← AI: Voice or Victim?
Loading summary
Transcript58 lines
- [00:00]
A
And the big problem with AI, I think, is if you don't know the answer, everything looks like magic. With AI, right, if you had no knowledge of math and I gave you, on a phone or a magic box, a system that, when you ask a question, give you a number back in less than a second, you would think this is magical. And I think this is where we really need to kind of build more of that capability, that knowledge of how the system works. Because amplification and AI as a co founder is really where you work with an entity that has some capabilities that can help you augment and amplify your skills, but still operates under your authority and your ethics and your kind of own responsibility and accountability. And I think that's what I mean by AI as a co founder,
- [00:55]
B
AI isn't the future, it's now. And whether you're in hr, sales, operations or leadership, the choices you make today will determine whether you thrive or get left behind. Well, today we want to get started with AI, Voice or victim. I am the host. Typically, I'm the co host with my partner, Erica Rooney. But today I'm going to be solo again. My name is Greg Boone. I'm CEO of Walk West. I'm an AI strategist and someone that's really at the kind of the forefront of understanding and having these conversations around AI, what it's going to do to the future of work and to our society as a whole. Today I have. I am just so honored to have a guest here, Arno Fraad, coming all the way from Singapore, I guess, originally from France.
- [01:41]
A
That's right.
- [01:42]
B
Arnaud is a global executive consultant, AI strategist. But now, you know you're also an author, right? You have your first book out. And so that's what's really brought me here to this conversation today. I'm so excited for you. And so with that in mind, tell the audience a little bit about yourself before we kind of get into our conversation today.
- [02:06]
A
Well, first thing, Greg, thank you very much for having me. I really appreciate it. I know it's early for you, so, you know, it's really extremely nice of you to have me. So I'm Arno. I'm based in Singapore. I've been in Singapore for the last 20 years. My background is in building or rebuilding business units and teams for professional services firms. So I've worked across the advertising industry and the market research industry, as well as the consulting industry. And today I work with a company called Pure Spectrum that I also work on as an author. I'm Also an author of the Sentient Startup, which is a book that is coming out on April 28th with Penguin Random House, but is a book really that is almost, I would say 10 years in the making. So these things don't happen really quickly. I started writing on AI in 2016, initially on a slightly different topic, slightly different angle, which was the idea that marketers were going to have to start to market to algorithms because those were powering the search that were actually delivering products to consumers. And so the idea of this kind of first attempt called algo marketing was exactly that is how do we best to understand the relationship we have with these algorithms that power search and power product selection. And then I kind of ran out of words and never really published this and didn't know anything about publishing either, but continued learning a lot about AI and went down the rabbit hole of engaging with researchers, which at the time was really, really easy and getting to understand better what was happening. The advent of ChatGPT or GPT1 in 2019, the rise of that, I would say consumer facing set of tools and chat based tools. And then in 2023, Altman, Sam Altman was at a conference and he kind of jokingly said that he and his Silicon Valley buddies were betting on who was going to be the first solopreneur to build a unicorn, thanks to ChatGPT. And to me that was really kind of a bit of an epiphany in terms of thinking about this idea that actually with some of these tools, some of these capabilities, you could actually scale up business activities. Not necessarily as a solopreneur, but really thinking about how a few people together with those tools could really build much faster. And that's the premise for the sentient startup AI. As your co founder. It's really the premise that we are opening up a set of opportunities for startups, for scale ups, for large enterprises to really grow much faster by rethinking how they're operating thanks to some of these tools. And obviously within a context of ethics and integrity, that's a really big part of what we have to kind of discuss when we talk about AI and within a context of human first. Right. And that's also another really big part of my beliefs on what we're doing here.
- [05:18]
B
Yeah, I mean that is awesome to hear kind of that journey and what brought you here. I think it's also this moment in time. I don't know if you've been following some of this New York Times story of the company medv with Matt Gallagher's on a run rate for a billion dollars in revenue. Started as one person, and he hired his brother. So when I, you know, I heard you, I was doing some research on you, and I know that that was something that kind of struck with you. So I don't want to get too much into that, you know, right now, but I do want to come back to that at some point because I was like, man, I thought that was years down the road.
- [05:55]
A
Like, you know, I think, I think the reality is it's. All of this is on a compressed timeline. Like everything is happening a lot faster than expected, or at least that some, some of us may be expected. I think we can talk about medv a little bit later if you want, because I think there's a. There's a side to the story, right, which is really what is underlying in the business. There is an FDA investigation into medv. So there's a number of things that are not just on the good stuff.
- [06:22]
B
On the good side, the headline does precede some of the stuff that's going on underneath. And I'd love to get to that.
- [06:29]
A
But I think it speaks to the, to the kind of the excitement and frankly the hysteria around AI. And I think, you know, where there is a propensity to latch onto these stories because they really kind of strike the imagination, but they're also real. At the bottom of the story, there is a reality which is. It is now possible to scale. Right? It is now possible to scale much faster than it was before. And I think that's the important part maybe of that first part of the episode.
- [07:01]
B
So before we get there, let's. Let's get to some of kind of your early kind of entry point into AI, right? You have the sentient startup, AI is your co founder. Just to kind of like ground folks in a lot of the conversation that you and I have had and I've seen you talk about a lot. It's like a lot of people are treating AI as a tool, but then what happens when it starts to act more as a business partner or as you describe in your book, a co founder. Right. And so before we dive even further into the book, I mean, what pulled you into AI, Right, like right now, like, you talked a little bit about the background. What I want to understand is the learnings and the education side, because a lot of people talk at a certain level, right at the surface, but not a lot of people like yourself are digging in and trying to understand what is behind this.
- [07:52]
A
Yeah, so it's a great question. So I Think for me, the background is first, my background is I'm not a developer, I'm not a programmer by trade, I'm a commercial and business leader. And so for me, the early part of trying to understand what this is about are really deep diving into the technical aspects of it. When I start writing kind of this first idea of a book back in 2016, I'm starting to correspond with a few researchers, I'm starting to read things that at the time to me are extremely technical, technical. And it's really about trying to really understand what's underpinning the topic as we kind of fast forward. For me, it's more of a journey of really trying to get stuck and get to understand those tools. I started with, I would say the vive coding phase, right where you start to. You don't really understand what you're doing. You realize that these tools have massive limitations. They have huge risk, they are security risk, there's executional risks, et cetera. And you start to understand a little bit. Today I've kind of progressed through that. I've started to learn more about the intricacies of coding. I'm working on some projects. For example, part of the launch for the book is a panel, a series of panels I'm going to do with some fairly well known companies. But along those guests, I have built an AI panelist from scratch. Now that is for me an incredible capability in terms of what it does. It's a completely standalone, it works off a mini PC, it is fully voice driven, so it listens to the conversation, it speaks in perfect tone, it uses cloud code, it uses 11 labs in the background. And the fact that I could put this together in basically two days for $35 is a little bit astonishing. When I share the solution with some professional developers and their feedback was three, four years ago, it might have taken them six months to put that together or more. And I think this is where we are today. We are in an era of acceleration and it's really about embracing those opportunities, right? There's a lot to learn, there's a lot of bad information, there's a lot of misinformation, there's a lot of hype, there's a lot of hysteria, like pure hysteria. And there are some, some very real risks. But I think it's it, it falls upon us to at least those that are interested to try to do anything you can, to learn as much as possible, to really come with an informed view. And for me, that's the purpose of the book. By the Way the book is not, you know, I'm not expecting to sell thousands of copies. I'm expecting that it creates a much better quality of conversation. Right. And that's really the sole intent of writing it and publishing it through Penguin Random House.
- [10:56]
B
Oh man, that's great. Color in context, for sure. Right? Like, I listen to a lot of folks talk about this reality that you can't have an AI strategy if you don't first have AI literacy. Right. It's like a lot of folks have no understanding of the capability. So my background is, my degree is in computer science. So I used to write code before I became an executive. Versus, like, it seems like you have been an executive, right, and now you're becoming a coder, right?
- [11:27]
A
I've got a GitHub. That's disturbing look.
- [11:31]
B
Exactly, right, yeah, I got it. Like, that's the thing. We have to start like highlighting with folks like, do you have a GitHub or not? Right.
- [11:40]
A
Do you have a repo? I have their little, little color square.
- [11:43]
B
You know, if you have no repo, your hands are not in the dirt. As my guy, Alec Coughlin always says, you got to have your, your hands in the dirt to really understand. Because to your point, you know, it's, it really is going to be treated like a tool or it's going to be very surface level if you don't fully understand the capabilities of what you can build. And not just the capabilities, but to your point, the risk. So I want to talk more specifically, like, how did you and I come together? Because it became over a shared learning experience kind of at different times, within the kind of the mo. The MIT family of professional education. So can you give a little bit of background on that?
- [12:20]
A
Yeah, sure. I mean, to me it's kind of the same, the same purpose, which is, you know, I think before you, you should sit at the table, you should have a better sense of what's being discussed, how to engage and what's going on. And so I wanted come to this not just as an author, but also having different kind of learning experiences. Some I've done on my own, some have done through the formal processes such as MIT or others. And for me, it's really part of the same approach, which is, I think this is an extremely complex topic. This is a topic that has ramifications about how we work, how we live, how we connect with others. The future of work is a really huge topic in all of this. Right. It underpins a lot of things and we see some terrible decisions being made in the name of AI, Right. That's a really challenging part of it. And I think you need to again, come from a perspective of humility and knowledge. Right? So building your knowledge gradually, building that experience gradually from wherever starting point you are at. Right. And also some humility in terms of knowing what you don't know, trying to figure it out. I talk to a lot of people. I mean, we're having this conversation because in the MIT course we both attended, in the version I attended, you're one of the use cases, you're one of the case studies, right. So I reached out to you, I thought you had a super interesting profile, really great achievements. And to me that was another step on kind of learning from somebody really interesting, you know, learning from somebody that had achieved something else in that space. And that's kind of what I'm trying to do at the moment, which is connecting with some incredibly interesting people, building engagement, building conversation, with the sole purpose of augmenting the knowledge, augmenting the understanding. And also, you know, when it comes to talking to companies about strategy, back to your earlier point, you're spot on. You can't have an AI strategy if you don't understand AI. But there's another dimension to it. You can't have an AI strategy if you don't have a strategy to start with. And so these are the two things that are really critical for companies today. There's a degree of confusion. And one of the reasons I wrote the book just to close on this is I attended a corporate event many few years ago in around 2024, just before I started to really kind of dig into it. And at that event, somebody was presenting their vision of AI and it was frankly, a lot of complete rubbish. It was a lot of nonsense. And the problem was the people in the room, some of the senior executives were absolutely petrif by that description. It was apocalyptic. It was just an exaggeration of ultimate proportion. And I think to me that's also the purpose is I think we owe it to ourselves, but also to the people we work with and to the companies we work with to bring a voice of reason, a voice of kind of caution, and a voice of a bit of a balancing perspective on some of the really good things this can generate and some of the challenges that are going to come from it and which we have to face headaches at first. So that's kind of where I'm coming from.
- [15:40]
B
I mean, that's a good reminder of the realities of whether it's the hype or the overselling kind of the. Either the promise or the downside. Right. And you see a lot of this hyperbole out there in the market and similar to yourself in your background. Right. When Dr. Westerman at MIT asked, you know, could they use my video case study for future MIT executive at Education students, I said absolutely, because I want it to be helpful. Like, I don't make any money off of that. It's not the purpose of that. The purpose of this was to show other folks the art of what is possible, how transformative this can and will be. And, you know, I have a book coming out later called AI the Speed of Trust, which is in a very kind of similar vein. And it's just about how do you get people to understand what's not just the capability, but what are some of the downside, how are some of the things. How are you going to navigate some of these challenges? So getting back to your book and how you're positioned this as AI as a co founder, what does that mean to you and what do you want the audience to take away from that phrasing?
- [16:43]
A
Yeah, I mean, I think it's really the dichotomy between delegation or amplification. Right. So for me, one of the books that I found most relevant if you know nothing about AI is Ethan Mollick's CO Intelligence. I think is. Is a book that to me resonated very well because he's a little bit on the optimistic side, maybe. But certainly I think his premise of AI as a co intelligence is really at the heart of my own view on where this is. And I think if you think about it as a co founder, what does that mean to me? Well, it's this issue of it's more about amplification than delegation. I think at a very basic level, if you look at AI as a delegation tool, you think, okay, I'm doing this task at the moment, it's done by somebody else. I can delegate the task to AI. I don't need the person anymore. This is why we see those kind of absolutely irrational decisions of letting go thousands, tens of thousands of people on the basis of fake or faked AI productivity. But the second part of AI, which I think is much more important than delegation, is amplification, is we have a set of skills, but these tools, these models can help us really amplify those. And the big problem with AI, I think, is if you don't know the answer, everything looks like magic. With AI, if you had no knowledge of math and I gave you on a phone or A magic box. A system that, when you ask a question, gave you a number back in less than a second. You would think this is magical, but you had no knowledge of the numbers. So two plus two equals seven. But the answer came back in less than a second. It's like magic. The problem is, that's where we are today. If you don't know the answer, everything in AI looks like magic. But the problem is you really need to know the answer to understand what this answer is and whether it makes sense and whether it's valid. And I think this is where we really need to kind of build more of that capability, that knowledge of how the system works. Because amplification and AI as a co founder is really where you're working with an entity that has some capabilities that can help you augment and amplify your skills, but still operates under your authority and your ethics and your kind of own responsibility and accountability. And I think that's what I mean by AI as a co founder.
- [19:13]
B
Well, thank you for that. And as someone who also follows Ethan Malik a little too much, I'll say. I got to see him on stage a couple weeks back in Las Vegas at the Unleash America conference where he was speaking to a room full of HR executives. Right. And he was showing the art of what's possible. He was showing himself jumping out of an airplane through like a Sora or, or maybe VO3 video. He was showing the art of what was truly possible to open the eyes of a lot of folks. He puts a lot of the research in the data out. His substack is, is pretty amazing. And I point people to that. One of the things I also, I took a little bit of a victory lap, you know, last year I used to spend a lot of time when I trained people on AI. I use the startup example. I show people like, hey, how you can create a startup from scratch, right?
- [20:06]
A
Yeah.
- [20:06]
B
And as part of an eight hour session that we'll walk through. And then I saw him post something a couple months ago where he does the same thing with his MBA students. And he was so impressed by how something that used to take them weeks and months to come up with, now they were creating in four days and they had like 80% of a working actual prototype of whatever the business or the idea. Right. And that's kind of the advancements over the last several years. And so kind of with that in mind. Right. You talked about the kind of the. What it means to be AI as a co founder. Right. What does that mean? Like, I want to jump into kind of the more human founders, right? What are the founders right now that aren't really deep in the weeds of AI like you and I? What are they getting wrong or what are they missing? And what kind of advice would you give them when you think about how you can leverage these things?
- [20:57]
A
First, I don't think they're doing anything wrong. I don't think, you know, it's, it doesn't have to be AI for everything. So I think, you know, there, there are people building amazing brands or amazing products or amazing services and they don't feel the need and they're operating in their own way. And I think that's perfectly fine. I think the reality, though is some of these models are going to be incredibly helpful to do things you don't really need to waste time doing. Right. And there's a lot of small tasks, for example, that you can start to automate. You start to think about agentic AI, right? The power of what agents can deliver for you. So the question is more how fast do you want to go? Who are you competing with? I think what access to these tools is going to deliver is a new breed of entrepreneurs, much faster, much more agile, competing with their very big boys and competing with big companies in extremely credible way. And you've touched on medv. It's maybe not the best example in terms of what this business really is about. And I don't know enough about the intricacies of it to really comment in great details, but there are some, there are some questions about the testimonials that he's providing. There's some question about the authenticity of some of the doctors that are providing panel information, et cetera. But, you know, maybe, maybe it's fine too. The point being it's allowed to scale and it's allowed to scale with effect, efficiency. And I think the question would be, yeah, of course you can do everything without AI today, but you're competing against others that are leveraging it as sentient startups. I was on a panel just a few days ago with OpenAI and a couple of startups, and both of the startups on stage, once they had heard about the way I talk about it through the book title, described themselves as sentient startups. And both of them are operating like that. They're operating with AI, taking some functions, some of the things that they, as founders don't necessarily want to do and they don't need to delegate because it's more about just augmenting their own capabilities. So I think it's not a question of you have to. It's a question of this is a technology and a point in time where it's really about amplifying your efforts. Why not?
- [23:33]
B
Yeah, I mean, it's such a valid point. I love the way that you articulated the, you know, you don't necessarily have to, but you know what happens when your competition does, right? I was having a very real conversation yesterday and I was, it was very refreshing conversation was having coffee with a CEO who reached out to me in kind of the marketing space and she wanted my take on AI and kind of how we were transforming things, how we were thinking about business and we were sitting there and I was talking about how it's going to be disruptive for every industry, right? And so the, the analogy or the anecdote that I was using and I was describing even plumb in H VAC companies, right? And I was saying, I said, look, if I was to start a sentient startup right in the plumbing space, I said I would have AI to do a lot of back office work, a lot of AI SDR operation type things. I said, I'm going to be 30% faster, 30% cheaper, I'm going to be able to pay the plumbers 30% more. I said, I am going to be highly disruptive. And I said, if you don't believe me, I said, there's companies like BlackRock who've gone all in and are buying these plumbing and H vac companies and like these holding companies. And it was so funny because when we, we got up, the, the two guys sitting next to us in suits, they said, hey, if you start that AI plumbing company, we want in because we actually both work at BlackRock. And I was just, I had no idea. I was just at a coffee shop and. But they were like, what you were saying was fascinating, but they also acknowledged, like, man, we don't know enough about this stuff. And I'm not saying the people at blackrock, I'm saying not every single person understands what's going on, right?
- [25:11]
A
But you know, to your point, I mean, even imagine you're a solo plumber, right? You can work for eight or ten hours a day and it's exhausting, backbreaking work already, right? When you're working, when you're fixing somebody's sink or whatever you're doing, you can't pick up the phone, right? So maybe you have an answering machine, right? And that maybe means that you lose some business because the answering machine people don't want to leave a message or it doesn't work well or whatever happens, right? Instead you just build an agent that is basically your always on receptionist, right? That agent in real human voice, in real voice is going to take the messages, start building a quote right away. By the time you finish your job, you already have three updates on your phone from your agent saying this person has called, this is what they need. This is I think the indicator quote we can give them. If you validate it, I can send them a quote right away. Your task now is you finish a job, you want to get a bit of a break, all you have to say is yes, proceed, and the quote is gone. That alone has improved your quality of life drastically. Right? And the skills to put that together are extremely achievable, reachable for I would say, an average level of understanding such as mine. Right? So I think this is kind of what I mean is not everything is about putting man on the other side of the moon, right? Some of it is going to be helping the baker, the local plumber, helping small businesses, helping startups or individuals build their dreams because they have an idea and now they have the capability to really amplify, demultiply their efforts thanks to a number of tools that are now completely within reach. Right? And of course it's a big leveler. How do we make sure that we provide better education in places where there is not enough education? How do we make sure that people have access to better quality of knowledge, pronunciation, learning languages, et cetera, et cetera. So there's a lot of very positive things. The problem of course is you've got the other side of it, which is excess, you know, the tech bros and a number of things that are very concerning and that maybe require more regulation or at least a better quality of dialogue within society. So these are the two balancing acts. But I think for the solo plumber, there's a really good model there, not even working for BlackRock.
- [27:48]
B
Fair enough. I started Mark Bavasotto, a business partner of mine and I started RDU Labs which is a public benefit corporation with this whole purpose. And what we are trying to do, you know, get across, is that to your point, there's going to be a have and have nots and the person who is doing the maintenance and upkeep in my home deserves the same level of access as the big company. So we're doing this to make sure that there is a community aspect, that we make sure there's a path. Because I'm highly concerned about broad scale AI literacy. Right. I'm speaking next week at a university where I know there's a subsection of the professors that don't want the students to be using these tools. And I was like, I don't know about that. You're going to make them unemployable. It's like you need to understand that AI is embedded in every single piece of technology that we will use for now into the foreseeable future, for sure. But at the same time, one of the questions I have for you, what I love talking to you about both in this conversation, but even when we were off camera before having a conversation, is just some of the ethics and some of the kind of the decision making around this. Are there things that you would describe that are decisions that maybe AI should never make or it should make less of these decisions? How are you helping people kind of navigate that and what are your personal perspectives on that?
- [29:10]
A
I wrote a whole chapter called Navigating the Ethical Maze in the book because for me, you can't talk about AI without thinking about ethics and integrity. Right. I think we also have to separate the AGI conversation from the kind of agentic conversation. So the AGI conversation is a totally different topic. This is technology that is going to be almost immediately vaulted, that is going to be extremely powerful, yet very dangerous, and needs to be controlled within a very strong official legislative body. So this is one thing that we put aside. Eric Schmidt speaks about the fact that the moment AGI comes, it goes straight into a military bunker somewhere in a mountain. Right. And I think that's true. And I think that's really where Congress or National assembly or whatever the government structure is in the different countries we know need to understand that they are going to have to legislate on the use and the access to these kind of technologies. So if we put that aside, agentic AI, everyday use AI. Right. There are still some ethics and integrity aspects to consider. Is, you know, we've, we've heard about this example. Is it fine, for example, to have fake testimonials because you can easily create synthetic Personas? No, it's not. Right. It's still cheating. And I think that's kind of your decision. In my opinion, ethics is going to be a huge competitive advantage when it comes to the sentient startup. Ethics is going to be how you really differentiate yourself for the long run. Right. The companies that have lasted through histories, they've lasted in most cases because they were built on some strong founding principles, some strong values, and that's no different. It's not because you're building a three person startup or a three person sentence startup with 25 agents that you have to kind of dismiss the requirement to think about this with integrity and with ethics. Right. And I think there's a number of choices that have to be made around today. We see, for example, a lot of companies firing people on the basis that AI and productivity have delivered more value. We know this is not true. There is a Harvard Business Review study that demonstrates that in all instances the reason for firing has nothing to do with that. It's to do with lack of strategy, it's to do with over hiring and it's just to do with maybe just justifying huge investment in AI and having to deliver some sort of outcome. Right. And I think that's the reality. And these are noted. People know that people understand that these are not secrets. And I think this is going to have a huge detrimental effect to some of those companies in the long run. Right. So that's really your decision as a business leader. Ethics and integrity isn't different because you're managing agents or you're managing people or you're managing yourself. Right. These are the same things you have to play with.
- [32:16]
B
Yeah. I mean, to that point there's a lot to unpack there for sure. I, I've been on stage several times in the last couple months having this very conversation. Whether I was in Bogota, Colombia speaking on this last month, I was speaking at a, at a local big AI conference called All Things AI and to in an executive track. And I was calling out the realities, right? Whether it was the over hiring in 2021-2023 that are now kind of being reset. Whether it's the reality that some of these agentic capabilities and workflows will supersede much of the traditional SaaS kind of workflow productivity tool, right? I always tell folks, it's like AI does not care about your process or your org chart. It only cares about outcomes, right. And so, and so whether it's Block or Atlassian, all of these companies, what I will say is that there is a reality that, that if a company like Block or Atlassian makes these types of cuts, whatever the actual rationale is, and the very next day their Stock goes up 15 to 20% for sure, more and more are going to do this, right? Whether we like it or not. And what I always try to do is people say that. I say the quiet part out loud, right? And basically what I've tried to make sure people understand is the rules of the game that are being played and so that you can help. I like that. Navigating the ethical maze. I need to think about, you know, how people are navigating that from a business context. That's what I like. That's what I loved also about having our conversation. Let me ask you something a little bit different. Like, we've all had our own aha. Moments, right, that have gotten us into AI. But then outside of the aha. Moments, there are these moments where it's like, hey, man, this changes everything, right? What is. Is there been a moment in the last six months for you that you're like, wow, I know I'm already. Already in this, but this thing here changes everything for me.
- [34:12]
A
Yeah. I mean, they're. They're absolutely there. There are two distinct moments. One is very shallow, which is holding the book in the hands, right? It's like it's real there. That's. It's here. So that's a very small, fleeting moment. The one that I think sticks most is. I'm coming back to this AI panelist I built is really having a conversation with a machine. The model is called Athena. Athena, because of the word that I have to use to kind of trigger it. There's a lot of complexities around that in a noisy space. So I had to pick up a name that would be easy and kind of unique in the context. But what to me was absolutely startling is, you know, first test of the system, having again, built this from scratch by myself and having a whole conversation about the book, about with somebody or something, in this case, with extremely strong opinions, with a very good perspective because it has ingested a number of extremely kind of detailed studies and papers and perspectives and articles and contradictory points of view. And you realize, wow, this is. This is really quite amazing. Right? And I think that, to me, is one of those moments where you really think, you know, this is not. This is transformational. This is really a moment where it's only going to get more complex and it's going to get more exciting and it's going to get more challenging, but I definitely want to be part of that. Right? That's the moment.
- [35:53]
B
That's awesome. I. I had a few different moments of the last six months myself that are very similar in some ways. One was when I first saw the kind of. The image of my book that was coming out. I was telling my chairman I played a little bit of college sports. I said, man, this is the same feeling I had when I first picked up my jersey and saw my name on the back of it, but from a more deeper Kind of technical level. There was two other moments is when my chairman, who's never written a line of code in his life, started sending me apps that he was building using lovable and Claude Coat. And then to the point where I had to tell him recently, I say, you know that you're burning most of our credits at the company level. Right.
- [36:37]
A
Because he's a company.
- [36:39]
B
Right. But then the other one is, and I want your take on this is the Peter Steinberger moment of open clause, right. When I've still not personally actually leveraged OpenClaw. I'm looking at some of the other kind of versions, if you will, of that a little bit more controlled. But once I saw that and I said, oh, the cat's out of the bag now, or the toothpaste is out of tube, you can't put it back, right? The big foundation model companies, Anthropic OpenAI, they knew that this was capable. They were trying like crazy to not let it happen too soon. Now it's out there and everywhere that I go, open claw. You know, people have been open claw pilled. And even Jensen, who I want to get into that conversation too, about Nvidia, said that every company, they have Nemo Claw now. Right. And every single company needs to have an open claw strategy. So wrong way of getting to the point of what's your take on open claw, that kind of strategy and then your relationship with Nvidia.
- [37:40]
A
Yeah. So, I mean, I think first, the OpenClo moment is incredible because I think a few months back I was getting a bit concerned that we would be maybe entering some sort of an AI winter again, and that there is, you know, waning public opinion and shifting public opinion. And then Steinberger comes out with OpenClo 100,000, you know, gold star on GitHub within a week or something like that. Absolutely crazy. And it becomes this kind of, I would say, social phenomenon in a very microcosm. So I think the first thing we have to say is this is super inner circle style, right? This is not. But it's starting to go more open. Close was on the front page of the BBC News just two days ago. And I think that's kind of starting to give you an indication that the cat is out of the bag in that respect. And it's really moved past that inner circle and is starting to become something a lot more, a lot different. What is really interesting about it is it speaks to the fantasy that we all have of Jarvis, right? The idea of that sentient entity in our homes. I think at the heart of it is the Star Trek transponder. Jarvis. A number of technologies where we talk to a machine as we talk to a person and that machine does things for us without having to be instructed or through voice. Right? And I think that OpenClo in particular, because it has all of those connectors to Telegram, to WhatsApp, etc. That are so easy to set up, has really caught the imagination. Now the reality with OpenClow is there are some really serious security flaws in it. Now it's perfectly okay because again, Steinberger did not build that as a product to resell. He built it as a hobby. He is somebody that built a company many years ago, sold it and started to produce a number of ideas. I think he did 40 or 50 projects that all failed before that one. So this is again, not something that has happened overnight like most of us. These are not overnight discoveries. Right? This topic is not an overnight discovery. And I think this is really important is some of us have been talking about this or learning about this for more than 10 years now, or in some cases 20 or more. So Steinberger comes to this with a tool that he knows is not perfect, is open source, is fully accessible. And of course, suddenly it becomes this phenomenon. In China, Tencent and others have organized public events where people from 7 to 77 have turned up to learn about how to install and set up Open Clock. And I think that really speaks to this thirst for learning, this kind of interest in the topic. People want to understand how this work, but obviously the downside to that is there's not enough education on the challenges. And I was talking to someone recently who was very proudly telling me they had set up an open claw only to kind of shut it down because it was kind of misbehaving. And when I said to that person, so how did you harden it against prompt injection? Their response to me was, what is a prompt injection? And I think this is kind of a really big issue because you can't jump into these things without understanding some of the way they work. I set up OpenClow on a VPS, but just to kind of understand the setup process. But I don't understand enough about server side security, right? And so I disconnected it after having tested the whole routine of setting up because I wanted to understand how the setup. I'm now going to set it up on a dedicated machine on a Jetson Orin Nano super, which is a really small, very kind of dedicated machine, to try to understand how in a self contained environment it starts to operate what can I do with it? What kind of permission I can grant it, et cetera. But I think you've got to kind of adapt those tools to the level of your skills. Right. And then you've got other examples. You have people like Nat Eliassen who have set up a whole business partner, right? So they have an open CLO called Felix, and it's basically running a business now, Right. It's writing books and selling PDFs and all this stuff. And I think that's really the fantastic aspect of it. You can have a lot of different people experiencing the technology based on their own capabilities. That's got to be a really positive thing.
- [42:23]
B
Yeah. And that's why I would say that it changed the game for me. I mean, as someone has been at technology and has been on the tech side earlier in my career, the moment it came out, and I had buddies come up to me that weren't as deeply technical, they were semi technical, and they were like, yeah, I gave a root access to this and gave it access to that. I said, oh, no, no, no. I was like, you have no idea what this thing is going to do. And one of my buddies, he. He texted me. He was so excited about it. And. And he says, well, first he was excited when it was called Claudebot, right? And then they got in trouble. Within 72 hours, anthropic came out. So, hey, you change the name. Then went to Moltbot. Then it went to openclaw, which became like the most public interview ever for him to basically sell his technology and go work at OpenAI, which I thought was awesome. But then my buddy text me a few days later, he said, hey, man, this thing now has access to call me. And it won't stop calling me. It won't, you know, and it keeps calling me, and then it's spending money. He's like, you know, I woke up and I had like a 200 bill. I was like. I mean, I said, I was trying to warn you. Like, I don't. Which. Which is wild to me, Arnaud, is that on one side of the fence, we have people that are using AI as a glorified Google search engine.
- [43:40]
A
True.
- [43:41]
B
And then you have people on the other side of the spectrum giving it access to. To do anything, you know, basically rob itself blind. But the challenge is the same individual has the same access and quick capability to set up both.
- [43:55]
A
But I think this is where, you know, back to your earlier point, this is where Jensen is completely right. I mean, you need to have. As a business, you need to think about whether you call it openclo, Nemo, CLO or agentic strategy, it's all the same in effect. But I think you need to really have your agentic strategy. And if today you're not capable to articulate the why and the how of that strategy, it's really going to be a sign that whatever you're implementing cannot really sustain itself. Right. Because it's going to be essential in a context where you can deploy agents increasingly in six months, in 12 months, it's going to be extremely safe to do so at the moment, it's still dangerous. There are still layers that we're not quite sure what happens there, or there's a lot of work to do to secure them. I think as we go forward, we're going to see solutions and implementation models which make it really, really secure. And when you start to do that, then it becomes something we do every day. Right. Agentic commerce is going to take over. Right. It's going to be very interesting. It's going to be about rethinking how we engage with certain software or certain pieces of technology which today we need to spend time engaging with. And maybe we don't need to do that anymore. And to me it's all about that amplification phase which we're hopefully entering, rather than just a delegation phase. And as such, I think he's right. A business needs to have an open clue or at least an agentic strategy.
- [45:30]
B
Yeah. And then that's a great clarification and great context because one of the things I say a lot to that point last year I talked to several highly regulated industries across healthcare and finance and they would say, oh, we're not going to use AI, you know, for the foreseeable future. I said, I said every tool will have at a minimum, AI assistant and co piloted. I said, the idea that you're not going to use AI is that what are you going to go off the grid? Right. And now you go from 2025, where everything had an AI co pilot to now in 2026. Almost all of these tools will have their agentic strategy or open claw strategy where they'll be. The product itself will be more agentic in nature. And one of the conversations I was having in the last 24 hours, I said, what happens when the tools that you use to run your business flip the switch and say that you can no longer use the old workflow drag and drop, click. You have to now talk to the machine and use the agentic way of working? I said, you don't have a choice anymore. Right. And that's the thing that is concerning to me.
- [46:36]
A
Yeah, completely. And also the models that are coming out are becoming so sophisticated that they're also uncovering things that we've been doing wrong for a long time. I mean, a case in point is what has happened today with the release or not the release, but the paper on Mythos from Anthropic, which is basically is seemingly geared towards cybersecurity and in running a number of tests has uncovered vulnerab in some cases that are 27 year old. Right. So they found a major flow in OpenBSD from a piece of software that was created 27 years ago. So this is a piece of software that has gone through multiple iteration, multiple checks, multiple testing, and still add within itself this flaw. Right. And this is what this model has uncovered. And I think this is a really kind of wake up call because what it suggests is some of the tools we're using may not be as safe as we think they are. Some of the tools that we've been using for a long time may require to be investigated through what artificial intelligence can do. I think we really need to understand this a lot better and certainly leverage those solutions in a more effective way for that purpose. For sure.
- [47:58]
B
You hit the nail on the head. I was thinking about this a lot that I've been thinking about it even back since December. I listened to a podcast talking about the. It was the head of, I guess at the point, at the point in time for OpenAI, he became the new head of security and vulnerability. I can't remember the guy's name, but he talked about finding how the Codex model found in OpenAI found a vulnerability to your point, that was 20 something years old that people thought, didn't even realize existed. And it found it. And they were like, oh boy. Right, right. And now you see the, the leak or the, the mistake of someone in the kind of the CMS group or the marketing group inside of Anthropic with Mythos gets leaked out. Which the. I. The irony of it is the whole point of Mythos was about security vulnerabilities and then it's get leaked. But it was a human error that effectively leaked the, the whole paper. But I think they found something like millions of vulnerabilities.
- [48:52]
A
Right. For sure.
- [48:53]
B
And like into your point. And so we're going to learn a lot more. It's kind of, as we start to kind of wind down here a little bit, there's a couple things that I want to do one. I want to make sure I give you space to talk more about how do you think the book is going to impact future founders. Right. But before we do that, we like to do one little game we call Last Chat.
- [49:14]
A
Okay.
- [49:15]
B
So it's a little surprise segment, but we ask you whatever. I never actually asked you this. I actually don't ask this question anymore. I used to ask what AI tools people were using, but there's so many tools and I try to tell people focus on categories, not tools. Like what are the things you're trying to accomplish, not the individual tool. But I'm going to break my own rule. What is your like daily driver from a, from an AI tool perspective?
- [49:40]
A
Oh, for me, I've switched completely to using Claude. So I think that is delivering the best in terms of the type of things that I use it for. I'm starting to investigate the much smaller models such as the Gemma 4 that just came out over the weekend or Qin 3B. And I'm interested in those small models because as part of the implementation I want to build on the Jensen or in Nano, I want to build a completely self contained system. So I want to be able to have on that a small enough 4B model that can run directly and try to see what it delivers. Right. So I'm just really curious about these kind of edge models. But on a really small form factor, that's what interests me at the moment. So that's kind of what I'm looking into. But on a day to day basis I use Claude. I obviously use also things like Google, Nano, Banana for some visualization sometimes. But I would say really Claude has been, I think extremely impressive. Obviously their throttling usage. There's a number of things there that are challenging, you know, but for the purpose I have of it, it's certainly doing its job. At least what I understand of the outcome.
- [51:04]
B
Yeah. I mean you have the French accent, so. Right. Calling it Claude or Claude.
- [51:09]
A
Yes.
- [51:09]
B
Yeah. When I see people, so I was like that's always cool. And I use Cloud quite a bit more now than I used to for a variety of different reasons. And I still use ChatGPT significantly but I use all the tools because one, because I'm training people. Right. So I have to use like I even have to use Copilot sometimes. I say have to because I have my own feelings around that.
- [51:30]
A
Yeah. Mistral. So I'm kind of getting curious about it. So again for me it's about curiosity. I mean you use a tool. Yes. Because it's a Bit of a default, but I have all of them. You know, I use, I use Deep seq, I use OpenAI, I use Mistral, I use Copilot, obviously the Google suite a lot. So for me it's really about kind of what's the go to and then the curiosity you have around. Not every model behaves the same. And I think that's also something that is incredible is you have so many models delivering so many, I would say niche expertise or specific type. There is some incredible work at the moment coming out on text to voice or direct to voice. Right. So there are some really interesting things coming out. There are a couple of topics that are really big interest to me. Anything to do with voice is a huge interest for me. Anything to do with small edge model is a huge interest. So that's kind of what I'm trying to learn about. So that's completely model agnostic. It's more about what's out there and what I can get my hands on.
- [52:35]
B
Yeah, I love the edge devices. I love the idea of, of air gapping, something that you can possibly put into your pocket or just kind of walk around with. But when we talk about last chat now, I have to ask you, what is the last thing that you chatted with Claude or Claude that you can share? And I'm going to share what I last chatted about.
- [52:56]
A
A friend of mine is kind of building a business and was asking me to test a number of hypothesis. So I was helping him do that through a couple of exchanges that took place and we were testing a voice interaction and a voice transcript and that's what I used it for, to help this friend of mine trying to kind of test what he's doing. That's the last thing I do.
- [53:21]
B
Nice. I had my last chat, I was going back and looking this morning and someone was reaching out on the HR side and they're trying to transform a fairly large organization to be upskilled in AI. And they had a big provider that works with our federal government here that is cutting edge, doing everything, you know, mentioning openclaw and first principles and air gap and all these, this terminology. And they were like, help me unpack and understand this. So I was like, all right, I can better define it and put it in my own words. But then I also want to say, if I had a high school education, right? What how would you convey this message? So I was trying to just put it in terms that made more sense for them and for their business context of a lot of this kind of, you know, we're all, you know, there's a lot of people that are in this bubble. Like we've, we've talked about like, and it's a very small bubble. And so I use it a lot of times to convey a message or to kind of make the complex more simple for folks. Right. But I also, you know, I build a lot of things with it. But my last actual chat and even the premise of the question assumes AI Basic is what I would tell people. I say your AI Basic is showing because I don't AI search, I AI instruct, I have it build things, do things, agentic flows. So with the last few minutes here, what do you want to tell our audience about your book? About yourself? What are you passionate about? How do they get the book?
- [54:51]
A
Yeah, I mean, I think I hope I've conveyed enough what I'm passionate about on this topic and what I'm interested by. The book is coming out on April 28th. I think at the moment it's on pre order in Asia and Australia. So I think it will come to Amazon in the us, in the UK in May. I think for me, really, I'm not expecting to sell millions of copies. What I'm expecting is that this is a great way to open really interesting conversations. So if you're interested in having these conversations, you can find me on LinkedIn and I'm always interested to engage on these topics and, and if that can help someone, if it can inspire someone because they realize that they themselves can become a sentient startup founder, they themselves can build a business that they never thought would be possible before. Then I think it's mission accomplished.
- [55:41]
B
To me, that's awesome. And I will be using that term, the sentient startup, right.
- [55:47]
A
You can use it. I relinquish the copyright for this express purposes to your benefit only.
- [55:54]
B
I appreciate that and I would like to say to the audience, right, if you're still thinking about AI solely as a tool, you are behind. There's a lot more that you can create. And I like, similar to Arnaud, I like to focus on really elevating the quality of work which then will elevate your quality of life. If you can do more things that you're passionate about, that are more fun to you, that are more personal, that give you more time back to your, your with your family, your friends, or the things that you deeply care about, I think that's what matters in business. So I appreciate the time, friend. Thank you for making the time all the way from Singapore. Best wishes on the book and hope you sell a lot more than a thousand copies.
- [56:37]
A
Thank you so much. Thank you. See you.
- [56:39]
B
Greg, Thanks for joining us on AI Voice or Victim? If you want to stay competitive in the age of AI, start now. Take one insight from today's episode and put it into practice in the next 24 hours. Make sure to follow us, share your thoughts, and subscribe for more actionable AI insights. See you next time.