# Transcripciones — agentsquad-4k-video1 --- ## clip 1.mp4 - **Duración:** 02:53 - **Idioma detectado:** en (confianza 1.00) - **Palabras:** 517 - **Archivo:** `/home/clawd/uploads/agentsquad-4k-video1/clip 1.mp4` ### Transcripción **[00:00]** you actually create a knowledge graph that these agents can traverse so that they can **[00:04]** stay at the level of embeddings and actually be able to act over reasonably long horizon **[00:11]** tasks. **[00:12]** So those are a set of technical challenges. **[00:15]** And that creates really interesting questions of business strategy. **[00:18]** So we had a very important conversation today, which is like, what is the value **[00:23]** of a control plane? **[00:25]** And won't we just be ceding ourselves to cursor as an example, which is a phenomenal **[00:29]** product? **[00:30]** And what you're having to navigate there is just, you know, in my language, in my understanding, **[00:35]** I would rather build MS-DOS and Windows, not necessarily, you know, Adobe Photoshop. **[00:42]** And in this context, I think that for AI, we don't have that control plane. **[00:46]** That's what I would like to build. **[00:47]** I want to have something that basically says the ground source golden truth for **[00:52]** all of these agents downstream will always be this hardware independent, database **[00:57]** independent, language independent, symbolic representation of what you want to do. **[01:01]** So go and convince business people of that. **[01:04]** They get that. **[01:05]** So we can sell the shit out of that. **[01:07]** But then translating that into a set of agents that can act on target and on task **[01:11]** is very, very difficult. **[01:12]** It's a very complicated problem. **[01:15]** The business challenge is then to convince them why those innovations should stay **[01:19]** with us. **[01:21]** What do I mean by that? **[01:22]** So now I go back to how do you build a trillion dollar company? **[01:24]** What was the term? **[01:26]** Network effect. **[01:27]** What is a network effect that has never been built before? **[01:30]** There's never been a network effect built around these kinds of secrets, around code **[01:37]** meaning if I was a hospital and I'm trying to diagnose a cancer patient or I'm an airplane **[01:44]** company designing a new wing, immediately where your brain goes to is those are two **[01:48]** totally different sets of challenges, two totally different sets of problems. **[01:53]** And what I would say is wrong. **[01:55]** I challenge you to say the following. **[01:58]** That may be a different ontology, but it's probably the same problem. **[02:02]** And at the level of code and at the level of assembly, it all looks the same. **[02:05]** You agree? **[02:06]** Yeah. **[02:07]** So what if it were actually possible for this company to be able to leverage that **[02:12]** code because it's just recompiled with the different ontology so that you understand **[02:16]** it? **[02:17]** Now all of a sudden you have a network effect. **[02:18]** Now the N plus first company gets to leverage all the secrets of the N companies before it. **[02:24]** So when Elon talks about abundance, that's how I translate it. **[02:27]** That's how you create just a logarithmic expansion of abundance. **[02:32]** You can do anything quickly, cheaply. **[02:35]** So we are navigating this very complicated business path to convince these businesses **[02:40]** to let us do that with them, to convince them why this is important. **[02:44]** Now that's a unique task. **[02:46]** But if we accomplish it, we are giving all of these technical people that work at 8090 **[02:51]** a chance to do something very profound. --- ## clip 2.mp4 - **Duración:** 10:13 - **Idioma detectado:** en (confianza 0.99) - **Palabras:** 1884 - **Archivo:** `/home/clawd/uploads/agentsquad-4k-video1/clip 2.mp4` ### Transcripción **[00:00]** as well like these agents like there's so much hype around like oh like open **[00:04]** claws running for 48 hours and solved my life or something like crazy like that **[00:08]** like it's clearly not true it's clearly an exaggeration and one of the big **[00:12]** reasons of this is as you were saying there's a lot of secrets right there's **[00:15]** landmines you have to bring in the retired developer from 15 years ago to **[00:18]** explain why something happened and explain why like chat GBT deleted this **[00:22]** line and now the whole like codebase is broken so in this path towards this **[00:27]** ultimate goal right we want a reason in English what are some of the biggest **[00:31]** challenges that you guys are facing that you think more people need to be **[00:34]** thinking about and so one is a technical challenge and the other one is one of **[00:38]** business the tech the technical challenge is we're struggling with the **[00:43]** same things that everybody else struggles with how what is the **[00:46]** interplay and the interface between where the agent starts and ends and **[00:49]** where the humans get involved how do you actually create the right evils **[00:53]** and the right guardrails for the agents to stay on task how do you **[00:56]** actually create a knowledge graph that these these agents can traverse so that **[01:00]** they can stay at the level of embeddings and actually be able to act over **[01:06]** reasonably long horizon tasks so those are a set of technical challenges and **[01:11]** that creates really interesting questions of business strategy so you **[01:15]** know we had a very important conversation today which is like what is **[01:19]** the value of a control plane and won't we just be ceding ourselves to **[01:24]** cursor as an example which is a phenomenal product and what you're having to **[01:27]** navigate there is just you know in my language in my understanding I would **[01:32]** rather build MS-Dawson windows not necessarily you know Adobe Photoshop **[01:37]** and in in this context I think that for AI we don't have that control **[01:42]** plane that's what I would like to build I want to have something that **[01:45]** basically says the ground source golden truth for all of these agents **[01:49]** downstream will always be this hardware independent database independent **[01:54]** language independent symbolic representation of what you want to do so **[01:58]** go and convince business people of that they get that so we can sell the **[02:02]** shit out of that but then translating that into a set of agents that can **[02:05]** act on target and on task is very very difficult it's a very complicated **[02:10]** problem the business challenge is then to convince them why those **[02:14]** innovations should stay with us what do I mean by that so now I go back **[02:19]** to how do you build a trillion dollar company what was the term network **[02:22]** effect what is a network effect that has never been built before there's **[02:26]** never been a network effect built around these kinds of secrets around **[02:32]** code meaning if I was a hospital and I'm trying to diagnose a cancer **[02:38]** patient or I'm an airplane company designing a new wing immediately **[02:43]** where your brain goes to is those are two totally different sets of **[02:46]** challenges two totally different sets of problems and what I would say is **[02:50]** a draw I challenge you to say the following that may be a different **[02:55]** ontology but it's probably the same problem and at the level of code and **[02:59]** at the level of assembly it all looks the same you agree so what if it were **[03:05]** actually possible for this company to be able to leverage that code **[03:08]** because it's just recompiled with the different ontology so that you **[03:11]** understand it now all of a sudden you have a network effect now the n plus **[03:15]** first company gets to leverage all the secrets of the n companies before it so **[03:20]** when Elon talks about abundance that's how I translate it that's how you **[03:24]** create an owl like just a logarithmic expansion of abundance you can do **[03:29]** anything quickly cheaply so we are navigating this very complicated **[03:34]** business path to convince these businesses to let us do that with them **[03:38]** to convince them why this is important now that's a unique task but if we **[03:42]** accomplish it we are giving all of these technical people that work at 80 90 a **[03:47]** chance to do something very profound if we don't have that it's a wonderful **[03:52]** business we'll do the same thing as everybody else does we'll raise money **[03:55]** it'll be a unicorn that'll be a Decker corn blah blah blah whatever but **[03:59]** that's not what I care about I want to prove this other thing there's a **[04:04]** network effect in the code that businesses can have a shared **[04:07]** cooperative approach to this thing that there is no reason why Boeing needs **[04:12]** to be afraid of Memorial Sloan Kettering in fact it's the opposite Boeing **[04:16]** and Memorial Sloan Kettering can sort of cooperate and both can thrive their **[04:19]** costs go down their value goes up their downstream impact on their customers and **[04:24]** patients go up that is a positive some view of AI there has to be some sort **[04:29]** of initial apprehension though with these companies sharing my secrets right **[04:33]** huge it's enormous how do you overcome that that's about trust and **[04:36]** reputation the thing that we do is we methodically first of all so this **[04:41]** is more about business than technology in sure in business it's very **[04:45]** important to sort of fish with a fish art which is to say that in every **[04:48]** technology adoption curve you'll always go through the same transition you **[04:53]** have the early adopters then you have this sort of like mass middle and then **[04:56]** you have laggards and historically there was a very pejorative or negative **[05:02]** way in which you viewed those three classes particularly the last two **[05:06]** classes and I've challenged these assumptions to say something different **[05:09]** which is everybody has a different risk posture it's not I think like it's **[05:14]** very unfair to call folks that wait till the end laggards if you're a **[05:19]** regulated pharmaceutical company I was reminded of this this weekend and you **[05:25]** do something wrong you can go to jail so it's not like they're like I want **[05:29]** to poo poo this thing it's like they have a responsibility to all of you **[05:33]** that when you land a drug inside of your body that it doesn't you know **[05:38]** grow an arm out your forehead do you know what I mean yeah so that's not being a **[05:44]** laggard that's being responsible for their customers one two I said this **[05:49]** earlier right now we have a very complicated moment in AI which is that **[05:53]** there is demerism on one side and full throttle capitalism for me on the **[05:58]** other those are the two choices like think of the compact of the internet up **[06:02]** until AI the compact of the internet was we would make products where you **[06:07]** would participate and contribute and the quid pro quo was you would get a value that **[06:13]** you would assess as being greater than what you are giving me you use Facebook **[06:18]** and Instagram you take photos well thank you very much it allows us to create **[06:22]** a network effect we're able to build a trillion dollar company you still **[06:25]** find value now think of what AI says AI says I'm going to learn I'm going to **[06:30]** tokenize that knowledge and then I'm going to sell subscriptions thank you **[06:34]** very much see you later that is not a positive some view nor is the one that **[06:39]** says I'm going to go and then use that to replace you and fire you that is **[06:46]** deeply and fundamentally irresponsible the positive some view is we're going to **[06:51]** work together we're going to take our time and methodically navigate this **[06:55]** complexity when you need help with regulators or the government we will **[07:00]** go beside you and help you figure that out together we're going to get to **[07:04]** the finish line in a constructive positive some view we're going to actually **[07:08]** show how you're hiring more people you're paying them more that's how we do it **[07:12]** so it's slow it's methodical but it's really working because it turns out **[07:19]** humans again are amazing resilient positive not zero some you know many **[07:26]** people just want to play an infinite game they want to take care of their **[07:30]** family they don't want to fucking get screwed over and so why are we here it's **[07:35]** just terrible leadership terrible fucking leadership so I go talk to the fortune **[07:41]** 1000 and I tell them my view of the world and they can vote with their feet **[07:44]** they vote with their dollars and so so far so good I don't have a billion **[07:48]** dollar company but I would sign up I want to ask you your opinion on all **[07:53]** the hype around like these local AI agents you have open claw open Jarvis **[07:57]** etc as the models keep on improving with like quantization efficient **[08:02]** architectures better on device GPUs we're going to see a lot of the AI **[08:06]** workload shifting to like local compute and local devices right where do you **[08:10]** kind of see this like split between like local and cloud dough does this **[08:15]** kind of affect like business creation or like the enterprise side of the **[08:18]** world do you think there is actually any merit in these or it's just like **[08:21]** all hype and you don't I think it's fantastic I think the most **[08:25]** important thing that is going to happen in AI which is going to be **[08:29]** ginormously disruptive is fundamentally open source **[08:34]** models we don't first of all we don't have open source models okay **[08:37]** we have closed source in America we have open weight in China that's what we have **[08:41]** we do not have open source models but number one is we need **[08:45]** an ensemble of open source models and then number two is we need a **[08:49]** fundamentally completely unregulated totally distributed **[08:53]** form of compute initially training and then inference **[08:57]** the combination of those two things is profound **[09:00]** how open claw and all of these local agents work **[09:04]** in those realms I don't know well enough to say **[09:07]** but man that is a huge huge trend subnet three **[09:12]** of bit tensor folding at home pluralis there's a couple of others **[09:21]** I have I have no stake in any of these but I would recommend you to learn about **[09:25]** all of them it is incredibly important I think **[09:29]** what's happening that is a it's a Cambrian explosion that's just **[09:33]** going to go crazy because then you are totally **[09:36]** delevered from government oversight and government **[09:40]** infrastructure there is no kill switch and why that's important is that if **[09:44]** these things do become able to symbolically reason **[09:48]** you cannot have that gate kept by five or six entities **[09:52]** because what will happen if five or six entities control it **[09:55]** is you get immediately this distribution of you have the people that are **[09:59]** aligned to the model maker you know sort of like think of it as like a **[10:03]** planet with a handful of moons rotating around it **[10:05]** and then you have everybody else which is forced to become effectively a **[10:08]** vassal state of that model maker and I think that's a very complicated **[10:12]** outcome --- ## clip 3.mp4 - **Duración:** 01:52 - **Idioma detectado:** en (confianza 1.00) - **Palabras:** 346 - **Archivo:** `/home/clawd/uploads/agentsquad-4k-video1/clip 3.mp4` ### Transcripción **[00:00]** And in this context, I think that for AI, we don't have that control plane. **[00:04]** That's what I would like to build. **[00:05]** I want to have something that basically says, the ground source golden truth for all of **[00:10]** these agents downstream will always be this hardware independent, database independent, **[00:16]** language independent, symbolic representation of what you want to do. **[00:19]** So go and convince business people of that. **[00:22]** They get that. **[00:23]** So we can sell the shit out of that. **[00:25]** But then translating that into a set of agents that can act on target and on task **[00:29]** is very, very difficult. **[00:30]** It's a very complicated problem. **[00:32]** The business challenge is then to convince them why those innovations should stay with us. **[00:38]** What do I mean by that? **[00:39]** So now I go back to how do you build a trillion dollar company? **[00:42]** What was the term? **[00:43]** Network effect. **[00:44]** What is a network effect that has never been built before? **[00:47]** There's never been a network effect built around these kinds of secrets, around code, **[00:54]** meaning if I was a hospital and I'm trying to diagnose a cancer patient or I'm an airplane **[01:01]** company designing a new wing, immediately where your brain goes to is those are two totally **[01:07]** different sets of challenges, two totally different sets of problems. **[01:11]** And what I would say is raw. **[01:13]** I challenge you to say the following. **[01:16]** That may be a different ontology, but it's probably the same problem. **[01:19]** And at the level of code and at the level of assembly, it all looks the same. **[01:23]** You agree? **[01:24]** So what if it were actually possible for this company to be able to leverage that code because **[01:31]** it's just recompiled with the different ontology so that you understand it? **[01:34]** Now a lot of a sudden you have a network effect. **[01:36]** Now the N plus first company gets to leverage all the secrets of the N companies before **[01:41]** it. **[01:42]** So when Elon talks about abundance, that's how I translate it. **[01:45]** That's how you create just a logarithmic expansion of abundance. **[01:50]** You can do anything quickly, cheaply. --- ## clip 4.mp4 - **Duración:** 05:12 - **Idioma detectado:** en (confianza 1.00) - **Palabras:** 914 - **Archivo:** `/home/clawd/uploads/agentsquad-4k-video1/clip 4.mp4` ### Transcripción **[00:00]** of business. The technical challenge is we're struggling with the same things that everybody **[00:08]** else struggles with. What is the interplay and the interface between where the agent starts and ends **[00:14]** and where the humans get involved? How do you actually create the right evals and the right **[00:18]** guardrails for the agents to stay on task? How do you actually create a knowledge graph that **[00:22]** these agents can traverse so that they can stay at the level of embeddings and actually be **[00:28]** able to act over reasonably long horizon tasks? Those are a set of technical challenges and that **[00:36]** creates really interesting questions of business strategy. We had a very important conversation **[00:41]** today which is what is the value of a control plane and what we just beseeding ourselves to **[00:48]** cursor as an example, which is a phenomenal product. What you're having to navigate there **[00:53]** is just, in my language, in my understanding, I would rather build MS-DOS and Windows, **[00:59]** not necessarily Adobe Photoshop. In this context, I think that for AI, we don't have that control **[01:06]** plane. That's what I would like to build. I want to have something that basically says **[01:11]** the ground source golden truth for all of these agents downstream will always be **[01:16]** this hardware-independent, database-independent, language-independent, **[01:20]** symbolic representation of what you want to do. Go and convince business people of that. **[01:25]** They get that, so we can sell the shit out of that. But then translating that into a set of **[01:29]** agents that can act on target and on task is very, very difficult. It's a very complicated **[01:34]** problem. The business challenge is then to convince them why those innovations **[01:40]** should stay with us. What do I mean by that? Now I go back to how do you build a trillion **[01:44]** dollar company? What was the term? Network effect. What is a network effect that has **[01:49]** never been built before? There's never been a network effect built around these kinds of secrets, **[01:56]** around code, meaning if I was a hospital and I'm trying to diagnose a cancer patient **[02:03]** or I'm an airplane company designing a new wing, immediately where your brain goes to is those **[02:09]** are two totally different sets of challenges, two totally different sets of problems, **[02:14]** and what I would say is wrong. I challenge you to say the following. That may be a different **[02:19]** ontology, but it's probably the same problem, and at the level of code and at the level of **[02:24]** assembly, it all looks the same. You agree? So what if it were actually possible for this **[02:31]** company to be able to leverage that code because it's just recompiled with the different **[02:35]** ontology so that you understand it? Now all of a sudden you have a network effect. **[02:39]** Now the N plus first company gets to leverage all the secrets of the N companies before it. **[02:45]** So when Elon talks about abundance, that's how I translate it. That's how you create **[02:50]** just a logarithmic expansion of abundance. You can do anything quickly, cheaply. So we are **[02:56]** navigating this very complicated business path to convince these businesses to let us do that **[03:02]** with them, to convince them why this is important. Now that's a unique task, **[03:06]** but if we accomplish it, we are giving all of these technical people that work at 80-90 a chance **[03:12]** to do something very profound. If we don't have that, it's a wonderful business. We'll do the same **[03:18]** thing as everybody else does. We'll raise money, it'll be a unicorn, there'll be a deca-corn, **[03:22]** blah, blah, blah, whatever. But that's not what I care about. I want to prove this other **[03:27]** thing. There's a network effect in the code that businesses can have a shared cooperative **[03:32]** approach to this thing, that there is no reason why Boeing needs to be afraid of **[03:37]** Memorial Sloan Kettering. In fact, it's the opposite. Boeing and Memorial Sloan Kettering can **[03:41]** sort of cooperate and both can thrive. Their costs go down, their value goes up, their downstream **[03:47]** impact on their customers and patients go up. That is a positive some view of AI. **[03:52]** There has to be some sort of initial apprehension though with these companies sharing my secrets, **[03:57]** right? Huge. It's enormous. How do you overcome that? That's about trust and reputation. **[04:02]** The thing that we do is we methodically, first of all, so this is more about business than **[04:06]** technology. In business, it's very important to sort of fish where the fish are, which is to **[04:12]** say that in every technology adoption curve, you'll always go through the same transition. **[04:17]** You have the early adopters, then you have this sort of mass middle, and then you have **[04:21]** laggards. Historically, there was a very pejorative or negative way in which you viewed **[04:28]** those three classes, particularly the last two classes. I've challenged these assumptions to **[04:33]** say something different, which is everybody has a different risk posture. I think it's very **[04:39]** unfair to call folks that wait till the end laggards. If you're a regulated pharmaceutical **[04:46]** company, I was reminded of this this weekend, and you do something wrong, you can go to jail. **[04:52]** It's not like they're like, I want to poo poo this thing. It's like they have a responsibility **[04:57]** to all of you that when you land a drug inside of your body, that it doesn't grow an arm out of **[05:04]** your forehead. Do you know what I mean? Yeah. That's not being a laggard. That's being responsible **[05:10]** for their customers.