Speaker Key:
DD Damien Deighan
KP KP Reddy
PD Philipp Diesinger
00:00:00
DD: This is the Data Science Conversations Podcast with Damien Deighan and Dr. Philipp Diesinger. We feature cutting edge data science and AI research from the world’s leading academic minds and industry practitioners so you can expand your knowledge and grow your career. This podcast is sponsored by Data Science Talent, The Data Science Recruitment Experts. Welcome to the Data Science Conversations Podcast. My name is Damien Deighan, and I’m here with my co-host, Philipp Diesinger. Today we are talking to KP Reddy, who is the founder and managing partner of Shadow Ventures Seed Stage Venture Firm focused on revolutionizing the built environment. He is also the author of a recent book entitled, “Creating the Intangible Enterprise. His previous books include, “What You Know About Startups Is Wrong,” which Debunks 11 popular myths about entrepreneurship. And his other book, “BIM, for Building Owners and Developers, is the Definitive Guide to Building Information Modeling.”
00:01:16
KP is a technology innovator and has been for over three decades building startups since the early days of the internet back in the 90s. Now his passion lies in identifying, advising and investing in market leaders that are transforming cloud computing, computational design, robotics, and AI. So, in today’s discussion, we are going to be getting the venture capitalist perspective on how the edge of AI is unfolding, and we will be exploring how AI will change how we build companies in the future. KP, it’s great to have you on the show. Welcome along.
KP: Thanks for having me, Damien.
DD: So, if we start with your background, KP, I mean, tell us how you went from a civil engineering degree to the world of venture capital.
KP: Yeah. I always kind of joke, my father was a civil engineer and my mother was a computer programmer, so a little bit bred this way. But I started early days of PC. Started writing code when I was 13 on the early PCs that my dad had bought for his company. And ever since then, I’ve just been intrigued and fascinated. And quite honestly, when I went to college, computer science wasn’t that big, and I was self-taught. So, I thought I’ll just go into the family business, get a civil engineering degree, because I already know to write code. Clearly things have become more complicated since then. But I figured I couldn’t like design bridges on the side, but I
00:02:52
could write code on the side [laughter]. And so, I started working in civil engineering and my first product was a web-based construction management product that I built in 1996 that I struggled with, but we pivoted and did well.
DD: And then from industry, what attracted you then to the world of startups?
KP: I think that when you start working in very parochial industries like civil engineering where your boss says, after five years you can do this and after 10 years you can do this, and in 20 years you can be a vice president, and you have a level of ambition. These traditional industries have a way of driving people with a little bit more ambition and entrepreneurial mindset out, because at 22, you’re very impatient by nature. And so, when you hear your boss’s story about how he’s a VP, but it took him 30 years, it’s just not that attractive [laughter].
DD: So, moving on to the AI conversation, how would you describe the current state of AI technology and its adoption in industry?
KP: Yeah. I like to describe it as a teenager. You’ll sometimes mistake in a teenager for an adult, and sometimes a teenager mistakes themselves for an adult. So, there’s definitely like signals of this adult maturity coming on set. And then every once in a while, a teenager disappoints you, ’cause then you remember this is a child not quite an adult. So, I think we’re in
00:04:52
that teenager phase of AI, but moving, maturing very quickly. If you think about the, not just the expansiveness of feature sets that keep coming out, but also the quality of what’s coming out, even on the older feature sets, been quite aggressive.
DD: Yeah. I think the ecosystem build out is on a really large scale to anything else that AI has been compared with like blockchain or the Metaverse or whatever. So, it’s clearly here to stay. I mean, what’s your take on its adoption and industry? Because I think that is very uneven and there a lot of commentators believe there’s a lack of enterprise use cases for particularly Generative AI. And I think it’s important we draw the distinction. What’s your take on that?
KP: Yeah, I think enterprises are using are related to early days of the internet. They’re using it in very, I would say, user-driven use cases, not enterprise driven use cases. So, most of my investors are large enterprises, they’re multi-billion-dollar companies. So, when I talk to the CEOs and I ask about AI, and they’re like, well, we’re building a strategy at the executive level, but then when I talk to the people in the trenches, they’re using AI a lot, but it’s for their uses. It’s for their use case. And I don’t think those individualized use cases have percolated to any kind of pattern that says, oh, here’s what we should be doing enterprise wide. So, I think there’s
00:06:09
pockets of value creation happening using AI by people that have become power users and are intellectually curious about how to do their job better.
But I don’t know that necessarily a lot of these have percolated up. And I think the second part of that is all of these pilots, nobody really has a good understanding of what the long-term compute costs will be. So, you do these things in a pilot, your compute costs are negligible, and then you think about deploying it over, the math of deploying it over 20,000 people in an organization, or a hundred thousand people in an organization. Sometimes that ROI no longer exists because of your compute costs. And I think in many ways, enterprises are waiting to see one about security and privacy and those topics, obviously. And then I think even compute costs, they’re, how much is this going to cost us, actually,
DD: Yeah. I think potentially closely related to that. I don’t know if you saw David Khan’s blog a few weeks ago from Sequoia Capital. Obviously, for anyone who isn’t familiar with that, he posted a blog. They’re one of the largest VC investors in the AI space, and he suggested that AI will definitely deliver a lot of value in the long run. But in the short term, it has what he calls the $600 billion problem. That’s his estimation on how much extra revenue needs to be generated to pay for the aggressive AI infrastructure build out we’re seeing and the compute costs and so on.
00:06:09
What’s your take on that? ‘Cause I was flabbergasted, to be honest, that a VC actually published something like that.
KP: Yeah. Look, we as VCs there’s nothing that we do is for the goodness of sharing knowledge [laughter]. Everything we do has some long game. We’re long-term investors in many ways, but I think everything we say you have to take with a grain of salt. Because there are motives, there are heavy financial motives in our industry to say the right things. And I think when you’re at Sequoia, one of your ideas is that we’re a large capital allocator, and we want to make it very difficult for everyone else to get in on it. And we want to push LPs to invest in our funds because we have an inside track, and we already have momentum, and those kinds of things. So, I think there’s a lot of motives involved.
However, I do think history repeats itself. If you look at cloud, we went from taking servers that were on-prem and moving all of our data into the cloud. It was supposed to be cheaper, more secure, more everything. And then now if you do the math, one would say, I don’t know if I’m saving money, maybe I should have my servers hosted in a data center. Maybe I should have my servers on premise now. Like, who knows? But I think there’s this cyclical nature of moving both to the edge and both from the edge and centralized back and forth. I think you’re seeing that now. But I think as far as the value creation and how people think about how we
00:09:30
monetizing these investments, what they tend to forget is there will be new business models.
And I talk about in my book, like we’re at the early stages, but we have yet to imagine the new business models. We looked at the internet and we said, Amazon’s selling books on the internet. So, we went from brick and mortar to online. Same, same. We never thought or imagined that Amazon was going to be the largest cloud provider and be in all these other businesses. So, I think the fallacy is that we’re applying a technology to do business better than we’ve, but the same business, but just differently and better. And ignoring the idea that there’s going to be big, big businesses that we can’t even imagine yet.
DD: So, I guess what you’re saying is just like we had web 1.0, 2.0, 3.0, we’re right at the beginning. We will have an AI 2.0 and so on, where all of the new opportunities and models, business models merge.
KP: Yeah. 100%. I think you have to look at different markets. I mean, if you look at what AI is already doing in life sciences and what it can do. I mean, it used to be if I’m an entrepreneur, I want to start a life sciences company. My startup, my seed round is 250 million. Now you’re starting to see, because of AI and what you can do with so many fewer people and niche plays. You can do designer drugs, you can do personalized medicine, you
00:11:02
can do all these things almost out of your garage. That’s a whole industry that’s going to be impacted, that’s been highly driven, kind of shared monopoly style. If you look at pharma, there’s big pharma, there’s not a lot of small pharma, there’s not a lot of midsize, it’s all big pharma. So, I think when you start looking at the disruption of what can happen and new business models, I think it’s going to be super fascinating. And I think if you don’t have the creativity to think about what those new business models are, you can get caught up in, oh, we have to create 600 billion up. I mean, say that to Amazon 30 years ago.
DD: So, KP, how can companies differentiate between what’s hype, what’s real, so that they can have actionable AI technology and outcomes?
KP: Yeah. I think there’s this balancing act of, are you building on other people’s platforms and leveraging their feature stack. So, one of the things I talked about 12 years ago with AWS was if you were a startup and you’re using just the core features of cloud, and maybe some DevOps, you were fine. But if you started going into the feature stack of AWS, there was zero opportunity for you to switch, first of all, to another cloud provider. And in fact, when you did the financial modeling, Amazon was getting more gross margin than you were. So, the title of my paper back then was, are you merely a feature of AWS? And so, I think that’s one of those challenges, like how are you using the big AI models?
00:12:49
Whether it’s OpenAI or Llama, or anything else. But also, how are you looking to optimize the existing feature set and from a cost perspective, and then how are you looking at how do I build features on top of it, and those features on top of it, how are they proprietary versus what might be out of the box? And I think that’s the real question. And some of that comes from markets. Like, is OpenAI going to go after a certain market? Like I think OpenAI is pretty clear. They’re probably going after the entertainment industry with a lot of their features. I don’t know if I want to go build something that competes with that. And if I am, I’ll probably get put out of business. I think the same thing we’ve seen these AI legal companies, there’s several of them, like doing legal and contract review.
My theory is by the end of the year that’s going to be called Microsoft Word, they’re not adding significant value above the legal document review above and beyond what you’ll probably get out of Microsoft Copilot on Microsoft Word. So, I think that a lot of it is both the technical stack of what you’re building as a feature, and then also understanding where the market is moving and where are these big companies going to. ‘Cause I think the big general use case is like legal document review. That’s such a big, huge use case that I don’t know if a niche player, unless they have significant feature sets that won’t be native to open AI or anything else, I
00:14:23
think they’re done. But if you’re doing something that’s super niche, super interesting, and technically difficult, you’ll probably be fine.
So, I think it’s a really classic balance of you just can’t build technology, you really need to understand where the markets are going. And I’ve seen a lot of startups raise 10, 20, $50 million. There’s one right now, the legal AI company one, they’re saying it probably doesn’t even work well. Like, it’s not even working. And then two, by the time they figure it out, it’s probably just a feature of Word and you can get it for $29 a month.
DD: Yeah. Interesting. So, what long term, I think you’ve touched on it, but what can larger enterprise, your typical investors, what long-term strategies can they adopt to make sure that they stay competitive in an AI driven future?
KP: I think one is a shift in thinking to focus on new business models. I think that’s one area that large enterprises don’t do a great job. You have a billion dive, one of my investors they do $2 billion a month, it’s working. So, they’re not thinking about the next new business and all that. And I think that’s a mistake. I think you have to be thinking about new business models. And not just how AI can disrupt your business model, but just how are there new, how can you go disrupt other businesses. Based on whatever competitive advantages you have. I think the second thing, and I
00:15:58
think to your audience too, is we’ve been talking about data for a long time. We’ve been talking about data for a long time. We talked about big data, we’ve talked about data lakes, and what I’m seeing at the enterprise is using AI tools to actually take a bunch of unstructured data and make some sense of it.
And I think that’s super powerful other than I’m not sure now that you have all this data, what are you going to do with, it goes back to the new business model cases. So, I think there’s some low hanging fruit around, I would say, big data challenges that we’ve been working on for decades now, that like, oh, we need data hygiene, we need this, we need that. Our data’s not worth anything until it’s clean. I think AI gives us an opportunity to really work with a lot of the unstructured data that we’ve accumulated and figure out whether it’s worth anything. Something’s only worth something if someone’s willing to pay for it. I think I wrote on LinkedIn the other day, there’s a saying, data is the new oil. My running joke is now I’m seeing AI data centers being built that are off grid, on fiber, off grid, 100% powered by natural gas. So, literally data has become oil [laughter]. It’s not like, and that it’s not an analogy anymore. Literally, data is now oil.
PD: So, KP you’ve talked a lot about the need for change in large organizations, you mentioned the life science business specifically. I think big organizations are notoriously slow in pivoting and adapting their
00:17:40
strategy. Even now, we see that just coming up with an AI strategy in itself is already a huge endeavor that many organizations struggle with. So, my question would be, do you have any tangible practical advice for organizations, maybe for a chief data officer, or CEO, or c-suite struggling with this? Like, what could be done to help them adjust fast enough to basically deal with the challenges that come their way?
KP: Yeah. I mean, I have a simple answer, which is to start thinking like a startup founder. Start thinking beyond and above what your current business is, what they’re doing, and how do you go into new businesses? Because there’s a trend happening right now, and a lot of it’s, I probably get three phone calls a week from private equity people, and their question is, I want to buy this traditional business. It generates 10% profit, it’s a commodity business. However, if I bought it for cheap and applied AI principles, innovation principles, could I actually turn it into a 30% or a 40% net profit business? And these are people managing hundreds of billions of dollars. And so, you’ve already seen it with companies, General Catalyst, good friends there at General Catalyst Ventures, and they basically, they’re buying hospitals.
So, think about this, this is a VC. They’re buying on a hospital saying, this is a commodity business, they don’t do it well. Can we apply technology, AI, and innovation to make a hospital highly profitable, highly service
00:19:23
oriented? So, there’s this trend to say, we’re not going to wait for a traditional enterprise to change, we’re just going to buy them, fire everyone, quite honestly, and put in new people that are going to approach everything differently. Because I think a lot of these large enterprises, that’s going to be the shock to the system.
PD: And maybe a little bit, the Meta question, you talk a lot about ventures and so on. Do you think AI will also have a big impact on the way new business models come into play and how investment decisions are made, how ventures are being evaluated, maybe how the diligence is being done? Will it play a role in that process?
KP: 100%. I think this is going to usher in a new wave of entrepreneurship. And I think I have been saying this openly, I think AI might actually disrupt the venture capital business. Because if I don’t need $25 million to build a team and do all these things, why do I need VC? And I think there’s this tone that says maybe I can build a unicorn all by myself. If you think about AI as your employees, like your agents, and all your co-pilots, do I need all these people. And the ability that people are running, and I’ve talked to some people between AI and using contract labor talent in India, Vietnam, wherever. Can I actually build a zero-employee company and get to a billion in revenue? And I think in VC, we’re looking at it saying, I think you might be able to.
00:21:03
And you might be able to find… So, I think that’s one dynamic. I think the second dynamic of business building as VCs, we have this idea. I mean, if it’s not a hundred-billion-dollar market, we’re not interested. But as an entrepreneur, can you go after a nice little half a billion-dollar market and capture 30% market share and totally bootstrap, not need a single dollar to go after those markets. So, I think it’s going to change a lot of that just in terms of how we finance companies, how we think about companies. And I think that’s where enterprises, their core business is going to be under attack, not just by private equity, and tech guys, it could just be a small competitor that they didn’t even pay attention to that just starts to scale and do things better.
PD: If your employees are actually AI agents, could you see a scenario where the AI agent makes also financial decisions, maybe gets a loan from another AI agent to build something or develop something?
KP: That would be really cool, wouldn’t it [laughter]?
PD: I mean, I’m seriously thinking about it now that you mentioned it. Because I would not be surprised if like human decisions, let’s say up to a certain budget, let’s say up to five, 10 million or so, human decisions might not be super-efficient there.
00:22:28
KP: I mean, why not? If you think about, if you go get a loan, what is underwriting anyway? It’s really humans putting things in spreadsheets and getting an outcome out of the spreadsheet. There’s zero thinking, there’s no critical thinking happening up here. If you think about credit cards. Credit cards, there’s no human in the loop that’s pretty much automated. Now, if AI can say, oh, based on your spend, based on this, based on your social posts, you might be getting, heading for a divorce, we need to reduce your credit limit. Like, whatever it is. I think we’re already kind of doing things. One of the other markets during the internet, I’m kind of a product of the .com. All we talked about was disintermediation.
Oh, if you can order something on the internet, why do you need warehouses? Why do you need this? Why do you need that? Why do you have all these people in between? And there’s so much margin in between that gets eroded. And that was the dream. So, Amazon did it a little bit in the consumer side. But at the end of the day, if you’re buying an air conditioning system, nothing has changed. You’re buying it from a contractor that buys it from a distributor, that buys it from a manufacturer, and there’s all these people in between. And if you ask the people in those big traditional businesses, shipping products and warehouses, why do you
00:23:56
have these people in the middle? They’re like, well, we don’t know how to predict demand, therefore we don’t know how to predict capital needs.
So, our distributors act as a buffer for understanding demand and handling the capital needs. So, I look at AI and say, well, AI can handle that. So, I’ve been meeting with multi-billion-dollar distributors that for the last two decades were afraid that Amazon was going to take their business, and now they’re worried that the manufacturer’s going to take their business. And when I talked to the manufacturers, they say, yeah, we might.
PD: So, we are talking already about changes in business models. I wanted to ask you this earlier. Do you see a specific type of business model that is emerging due to AI or Gen AI advancements?
KP: Yeah, I think broadly, the most susceptible businesses are the very traditional businesses. So, that’s why I focus on things like construction, and engineering, and very traditional building products and all of that. But I think there’s a lot of traditional industries that we’re already seeing change. If you think about professional services, accounting, legal, that are very people oriented, and if you’ve ever worked with a law firm, 90% of the billings you get are generally worthless. It’s a bunch of kids billing on your project, doing some research or whatever. It’s really that 10% of advice from the partner and the advice they’re giving you and the insights
00:25:36
they’re giving with their life experience and business experience is really what you pay for. But you have to pay for all these other people.
So, I think when you look at accounting and legal, any professional service, architecture, civil engineering, all of it, they’ve built these apprentice-based programs where you start in, you don’t know anything, and then over the years you move up. If I have an AI agent that I can train once, just like I trained someone straight out of university, then not have to hire another young person, I’m going to do that. And we’re seeing that, we’re seeing, starting to see, I think IBM has come out and said they’re not really hiring anyone new. I think you’re seeing it with the consulting companies, McKinsey, Booze, Bain, they’ve all retracted from hiring fresh grads this year and last year. Some of it’s on the demand side, because also the way they’re getting affected on the demand side is a big company paying McKinsey to do a bunch of research and give them 10 million bucks.
They’re finding out they can get just as much information faster out of Chat GPT than hiring this management consulting firm. So, there’s a little bit of pressure coming in on the demand side where the customer’s like, I can do it myself. And then on the supply side, they’re just saying, why do we need all these kids? We’ll pick the top of the top and what skills are we going to hire for? We’re not going to hire for analytical skills, we’re going to
00:27:10
hire for human skills. We’re going to hire for their ability to build relationships, and manage a customer, and do business development. They need to be good enough, but they don’t have to be the best. They need to be the best at how do I drive business and be business leaders not being the best analytic person. Same thing with accountants. I’m seeing accountants hiring people that are not the best accountants, they’re okay, but they’re very good at talking to customers, and getting customers, and thinking about new businesses because a staff accountant is now an AI agent.
PD: So, we are talking about the workforce impact now, which is kind of the other side of the medallion. So, we talked about the organizational, like changes top down and the impact that we’ll have. And now we’re talking about a little bit the employee perspective and the workforce maybe present students who are thinking how can I actually position myself in the market of the future? How can I actually be hired? What are jobs that are interesting for me? Or what are opportunities that will open up now through all of these changes going on? Do you have any advice in that regard?
KP: Yeah. I think the last, call it five, 10 years, we’ve produced a bunch of laptop workers. There’re people whose full-time job is to update the Asana board up, update the Trello board. That’s their job. And so, I think what
00:28:31
we’re really moving towards, I mean, it’s bizarre to me, but these laptop workers maybe they have English degree and they’re updating, they don’t know anything about software or product development. They’re just in charge of being on standups and updating the Asana board. And I think those jobs are gone, but I, what I tell students and young people, and I’ve got three sons, one’s 24, one’s 22, and one is two. So, the 24 and 22-year-old is really building deep expertise and understanding of different markets models and aspects of that and really understanding those deeply.
And then being a power user of the tools. And what’s going to happen is, I think as long as, and by the way, be passionate and interested in whatever the subject matter expertise is. And I think we’ve gone away from that. And my son did an internship last summer at a structural engineering firm, and he was using AI power user, AI power user in the firm. And he was getting his job done by 11:00am like his tasks, which are like intern tasks. And he was done by 11:00am. And he went to his boss, he said, I’m done. What do you want me to do now? He’s like, why don’t you come to a meeting with me? Oh, we’re an expert witness on this lawsuit.
Why don’t you come to a deposition with me? And his conclusion was, if I was sitting in my cube doing entry-level intern work all day, I would never
00:29:59
have the opportunity to go to meetings to actually get mentorship, to go to a job site, to do all these other things. My son’s ambitious. A different kid would’ve used AI and just sat at their cube all day, and be done at 11:00, play video games and until 5:00 and leave. So, I think that’s going to be the difference is how much people care about what they’re doing and focusing on jobs that where they care what they’re doing and that they’re passionate about, and then their level of ambition.
PD: So, we’re talking a little bit on the attitude side. We’re talking about being passionate, being interested, being curious, being ambitious and so on. How about if you think about skills that you can acquire, like what skills would you recommend like students now to acquire, to be fit for the future?
KP: Yeah. I think people skills are back. Writing skills are back. As an engineer, I type things into Chat GPT and I don’t get great answers. My wife, she’s a writer. She’s one of the best people I know at the English language. When she prompts, she gets better results. You would think, oh, engineer code. Like, I would get better. I don’t know how to describe, I don’t have the base of vocabulary to describe things in such a way to get the feedback that I want. She’s much better at it. So, it’s interesting we might be experiencing a new renaissance where words, and arts, and creativity is the new thing. And the idea that the gap between idea and
00:31:38
execution is shrinking. And so, being that creative idea person and being able to execute on it quickly with these AI agents looks more like the future. And then I think also, like, as leadership has to ask less about, I talk about a job description these days looks more like instructions to a robot. It’s not a person, its activities, and tasks, and capability.
And I think if you look at that, if an AI agent can do all those tasks, then what are you hiring for? Who are you hiring this person? And now I think like, culture starts to matter. How’s this person going to add to the team? I’ve been in companies, I know, the biggest company I ran was 1200 people. I had people that were okay at their job, but their positive nature and they’re working hard and being positive, I kept them around. I mean, I could have hired better executors, but they somehow made everyone perform a little bit better. They somehow made the entire team of people that were very good at executing a little bit happier to be at work. And there’s value there. So, I think there’s a reason why I named my book what I did. There’re these intangibles. The intangibles are, what are going to be the difference makers? Because all the tangible, tactical stuff, it’s a level playing field. Everyone has the same things.
PD: Earlier you mentioned one example, writing skills, arts, creative skills. I think it’s a little bit surprising answer. Everybody sees how amazing Gen AI is becoming at that, lots of artists are really scared of being replaced. In
00:33:27
your vision of the future, do you see a place where AI and humans compete, where the human has to outperform AI in some way to be relevant in that market?
KP: I mean, in the commodity world, yeah. I deal with a lot of architects, I’ve worked with Frank Geary, and I used to work [inaudible 00:33:48] back in the day. And these are massive creative people. They never worried about anyone copying them. They never worry about that. But there’s also architects that build fire stations [laughter], they design fire stations. Those are easily AI friendly, AI can do this. So, I think there’s a shift at. But to your point though, I have this friend, he’s like this famous Hollywood stunt man, and he’s, my age. He’s like 53, 54. He’s like, KP, I can’t keep jumping out of buildings. Like, I’m getting too old for this. I don’t recover like I did. I can’t keep crashing motorcycles for a living, it’s just too hard. And I was like, well, what do you see happening? He’s like, well, I’m writing screenplays and I’m designing being a stunt designer. ‘Cause I do that. And I’m like, oh, how are you doing that? He’s like, I’m using Sora.
I’m like, really? I was like, surprised. I mean, ’cause these are, if you ever met stunt men, they’re kind of rough and tumble type of people. You kind of don’t even imagine them using computers. They’re just these very physical people. He’s like, no, no, I’m using Sora. I’m creating my entire storyboards with Sora. And then he is like, I’m also doing stunt design in
00:34:56
Sora, because it turns out Sora is actually pretty good at physics and body movement. And so, he’s like reframing his future by using these tools. And he’s like… So, I think there’s this idea that maybe I can’t jump through a building, but I have ideas of how you could jump through a building that creatively would be interesting. And maybe I’m feeding that in the Sora. And maybe you’re not great at painting or photography, but you have ideas. But the ability to execute an idea from your brain to your fingers, maybe you don’t have that.
DD: I would like to take a brief moment to tell you about our quarterly industry magazine called The Data Scientist and how you can get a complimentary subscription. My co-host on the podcast, Philipp Diesinger is a regular contributor, and the magazine is packed full of features from some of the industry’s leading data and AI practitioners. We have articles spanning deep technical topics from across the data science and machine learning spectrum. Plus, there’s careers advice and industry case studies from many of the world’s leading companies. So, go to datasciencetalent.co.uk/media to get your complimentary magazine subscription. And now we head back to the conversation. You mentioned the book, “Creating the Intangible Enterprise,” KP. What are the key principles or intangibles you refer to in your book that people should be aware of?
00:36:31
KP: Yeah. I think a few are the ability to generate new business models by deeply understanding markets. So, it’s all going to be about not just your markets, but understanding everybody else’s markets. Two, I think the creative class, integrating the creative class into your organization. You might be a chemical manufacturer. It doesn’t mean that you don’t, I think AI’s going to allow the creative class and people that have ideas that maybe don’t have a chemical engineering degree, they can actually add value because the chemical engineering is going to be done by AI. So, I think what you’re doing is focusing on attracting more of the creative class more than ever for every industry. Third, I think I call it cultural alchemy.
We have a way of scaling companies and hiring people that all look the same and all have the same skill set and the same upbringing. In my book, I talk about GE, GE if you think about it was the example of how to build a large business for years. It was Jack Welch this, and Optimize this, and Six Sigma that, and almost every vice president at GE almost looked the same, dressed the same. It was like this, they were like robots. And where’s GE now? They’re gone. So, I think companies that just sit there and focus on optimization, I think AI gives you that as a core feature anyway. And I think if that’s all you’re using it for, it’s not that helpful. So, I think as you think about innovation and as you think about new ideas, you just have to launch these things and kind of get out of the mindset of
00:38:16
Optimize, and KPIs, and all these, we’ve created all these things to manage businesses that I think are generally irrelevant.
There’re rooms of people creating reports for rooms of people to put into five slides for the board. So, I see a lot of that going away. So, I think it’s going to be more about creativity and niche understanding of markets versus macro understanding of markets. This macro, looking at a macro market, that’s fine, but if you can understand all the, there’s a saying in niches create riches. And the reason most companies avoid those niches is that it’s too much trouble. It’s too difficult. If we’re a billion-dollar company, oh, let’s go after a $50 million market, they like, it’s not worth our time. We can only go after $5 million markets. Well, now maybe with ai you can’t actually go after niche markets. So, I think it’s really going to be people that have a good grasp of niche markets and understanding of those markets.
DD: And in the book, you talk about the four stages of technology progression. Can you perhaps talk about that, give an overview, and is that something that can help people make sense of how to prepare for this AI driven future?
KP: Yeah, I think so. I think this idea like we were talking earlier, where are we in the AI cycle. And there’s people that think we’re there like, oh my gosh,
00:39:44
we’re here. There’re some people that think, oh, I don’t think it’s going to last. I’m waiting for the AI winter. So, I think that we do know, we’ve seen this movie before when spreadsheets came out, Lotus 1, 2, 3, Quattro Pro pre-Excel. These things came out and accounting firms believed that they were, everybody said, oh, accounting is out of business. Spreadsheets will replace accountants. That didn’t happen. It adapts. Arthur Anderson accounting started Anderson Consulting. They implement accounting systems and technology, and that company is now called Accenture. So, I think understanding where we’re at and what we’re going knowing that you do have time. You shouldn’t give up, nor should you be too relaxed about it.
It’s really understanding the cadence of how things are going. And it’s a lot to keep up with. And I think the difference, during the .com and the internet, there was just a lot of disbelief. There was tremendous disbelief that any of this mattered. Oh yeah, we’re going to buy houses on the internet. Sure. There was just a lot, oh, I’m going to check my bank account on the internet. Sure. Like there was just a lot of disbelief and all these ideas. And I feel like this AI cycle, people have gone the other way. I think they’re not believing enough of how fast it’s going and how to think about adoption. But I tell all my clients and investors, start small, get engaged, understand it. ‘Cause I’ll tell you, I think the funniest thing I see
00:41:24
when I talk to CEOs, not since the Blackberry, have they been this excited about a technology.
IT is not going to the CEO and saying we should do AI, give me a budget. The CEO is going to IT saying, what are we doing about AI? I just downloaded Chat GPT, I think it’s amazing. I think it’s the future. We need to deploy a million dollars to understand how we’re going to use AI. So, it’s not coming from the nerds, it’s coming from the business people. And that’s very different. The internet was, IT saying, I think we should have a website. We should have an intranet, it was IT driving it. CEOs didn’t care, the boards didn’t care. They’re like, what are we some kind of .com, we’re not a .com, we sell lumber. That’s not happening this time. The CEOs are driving all the behavior right now. And I think that’s the difference. Right or wrong, when your boss tells you what to do, you do it.
DD: So, you’ve talked obviously about what is different, I mean, what do you think AI won’t change? ‘Cause we’ve talked a lot about the changes that are coming. What’s going to stay the same?
KP: Yeah. I think what is going to stay the same is a little, unfortunately right now, I don’t think it’s a good thing. I think a lot of the human connection that we’ve lost through the pandemic, through social media, through all these stuffs. We’ve, as humans become more and more, we’re two
00:42:54
minutes from each other, but we connect via Facebook. It’s that thing. I don’t think AI’s going to help that, unfortunately. I think when there’s fewer and fewer reasons to connect with people, that’s one of those things that I get worried about. I think it’s going to be bad. That just the human connection will continue to be a challenge.
And I think there’s some markets, there’s a lot of people super excited about humanoid bots, making their bed, doing their dishes, doing their laundry. I think we’re a faraway away from that because quite honestly consumers don’t have the budget that a corporation has. So, I think the first deployment of a lot of the humanoid bots, even like when Elon Musk at Tesla says that they’re launching a humanoid bot next year, which we’ll see, it’s for factories. It’s not to be at home and take care of us. So, I think there’s some social challenges ahead of us with all this.
DD: And you mentioned earlier, obviously we talked a lot about AI agents. Have you seen any really compelling implementation of AI agents either in your investor circle, or startups, or large companies? Or is it just too early for those things to be really making an impact?
KP: No. We’re seeing some interesting use of AI agents, once again, by executives. Executives shifted from, I want to see my financial reports on my iPad to now they’re saying, I don’t want to ask you for financial reports.
00:44:33
So, you’re seeing a lot of AI agents where the CEO can say, who are our 10 best customers? How much are they spending? They can just interrogate their data and not have to get someone to create a report, and do all these things, and ask 10 people. They’re interacting directly with the data without a buffer. So, we’re seeing a lot of that. We’re seeing a lot of that in the enterprise stages. We’re also seeing a lot in computer vision, and I’m a little biased because of my industry.
I deal with construction, and buildings, and bridges, and sewers, like physical things. And of course, computer vision has massive application around physical things. And I think where that’s manifesting is how we think about inspecting, whether it’s inspecting a bridge, inspecting a road, inspecting an airplane. We’re just seeing a lot of computer vision at use. I think most people thought about computer vision and they thought facial recognition, that type of stuff. But where you’re seeing it used these days is on everything. Warehouse management, you’re seeing a ton of computer vision, safety management, you’re seeing a ton of computer vision. So, that we’re seeing a lot of practical applications of computer vision.
DD: And interrogation of financial numbers and data. What are they doing? Are they using Chat GPT for those executives that you’re seeing?
00:45:54
KP: Yeah, they’re using some, either an OpenAI interface or a Llama interface. I’ve been really impressed with how quickly Llama is moving and its open source. It’s a challenge, during the .com, we all kind of picked sides, either open source or Microsoft. I was very open source early on. And then when I started having to hire people, I couldn’t find enough Linux developers to scale with. So, I had to shift things over to Microsoft because I could hire lots of, there was a labor force, even though I thought it was an inferior product, but I can hire people. So, I think it’s going to be interesting to see how enterprises. They’re all playing with open AI because it’s easy out of the box, but we’ll see what those costs look like. But Llama’s doing some interesting things. But I think if you look at what people are, people just want access to their data without about a lot of intermediaries looking at it.
PD: Imagining data, you talked about earlier data as the new oil. What’s your stance on data quality versus quantity?
KP: I think that’s one of those things where when you look at like quantity of historical data, I always have a question, like it’s historical data. Is our past really predicting the future when it comes to data? Given how economies shift and the types of data you collect changes over time. I’ve seen some people really using AI to say, let me parse data that’s not of quality. Extrapolating data sets that may be weren’t that clean and coming getting some interesting results there. So, I don’t know. I think we overplay the
00:47:40
quantity aspect. You look at a company like Amazon, they have a lot of quantity of data, they have a lot of data. Do they need that much data. Does the quantity of data have diminishing returns? There’s a tipping point where you need to have enough data to be able statistically to be valuable. But then at some point it doesn’t matter. The increase of data, having twice as much data maybe doesn’t make a difference. It’s diminishing returns.
PD: Yeah. I have two more questions. Maybe quickly. One question is around the regulatory environment that’s shaping up, like what’s your stance on it? Can it have a positive impact on the development of business models and let’s say maybe the utilization of AI, maybe even the commodity AI, or is it at the moment developing not fast enough or in the wrong direction? Like what’s your stance on that?
KP: Yeah. Look, I think what we continue to see is very large companies are embracing, and doing and experimenting, and have the resources to experiment and fail. And, oh, let’s throw a million dollars and see what happens. They can do that. And then I’m seeing small companies, SMB type companies really using different tools for, very effectively. There may not be writing code, but they’re using these tools to be very effective. I think it’s the companies in the middle that are really struggling where they’re so big, we have to do things the way we’ve always been doing
00:49:18
them, but they’re not big enough to take the risk to do something differently. And so, I think it’s kind of this dumbbell approach where on either side of the market, there’s a lot of adoption on a lot of interest and effective use of.
PD: Do you see anywhere a lack of adoption due to the fact that the regulatory environment is still quite uncertain?
KP: Yeah. I think anything around consumer data is really tricky these days. I think it’s always been tricky, but when you look at privacy and data, privacy and how you’re using my data now all of a sudden with AI, I can do bigger things with your data. There’s a question of, if Amazon delivers, I wear one of these Oura Rings and I have all these IOT connected to me and things. If Amazon delivers lunch to me because it senses my blood sugar’s low, is that helpful or is that creepy? I don’t know. In one hand… And I think these are the constant conversations with consumer privacy and data is at what point does it go from being helpful to being intrusive? So, I think there’s a little bit of pause there. And then I think the second part too, who owns this data. I was talking to a large credit card company and they lease a lot of data.
So, now they’re leasing data from other data providers, how are they allowed to use that? You know, how are they not allowed to use it? And I
00:50:59
was at a conference the other day and the CTO of a big software company was asked by his customers at this user conference, are you using my data to train your AI? And his answer was a non-answer. Oh, well, we use synthetic data based on customer data, but we don’t actually use your data. And then he is like, have a nice day. I’ll see you later, and left [laughter]. So, I think how you’re using my data, and I think there’s this weird thing too. I think customers, whether it’s a business or a consumer I think that their data is more proprietary than it is, especially corporates. Corporates believe their data is so proprietary that it’s so valuable. And I don’t think it’s always the case. In fact, it’s rarely the case that it’s so valuable.
PD: What are some future AI trends that you are really excited about?
KP: I still get excited about computer vision. I’m just huge fan of that. And I think autonomous robots, I think we’ve had fringes of robots and robotics that have been very valuable. In industrial situations, we’ve invested in three companies that are in the construction space around robotics. But when you really want to look at an autonomous robot that can adapt to variabilities. It’s one thing to have a robot that just puts buttons on a shirt all day. That’s different. But if you can have a robot that puts buttons on a shirt, and puts on a zipper, and does something else. There’s utility to
00:52:34
these capital investments. So, therefore it can adapt to different environments and different orders and use cases.
I think that’s where AI is going to be really strong, because in robotics, the idea of buying a single robot that does one or two tasks, it’s very expensive. And only certain companies can scale at that level to say, I want to buy this robot for a million dollars to put that sews buttons on shirts. So, I think with AI combined with robotics, I’m really excited to see. I’m seeing a lot of things. I get the benefit of seeing a lot of things just that how using AI robots can adapt to an ever-changing environment of both the work they need to perform, or even environmental, adapting to like computer vision. Lighting matters, but how can you use AI to like really not worry about lighting, and adapt appropriately, and run predictions around that. So, I’m really excited about that interface of AI and robotics, I think it’s going to be really, really interesting.
DD: KP, in your book, “What You Know About Startups is Wrong,” what are the key lessons from that book and how has that informed what you look for to invest in a startup?
KP: Yeah. I think a lot of my book is based on my own personal entrepreneurial experiences, and I think one of the things that I really learned was how you manage your relationships. When you’re a founder,
00:54:08
your entire life gets consumed by this startup, and this startup and this problem that you’re trying to solve. You’re just massively consumed by it. And that’s what makes for a good founder. However, it doesn’t mean that your employees are as consumed with it and that they’re going to go do something different. And just because someone’s not passionate about your passion doesn’t mean that they’re dead to you. Like, oh, I don’t like that guy anymore. So, I think the thing as a young founder I say is like, hey, just treat people well on the way in, treat them well on the way out, because you’re probably going to run into them.
The startup community is very small. The investor community is very small, and just having a better idea of like how you want to treat people and your culture tends to become a very important attribute. So, a lot of what I look for is do they have a balanced approach? Are they passionate about the problem versus being passionate about their solution? And I think that’s where a lot of founders really miss the boat is they get so infatuated with their solution that they stop studying the problem. I want to see people that are really passionate about the problem and have experienced it. I think opportunistic founders that identify a problem but don’t really have a personal involvement in it, I think can be successful, but it’s not the same.
00:55:47
DD: And you talk about the investor mantra in the book of the, that they’re betting on the jockey, not the horse, but I think you’ve got a different take on that. Do you want to explain that?
KP: Yeah. I think I say I want to own the racetrack [laughter]. I don’t care about the jocking the horse, I want to end the racetrack. Which I think I’ve executed on a little bit. I have over a hundred investors behind me, all in the same sector, working kind of as a team. So, when we invest, most of our founders automatically have customers built in. They automatically have mentors built in. And as long as you’re in my race track, you’ll do well, so to speak, as long as you’re. But it’s interesting because founders, when you think about it, there’s like generational shifts. It’s a people business. And so, the founders you see today are different than the founders you saw five years ago. And for me at my age, sometimes I forget that a 25-year-old founder five years ago was in college.
And for me five years ago is a blink of an eye. So, really understanding what generation and what mindset they were brought up in really changes like every three to five years in terms of thinking like we went through all of us went through a phase of like, oh, these millennials, they’re the worst. Like there’s all these little tropes that people get into. But I do think every generation has a different idea of what they think a great startup is, what they want to do, what they’re passionate about and what their level of
00:57:35
ambition is. I think ambition is just one of those things as an investor, we try so hard to gauge, but it’s really hard. And so, sometimes that just means I spend a lot of time with people to really understand how they’re thinking about it. I look at founders and I try to look at their own lives and like how it may or may not change.
I’ve had founders that are single, unattached working a hundred hours a week, but what I know is that will change some point in the future. They will meet someone, they might get married, they might have a kid. Because we’re in these companies for five years. A lot happens in five years. And sometimes I feel like it’s my job to look out for like, hey, these life changes will happen. You may think you’re going to work a hundred hours a week for the rest of your life. That is not true [laughter]. And it’s my job to know that that’s not true, even though you may say it. So, I would say there’s a lot of like life experience I try to bestow on my founders and help them navigate, not just… We had one startup that we shut down a couple weeks ago, and in the conversations with the founder, I was like, what do you want to do next?
He’s like, what do you mean? I failed you. I’m like, you didn’t fail anyone. That’s not what happened. Like things changed, the market changed. We collectively made a bet. You made a bet with your time, I made a bet with my money. It didn’t work out, but I don’t see that you did anything
00:59:08
incorrectly. Like you didn’t, oh, there’s some lessons to be learned? Absolutely. But we learned those together. I’d rather invest in your next deal than some stranger off the street. I’d rather back you on your next idea because why would I give up on the learnings that you had, and together, the relationship and how we work together. Now we know how to work together even better. Why would I enter a new relationship when I already have one that’s predictable in many ways?
PD: Yeah. I think also with age that’s not only more commitment maybe to family and so on, but oftentimes you also see, in the right people, you can see that they prioritize better, they become more efficient, they can delegate better. And because of that, the output does not necessarily decrease when private commitments come into play.
KP: Yeah. I think between laptops and internet, like all the stuff that we have that makes us highly efficient and highly, that we can work 24/7 if we want to. I think what that does always say like, constraints are such a great forcing function to prioritization. And you say it a lot with fund, there’s a reason why a lot of companies that get overfunded fail. They pay people too much, they spend on the wrong things ’cause they have so much money. Whereas companies that are a little bit underfunded, they tend to figure it out. Any kind of constraints I think drives cleverness. It forces you to solve a problem when you have constraints, it’s a forcing function.
01:00:56
DD: Okay. So, that concludes today’s episode. Before we leave you, I’ll just quickly mention our magazine, “The Data And AI” magazine. It’s packed full of insight into what some of the world’s best companies are up to with their data and AI initiatives. And we will be featuring an edited version of this conversation in an upcoming shoot. Subscribe for the magazine for free at datasciencetalent.co.uk/media. And it just remains for me to say, KP, thank you so much for joining us today. It was an amazing conversation.
KP: Absolutely. Thanks for having me.
DD: And thanks also to my co-host, Philipp Diesinger, and to you, the listeners, and we’ll catch you on the next podcast.
