in general, write your opinion as to how you see these skies and how far you believe it is a piece of technology, a breakthrough.
I saw it, and, Manus, I saw the price detail first, and I thought I had done something ultra-exclusive and ultra-threat. I've had a feeling somehow. And, uh, originally. And there's still some news in the OpenAI. Sam Altman said that, uh, they'd let the agent out. And that's very interesting. If you remember last year we talked about before the pro two hundred-dollar version appeared, we were talking about their possible $20,000 agent per month or even 2,000. A dollar a month, some version will be worth, some version of ChatGPT. And they've released the description, and now they're saying the agent would be worth $20,000. Well, there's a range of decree-casane, but I want to focus on that $20,000 accent. It's a statement that shows that there will be... Well, first of all, it shows that the system will not be accessible to all. Secondly, it shows that many professions and what de-bes artificial intelligence will cost them much more than a normal person. Which makes sense, right? Well, the idea that makes sense is that there's some agent who--
Well, I mean, I guess, I mean, basically, you know that there's an executive assistant, a good executive assistant, and he still pays, like, 10,000 a month, like, for example. I mean, it's a thing that doesn't lose clients, which might be shorter, which understands the business, yes, but which, in fact, just sets out the meetings, yeah. The management calendar, but full management calendar and still doing some tasks in the physical world. I mean, I mean, you know, anyone, I think that it's exactly what it's like to think that any professional will spend, like, 10 percent of the professional's salary on the AI system. I-I'm-I'm having a hard time imagineing a world that's not gonna be like that. I mean, because it's like that, it's about that, it's about to work, really.
No, somewhere, if a person sells his services, for example, he can spend more. Given that you said a good thing at the beginning, that if now, conditionally, the person who is not a real executive, and he goes to get a MBA for some kind of crust, for a name, To become a consultant, the idea is that the people who really do this are dedicated to them, they will be extremely hard to grow through such systems, and more money can be spent to expand their capabilities, Yeah. If you've sold your time and you had to work for an hour, you can sell your hour and do it in ten minutes, probation, work, because you're gonna need an extra AI. Even now what they do, deep research, and if you can use these consultant systems, I think there are people in the world who, well, have earned more of their efficiency. But it's obvious.
There's a direction even vibe coding called when you write a code in some IDE classor, where you just write a lipstick, and you get it all generated. That's what's a special term called vibe coding.
Yeah. So many people still can't figure out, well, like, what, what, what, what, what, what, what's worth, what to buy, what to do easily, what's hard. Many things remain old prices. Someone comes in, says, this page on the website is worth a thousand dollars, although someone was doing it before on CMS in five or ten minutes. It's like the next level, yeah. You can tell that man, this algorithm is worth money, and he's got it in one minute or a half minutes, and he's already been. Another thing is, we're talking about some simple things like that.
In fact, in the consulting, I imagine, yes, if we're talking about the selection, like marketing or sales analysis, or the analysis of different systems in the business, or the creation of presentations. But it's just a nut job. One of the interesting cases we've been dealing with now is the case that's actually hitting. It turns out there are people who love to have a big presentation at the meeting. And it turns out people are wasting their time on big presentations. I'm not a fan of making big presentations. I'm all the presentations that make, I'm sending them to, uh, LLM-ku and asking for the summit to do. So what's going on? Which means there are people who say, prepare a presentation, use AI, write a lot of it, so there's a text, and then there's a man on the way out who makes the presentation and doesn't read it, - I'm not sure. So you could have straightened out a short summit, worked it out very efficiently, and the presentation would not be necessary. We're going to be in a very interesting world right now, yes, when a large number of people really start here, these big documents or the large volumes of data are going to be analysed briefly, yes, that's how the articles are, Like. I must be seven articles from ten now, and I'm very much in English, consuming content, seven articles from ten, I'm sending to the system and asking me to make a short summit. And then, if you need to, I'm studying something further on this subject. But in general, the summit is just, I don't read it-- uh, you come in any, totally any media, I read ten pages, uh, this article.
And here, uh, what, what, what, what, in the end, what, what, what kind of scenarios are gonna work, how does it win? Well, that's a very strong restructuring. We'll be back by $20,000 again. If Sam Altman claims $20,000, it's his speech again. Same as GPT chat four and a half. So far, by the way, GPT chat. Write in the commentaries who used four and a half. I, I, uh, stopped using four and a half. I can honestly say, I'm in person until four and a half of this is good, because every second time he gave me a bloke. Just, just a bra-- well, a left turf. And I'm gonna, uh, stop it. I'm mostly using, uh, O1 Pro and, uh, add-- someplace I use Deep Research, and somewhere I'm just going back to 4O because four and a half worked really weird. Uh, so write, viewers, comments. That's very interesting. $20,000. Sasha, from the point of view of the strategy, you're still in the market. Is that or are you going to have a skeptical attitude to Sam Althman's statements again?
Twenty thousand? Well, I think he should get to know Manus first, right? Then talk about $20,000. Still, it's all in the directions. I mean, now, well, they haven't had a major improvement for three months. Well, it means that there was, uh, a shipmas, and then there was nothing from the OpenAI. And basically, well, they're, uh, standing, well, they're not leaders in all categories now, are they? I mean, Voice AI, maybe they have the most adoption, but I haven't seen any lucky user. Everyone thinks he doesn't understand that, um, doesn't understand the intens, don't understand-- well, interrupts, uh, doesn't answer that very quickly. So there's no skies where the GPT chat, uh, has a significant, high-value, where the OpenAI has a significant advantage. So there's no pictures, no video, no audio. API is much more expensive than DeepSeek. And the reasoning is worse than Grok. Oops, the cameramen, the manus operator works, they don't work.
DeepSeek is--
About reasoning.
Well, I think so. Uh, no, I'm saying Deep, Deep Research.
Mm-hmm.
Yeah, yeah, yeah. Well, in my understanding, the reasoning is Deep Research. It's like a plan of what's useful, real useful.
Yeah, yeah. OpenAI, however...
What's the look on you?
Yeah, OpenAI is still Deep Research. He's kind of fun, but it feels like it's a standard that should be on the request. I'm just using it now. Maybe not a lot of requests a day, but I have a feeling that this is a standard answer. Here. Whoever Grok used Deep Research, you're gonna write down what you think about Grok Deep Research. I didn't use it very much, by the way. I'm gonna get this c-- there's a full-time version of Grok tonight. I'll see if I can use Deep Research in Grok for a few weeks, because I didn't use it very much. You, Ilnar, used it, didn't you?
I mean, $20,000--
Yeah, $20,000. $20,000 is just his statement to show you what he's doing to some spaceship.
And he's pumping, and he's moving just the window of Ebert. Yeah, well, he said that, like, 2,000 before. I finally made two hundred.
I finally made two. I'm just gonna swing at two-- on, on, on, $20,000, maybe there's two thousand going to be there in six months, and then maybe something can be released in a year. You want to add something? What, besides what Altman wants to do, and some tasks will be done or not, you have to add that engineering has to be some justification for that cost. Because you and I talked about a few episodes back, didn't we? What, the conditionality was Reasoning, then, in fact, DeepSeek showed up and made it all free. Same here. If there is no genetic engineering solution that cannot repeat the rest or some large number of computing capacities required to accomplish this task, Altman may I'll put any value in it. Then there's another player who's gonna make it just a little smaller, and it's gonna be worth, there's not two thousand, but, well, the same two hundred dollars, and it's gonna be all decided. So Altman can, of course, hack what he wants, but that must be all right. In this case, they have ceased to be monopolistic. And even like Sasha says, they stopped being the main player who's all over the baccalaureate upstairs.
So we wait here for him not only to claim value, but also to have some engineering justification why it would cost so much.
No, they just released API-- the operator- it's kind of a lame thing, right? They've released four and a half now, I don't understand. Well, I'm not saying, there's a lot of things. And Ia, see how Manus really works for the operator in terms of mass use, right? We still see some descriptions, briefcases, even though there are people who write that it's super cool. And, uh, maybe it's not for a reason that people spend thousands of dollars or tens of thousands of dollars for getting an inquit. Easily, uh, what is this system that costs that money? I mean, she should be, like, some kind of straight, well, super huge. And, Sasha, about the operator. Is there a skid in Manus operator now, they're designed to program or very much not to programming?
No, I mean, there's definitely a place. They're writing data analysis, which means a lot of code. I mean, data analysis, a lot of code.
Well, it's like you're on the deep research that OpenAI has, you're adding up the ability to write the code, to do the code, to collect some services.
And it is, and it decides that there is a need to collect services, and it decides that this service needs to analyse these data, and it is the service itself that analyses these data and does it further. Well, like, saying, like, that it's different, like saying, that it's not gonna be easy to do, right? And that's why they add--
Well, that's what makes this case a better approach.
Well, that's, uh, that's right.
Set up the schedules like Sasha says.
Well, it looks like some super agent, right? Well, now, from the point of view of the current time, that's like it.
Well, that's just that everything is well-equipped as it is at first sight.
Write in the comments, as correct.
Yeah, I mean, well, Sasha's about--
I think they're talking Manus in the presentation. I mean, I think they're gonna, like, not need them, so, uh, I mean, I think it's real, um, I mean, I guess they don't really need them, but, uh, in reality, Manus, well, they're not gonna be that way. I think it's a lot of money. Manus is also talking in the presentation. I don't think they'll be so much more charded.
I'm saying that Sam Althman said it specially to balance the installation, that you're kind of waiting for the Open-ended. It's very interesting in real marketplace to update. I mean, Elon Musk is Grok out a month ago and says that this is the best version, the coolest of all. OpenAI releases, there, deep research and goes on, well, it's still a renewal, in any case, sufficient update that's been made. Uh, and everybody's running around in parallel, and someone's coming in. Remember what happened to DeepSeek a month and a half ago when everyone said, "Oh, shit! Now, there's a new system. Everybody says, "Fuck it!" Again. And it's always very interesting how a new company really is in a way out of the other. Because, when DeepSeek came out, we saw they were out of the way, they did something, maybe something like a cheaper one, and that it could be easier to re-establish. They're not in superpowered, however. How, uh, now it's like someone who's gonna be in a bigger break, though I don't believe that much. Given the infrastructure, I think we need iron still to get a lot of people out of here to process all this data anyway, but... so long as we need a lot of this.
Well, then why would these investments be made for trillion dollars, er, electricity, factories, chips and everything?
When I hear about $20,000 for this kind of assistant, I always have stories in my head from Computer Vision. Oh, there are hogbows that open on a car cell, they don't open for any of them. I mean, let's say, open for those who have a number in the base or, uh, for emergency vehicles. So that thing works qualitatively, no matter what weather, lighting, and other than a system that recognizes, there are still some operators in the depths behind the low-cost money that are controversial. The situation is solved by hands. I mean, some unremunerated staff who help this system work properly. And now that we have a budget of $20,000, I still have doubts in my heart that it's a complete AI system, not a conditionality, there are thirty, forty-north high-paid people out there. The silicon valley, and from another part of the planet, which sort of things out for this system. Not well, not badly, in terms of what you're doing, you're paying for it, but this is a story that doesn't really get it. It was like, you know, a bike in competition, uh, cars or trains with steam engines. The first thing that came to the finishing line was a carpet inside of which the horses were inside. So it's actually just that they were able to get away with it, right? So technology has not allowed this result to be achieved. But there's a tabun horse in the old way that made this story work. And here, when budgets allow, this kind of story can happen, too.
But there's a place to eat. You said a very interesting thing. There's a friend-- another, another state. Yet we have seen two years, especially last year, that a large number of people would think that by artificial intelligence they would lose their value, replace people, increase their effectiveness. It's such an endless subject. Twenty years we've been improving efficiency, the world's population is growing, people get bigger, people work even more, more harder, more work, not enough. Here's Sasha Thybov, I think it's working even more, and it's probably 50, 1,000 times more efficient, yeah, inside. Well, if you look at the number of turnovers, Sasha's working a lot. So he's doing the start-ups. And then yesterday Sashin was meeting his wife at the café. Where's Sasha? Sasha's working.
I don't want to ask you anything, do I?
Yeah, but Sasha had a birthday, by the way, on the day of the show, congratulations Sasha on your birthday. Yeah. So, here we go.
Thank you!
Oh, by the way, Sash, it's your birthday! You can tell, we're shooting in a couple of days in advance. So... So I was gonna say one thing, one thing interesting. I had a partner, he was a Russian Pittter Network developer, a man like that, a Zarab-- a working billion dollars. He's a very good one to me, I think he said something. He said, "We don't do automated warehouses in Russia because in Russia, a warehouse officer is cheaper. If we do in Switzerland, we'll have an automated warehouse to do the best because there's a warehouse officer worth four to five thousand euros, pounds, there's, uh, $5,000 parole, and it'll be paid. So the automated warehouse will be paid. If a warehouse officer in Russia is worth $500, for example, it's useless to make an automated warehouse, you won't buy it." When we see $20,000 with you, it's a very interesting statement as a whole. And that's it.
No, I think it's stupid. I don't think he's wearing a dummy.
I mean, look, I, I don't-- wait, I'm not saying what he said.
Let's see, Altman. What's the subject? So far, everything that's moving is still the feeling that the AI agent is supposed to be cool. At the same time, what happens? If the AI agent is worth money like you, Sasha, now you're saying that if you pay $100 $20,000 a year for the executive assistant, $10,000 a month, yes, $100 $20,000 a month. A year. And whether you're willing to pay for the highest-quality, high-quality, artificial intelligence of the executive assistant, for example, $10,000 a year or $20,000 a year. But it's the amount of money that's completely different, isn't it? I think most people who were watching us had jaws to fall off the point of view of the assistant director-general, yes, worth $10,000 a month.
Good executive assistant, he still pays, like, 10,000 a month.
And that's why...
It's a very good system, isn't it? So we're not saying that there's not some assistant who really understands the business that was--
Yeah, but that's you talking from the world. Look, you're talking from the world, and you're talking about ten percent where the executive pays a million dollars. But when we look at the world, the executives don't make a million dollars. Executive makes a thousand dollars, $2000. Well, I guess the hospital's gonna take it all, right? Well, that one, what's called executive. I see I'm talking in the morning-- well, there, the director of the store, I don't know, some kind of thing.
Well, it's a business owner, that's an executive in my understanding of business owner, that's a man who's right.
I mean, a man who's really good at PNL, in a rude way.
Yeah. There's just a PNL word, yes, I guess. Here. And here, it's like this, what happens, is your opinion that this very big breakthrough will be, that access to the top assistants, will be very expensive, too, as access to the cool staff? Just some cool staff, actually, and people really know. And I think Sasha, you said today that people who want to get a kind of education, but they're not inclined to, actually, they can't live inside, they can't even survive, they don't even. They can use it. They won't win the top. Oh, that's it.
It's gonna be really hard, yeah. I mean, in my opinion, Sasha, I mean, I think that's just gonna be very strong as a differentiation, right? And you'll have a skirt, which, well, most people, I don't want anyone there, I don't want to look smeared or anything, but for most of the tasks, there, create a bill, well, there, shorter, for the most part of the job, Most GPT-4 tasks are already appropriate, for the majority. It's like just to avoid making some stupidness obvious, probably GPT-4o, as some extra check, well, not yet, but very soon, there's literally a couple of small incremental improvements, and we're... Come on. And that's the one that's been working for a few dollars a month. I just think that everyone has this, all of it will be integrated, it's gonna be directly integrated into the laptop, and there's gonna be some big corporate design and some big fan, I don't know, not workday, and some big one. corporate vendor, which is how things are set, you know, there's a snapchat in Snapchat that puts you in a pirate movie to keep track of you. They attacked from outside. And the same will be the installation of these assistants, and the entire infrastructure will be built under this, maybe special chips like Grok and so on. But there will be assistants for $20,000, $20,000, $20,000, and $2,00, yes, that's the most complex systems that are very difficult to configuration, yes, you should. Give, like, a normal task. These systems will have a very good one, and more access to sensitive information will be made. I mean, it's gonna take a re-examination, you know, like revolutionizing all the glass. I mean, just an assistant I would pay $20,000 a month for, so I would, of course, but I should have, I would be scared if the information was leaking, right? I mean, I just need to get him out of his life, actually, and then I just need to open up and make sure he can really work in the offline. I mean, if the assistant can work in the field, yeah, I mean, you tell him, I had priorities for a week. I got this kind of water, and there were some peddies in the market. How you recommend changing that information, we've learned from, uh, our user data. How do you recommend changing the roadmap, like, or something? Or you can run a test there, see if Manus really needs, how useful it is... in the places that Manus has, uh, what they're writing about Manus, how much angeagement is real or that's what it is. Is it a little fae? So, yes, I mean, you're gonna have to run an analysis, so deeper analyses that require the use of some internal data, some benchmarks that are inside, with external data of what happened in the field. industry. And then, on top of that, making some reasoning, it's getting very complicated, in my opinion. And that's what you can pay for, of course, 2,000 a month. It's easy. But I think so. I mean, it's a super-quality analyst, super-quality analysts more than 5,000, probably not worth it.
It's probably gonna be hard to pay AI as a man, but that's, like, a few thousand a month. But we're not there yet, unfortunately. I mean...
And all that depends, look, it depends, you know, if you make simple requests, well, quite complicated, now, and they're not really worth anything, a couple hundred thousand dollars or $20,000, Fifty dollars, there, and so on a month, so it's like there's got to be some sort of breakthrough. And there are still skies that seem to understand that they are the AI agent, but there are skids, I don't know, solved by an agent. For example, help me make a decision, you have to buy this business or not. Or help me make the right schedule, if I have a whole series of business, I have a lot of business, meetings, details, and everything.
I don't even know what it's supposed to be like now for a system that it should have to do with the volume of data, how it should analyse everything. If that system comes up, it's some kind of top pilot. Now, as it feels like you're gonna notice, here, Anya, you know, Manus feels like if you, for example, understood how to use them completely, you know, as fundamentally as possible, then, He's a very unbelievably nara-- that's how he's gaining competence. So you have your very strong professional competence, for example, in the design of interriers, and everything that's connected to it. And if there's a tool that can still write some codes, analyses or re-establish some systems, or some kind of stuff, or something, or something else, it's probably just some top pilot. I'm getting it. And you usually say that if you're giving someone or people something, quality, quality, quality problem, yes, quality-wise.
I think that's what it is for any man who's making up his mind, and it always raises a big question as to present and broadcast this picture to his client. And by the way, I've been watching how everyone is doing rightly to sing the manus, rightly to strike A. So, manus isn't a manus, manus is, I think, Sasha, you're on some French manner. That's what this term is from zoology. So it's the final, final segment of the front limb. So animals have one thing, and a man has a wrist and wrist. Well, that's the fingers, respectively. There are many theories that a man has made so far because of his ability to use, to use his fingers, his limbs. So it's funny that this is the point and context that Ilnar was telling about Grok that time.
That's a pretty big term. Here. But, back to the question you, Sasha, said, yes, I am now, when I saw that presentation, what do I see? First, it is a great opportunity to better present the materials, and it is a great opportunity, there, to find the design solutions, some of these technical details. Well, I mean, I was really interested in that. I mean, I tried to squeeze something out there, too, but I certainly didn't. Well, plus there, half was on Chinese, well, yeah. I'm waiting too.
Well, actually, in your work, having systems like, uh, deep research, yeah, and with some extra possibilities of some sort of analysis or re-establishment, it looks amazing.
Well, I mean, at least I've been telling my old edition about the dog coaching, how he found the information quite exclusive, but for this, I've done so many different cases this week. I'm looking for you. And yet, well, it's really cool. Well, that's quality.
Yeah.
And I'll try Grok deep research. I'll have to try Grok deep research and see what he's doing there. I think that if I add a little of this quality at the end or, for example, when I ask you what I want to buy, well, so far, in ChatGPT, these glue that I'm looking for something, he's looking for it on Russian. websites. I'm saying, "I need the product that sells to the U.S.." Well, that's a big difference. Goods that are sold in the USA. I'm telling him specifically about Amazon. I mean, I think he should give me the list at the end of Amazon's purchase, show me the prices. The task is not super-complicated, but then there's some super-implant to shape. What are we going to do now? And more and more systems will come in. And I think it's a manus. More systems are about what's a manus.
I'm saying you have French.
Yes, French, because my dog's called Lalik. It's a French name, and it's all called Lalik. And that's why I am.
You see.
You know, all the followers are gonna start talking.
Yeah.
And I think that what he does on top is this extra one like an analysis or some kind of construction, it's really cool with the real challenge that a man has. Because if I was looking for a conditionality, like, some stuff to buy at Amazon, and he knows where I live, he's at the end, well, he's making a straight-up list with ready references or maybe he's throwing me all over it. Right into the basket, yeah, or by--
Right into the basket, right? Or show me where it's more profitable to buy. They also compare and show where the goods are more profitable to buy or what, or where they are, what, what, what, what, what, what, what, what, the amount of money they buy. I mean, for the rest of the day, this case would re-establish. Because writing such a prompt, you know, impossible. People in the world can't write. As much as the proms' writing profession is now, but I think it's all very illusory, too, because, in fact, the coolest prompt can only write the LLM system itself. I mean, my personal understanding, that's-- I understand, professional prompt, where I'm programming and so on, I don't have any questions.
But the proms that deal with work, that's, uh, other kind, they're gonna be written by the systems themselves, they're gonna have to clarify questions, try to get in, get in. To say, OpenAI recently, I think we're gonna end it, so they're doing a great job of memory. And they're putting them in their Instagram. I'm even in the middle of a fire tonight because, well, they're in Instagram, putting a skate like they have a memory at ChatGPT. I think they've done some new memory. It's a year since this memory, I think, right? Long ago, remember, it's been a long time ago, a reservoir that works very hard. They put out the same thing that works very hard, and ChatGPT 4.5 didn't read it at all. Well, I'm in ChatGPT 4.5, and I wrote, "Look where I live." He knows where I live. He gave me everything. And 4o, their simpler version, which is standing there, I don't know, 50 times cheaper, like, if API uses it, she knows where I live in Los Altos Hills in California, and right away, I live in the same place. I've got it all figured out. Here. I think this is a memory, a memory, a context of humans, real skies, a real question. We're going to do this, and we might have new players in a week. Today, I think that's a very different issue. Obes-- we ask you in the commentaries to add your findings, reflexiations from the point of view of systems, skies, your own applications. And, well, it's important to you, for, uh, to all other people in terms of their own. Look, you've been writing the last time you've been using your systems and who's using anything. I wonder if that's a big deal. So, really, really people don't use that kind of thing, what kind of superstructures don't they put on themselves, huh? And sometimes you don't even know why. By the way, add, it's interesting to always be, if in the skies, add to what country you live in, what city you live in. So that you can understand why it's happening. Or you use it in your personal life or in your work. It'll always be helpful. Until the next meeting in exactly a week. We're on ToTheMoon Channel, technological news and sites from the Silicon Valley.