Building Billion-Dollar AI Tech | Jess Larson

Speaker A: Welcome to Innovation Leadership. I'm Jess Larson. On this episode we get to have Andy Harrison. I'm going to start off with a little bit of an intro and then Andy, I want you to hit any of the high points that I miss, if that's okay. Speaker B: Okay. Speaker A: So Andy Harrison is a seasoned company builder and investor in technology companies who's created and funded numerous billion dollar plus market cap businesses and managed venture capital and other investment funds in excess of $4 billion on behalf of top tier corporations, pension plans, endowments, family offices, high net worth individuals. He's a seasoned operator and business and corporate development executive who's ideated and closed hundreds of transactions including global partnerships, joint venture collaborations, development partnerships, mergers and acquisitions go to market campaigns valued in excess of 2 billion. He's an engineer and named inventor on a variety of patterns patents. Started off at Royal Bank Canada in programs with Draper, Fish Fisher Jarvison led investments in early stage companies in healthcare, networking, nanotech, fintech and big data. He's a founding member of their VC program scaling a program to over 600 million, investing in AI, healthcare, energy SaaS and managed a billion dollar PE and distressed credit portfolio. As managing director at Fintan, he partnered and helped. At Fintan Partners he helped raise and deploy in excess of a billion in capital. As managing director and CEO at Lapis Advisors, he led partnership and business development in tech and healthcare. At Google as at Google he started in healthcare and AI executing over 30 joint ventures and helped raise billions for the company Verily Ventures and was part of the founding team at Granular Insurance and represented Alphabet Verily on multiple boards. Then as an executive at Google X, he worked in AI data analytics, fintech, robotics, quantum computing, computational biology and other disciplines and helped spin out Sandbox AQ in Quantum and AI and also the AI robotics company Intrinsiq. Now as CEO and General Partner at the VC firm Section32 he focuses on AI data and software application, LLMs, data infrastructure, enterprise solutions, cybersecurity, quantum and computational biology. And that's long. That's a long mouth. Speaker B: Yeah. No. Speaker A: What did I miss? Speaker B: Yeah, well you probably didn't miss anything. That's basically read out my whole existence there. But the most notably I'm known for, you know, having been an Alphabet for five years before this. Part of the team at Google X that spun out Google Life Sciences, that became Verily life Sciences. There I was the head of business and corporate development where I did 13 go to market, 26 partnerships, 7 JVs that helped us raise 3.5 billion in outside Capital from a variety of large investors and scaled the place up. I did help build Granular Insurance in partnership with Swiss Re. And I was at Google X where I work with Sandbox and Intrinsic and some of the other ML and AI teams under the hood at X. And that's when Bill Maris made the offer to come over here, section 32, which I gratefully accepted. And we have been growing and scaling and our performance has been very strong. But given some of our Google roots and our other connections to other big companies like Meta and Apple and others, we do a lot of work in AI and the application of AI and we're early participants in the space as it exploded over the last couple of years. Speaker A: How exciting. So I saw at rbc you started in Toronto. Are you a fellow Kanak? Speaker B: I am, I am. Where did you grow up? I grew up in Nova Scotia. I have Montreal roots, but I am. I am a fellow Canadian. Speaker A: Really? I grew up a snowboarder out in Alberta. Speaker B: Wow. Well, that's good place to learn how to snowboard. Speaker A: Yeah. Although I do like these days outside of Park City because we get the armpit deep powder without the negative 30. So I'm a fan. Speaker B: Yes, we go to Montana also, but they're all nice places. Speaker A: So I really want to zero in on what's most exciting for you these days. Speaker B: What. Speaker A: What are you most focused on? What are you most excited about? Speaker B: Yeah, I am most excited about still the application of AI models. But I think we've moved past just text. I think two years ago, text. Super exciting. We saw the explosion of companies that came out of DeepMind and other places doing large language models. But text is only one data and it's only one human communication. We're now seeing applications of AI to, to agentic behavior. So this is the ability to take action with these models. Physics. So interactions using these models, spatial awareness for robotics and for other applications, mathematical reasoning, biological structure and function, all the way into the more funky stuff around quantum mechanics and new quantum behaviors. And so I think that that's very exciting because I think AI we kind of forget is still in its infancy. And we've explored some of the ways these transformer systems can be used, but we haven't explored all of them. And there's a variety of new applications and a variety of new research coming out that I think is going to be incredibly exciting for humanity, but also for a variety of personal and industrial applications. Speaker A: Yeah. Thinking about investing in AI, what are some of the rookie mistakes you see either from VCs or other investors right now? Speaker B: Yeah, I think some of the calling them rookie behaviors, I'm not sure they're rookie behaviors, but we've certainly evolved our thinking on how to invest in AI companies. I think two years ago, right. We just invested in the best modelers on the planet. I think that was the choice that people had and certainly that has worked out for many of these companies. So rookie or not, that was the choice we had. About a year ago as they started to do more proof of concept work, we started to combine those modelers with operation and business development specialists and really augmenting the modelers with folks who could actually talk to customers and start to get these POCs organized and to really work on gathering the data required to prove roi, which at the end of the day this stuff's software and it has to do. About six months ago, product specialists became more important because it turned out when the POCs convert to actual deployments, you have to pass a vendor audit, you have to understand hipaa, GDPR stock, two, a variety of things that a product needs to be viable. And that's when we brought in product specialists. So it's this alchemy of human beings that we've added to the modelers that have really helped us. Speaker A: And for people who are not specialists in the space, can you explain what POCs are? Speaker B: Yeah, proof of concepts. Sorry. So what happened in the early days when the proof of concepts, you know, we're done. We're really just AI modelers talking to customers and saying, where would you deploy this? Here's the functionality of the system, how would we use it? And then the two partnered together to see if they could prove it to one another that it actually worked. And so those proof of concepts were sort of the early interactions between the AI modelers and their customers and now those are converting into scale and sales. Speaker A: Yeah. Can you give people an example that they might not be as familiar with? Speaker B: When you say example, do you mean in general or do you mean in some of the newer models that are coming out around different things? Speaker A: Yeah, like a real company, a real model, something you guys have something in the portfolio or something that you think people might not have heard of that is a predictor of the future here? Speaker B: Yeah, I think Sandbox AQ is a great example. Obviously I worked with them at Google X and now they're out very well funded. Eric Schmidt is the chairman. It's a very forward thinking business. But they just announced a quantum positioning and navigation tool that uses the magnetic fields of the earth versus satellites in Space. And this is the same way that birds and sharks navigate the ocean. It promises to be much more accurate, less jamal, which is a big problem these days. There are large parts of either conflict areas or urban areas where or there's atmospheric distortion that makes GPS less accurate and this is not prone to that. The problem is you need these new quantum sensors and the amount of data that comes off of these quantum sensors is immense. And so you need these new transformer systems to be able to parse through that data and turn all that data into exact positioning. And I think that's an extremely excellent use of next generation transformers that are working with quantum mechanical and physics data rather than text. Speaker A: Earlier in the year we had the kind of famous UK billionaire investor Jim Melanon, who's been investing in quantum. I would love to get your thoughts as far as how far do you think we are away from a ChatGPT type moment for quantum. Speaker B: Yes, it's interesting because the physics, we're seeing improvements in the physics all the time. But I like to break the quantum space into two buckets. So one is the quantum hardware and two are the quantum applications. So the quantum hardware is obviously making a fault tolerant quantum computer that we can use to do computations that we can't do on classical computers. And quantum applications are the use of quantum data and potentially quantum computers to do new things like the navigation, quantum navigation tool I just told you about a sandbox on the hardware side, I think daily or weekly I would say there's new literature coming out, people are making improvements, we're approaching areas where our error correction is strong enough and it's just slowly grinding away. When we're going to have a quantum computer, I cannot predict, to be honest. We're not big investors on that side just because there are 40 physical approaches. The end of the day we think you're going to buy your quantum compute from GCP or Azure or AWS anyway. So we've stayed clear of the hardware side, but we are extremely interested in the application side. And that's why we've been working with groups like Sandbox, not only on quantum sensing, but on quantum post quantum cryptography. So protecting computers against quantum attacks, biological and materials modeling using quantum techniques and a variety of other applications which you can actually do and simulate on Today's hardware. The H1 hundreds and the TPU V4s and others were designed as physics processors and they're very good at these kind of computations. So we can simulate some of the behaviors of quantum computers long before we actually have a quantum computer. And should quantum computers arrive, these applications will just work that much better. Speaker A: Yeah. For people who don't know sandbox aq, can you give an example of why you know, what's the value of quantum navigation? Speaker B: Well, the value is the jammability, but also the accuracy. So imagine you're an autonomous car, right? Well, atmospheric distortion and other things, GPS can create some variability in where you are. Every millimeter matters for an autonomous car. So imagine you could have millimeter accuracy. The other is for airplanes or defense applications where they may be moving through jammed space. Obviously then it's imperative to be able to position yourself. So these are extremely strong applications of this technology and why it's so interesting in my opinion. Speaker A: Yeah. Thinking about individuals who are getting into investing in AI companies who don't have the PhD backgrounds and haven't spent haven't been a million dollar a year AI engineer somewhere, how would you help them with the decision tree on separating out the companies that sound good versus the companies that have a high probability? Speaker B: Yeah. So the way we've been doing it is to follow the pedigree. I mean, at the end of the day, you know, as of two years ago, there weren't many humans on the planet who could actually build LLM systems. Right. You could say the same thing about quantum, you could say the same thing about computational biology and other areas using AI. There was just a rare, rarefied group of people who, who could do it. And following their pedigree, you know, they tend to move in packs. One person moves and then nine other people move with them. And you could, if you knew the people and you knew the organizations, you could really follow these people through the system and then you could tell the difference between what was cutting edge AI and what was an application of AI. Now that just may be good applications. So you might not need, you know, a rock star to do that. But mostly this stuff was, you know, the really cutting edge stuff was done by the rock stars coming out of these bigger organizations. And so, you know, that's how we do it. We track the people and then the people are referenceable and we can understand how good we think they are at it. But that doesn't mean that the general entrepreneurial community won't find excellent things to do. Yeah, with off the shelf LLMs, let's. Speaker A: Talk about the computational biology side. Can you give the basics for people not familiar to start with? Speaker B: Yes. So the application of AI to biology was pioneered sort of by the group at DeepMind when they first created Alphafold So when the transformer came out in 2017, there was obviously a bunch of use cases for it, from content moderation to predictive text, all the way to the creation of AlphaFold. Now AlphaFold is a program that determines the structure of proteins in the human body. And what they did was feed it with 300,000 solved crystallography structures that humanity had sort of painstakingly figured out over 40 or 50 years. And the crystallography structures told you what the structure of the protein was and they took the 300,000 solved structures and they stuck it into this program and out came approximately 28 million other structures of proteins. And this is Nobel prize worthy stuff in my opinion. Right now. Used to be if you were a PhD student, you would just pick some proteins in a, in a disease cascade and you would study it for six years and hope that one of them had any effect on the disease. Now you can just punch those, all those proteins into AlphaFold and see what their reaction site looks like, what their binding sites look like, that they may have, you know, a much higher chance of having a role in that disease. And then you're not wasting your six years getting your PhD on something that might not have effect. And, and it's really revolutionized medicine. I know companies out there, instead of taking three months to get their structures, have solved it in two weeks with AlphaFold. And so obviously that's in proteins and Google was kind enough to open source a lot of that work. But there work, work being done in small molecules which another big part of the pharma industry. But now we're seeing more advanced work happening and rna, RNA medicine post Covid obviously a huge explosion in MRNA medicines, antibody productions for delivery of agents, medical agents to specific tissues and other things all the way to linker chemistry and biodiversity and all these new areas. So not unlike text to an LLM. And now we're branching out into these new forms of data. You know we, we started with proteins on the AI side and biology and now it's branching out to a vast number of different molecular types and giving humanity more tools to, to generate medicines. Speaker A: Yeah, I have an analogy. I don't want to, I want to have you like tell me a better analogy. Okay. For me it feels like the lack of precision before so much biotech was almost like, you know, trying to figure it out, but with boxing gloves on where now like having the precision. It's almost like a different game. I don't know. Do you see it differently? Speaker B: No, I don't. I think that's an apt analogy. You know, an example is the COVID vaccine. When it first came out, I mean, you know, we had to put in a lot of it into a human system. It was sort of mildly toxic, you know, and it was moderately efficacious. Now inevitably very effective. But if we had the tools that we had, you know, in proteins and others, we might have been able to design a much, much better one, you know, that made people less sick or was more efficacious or so, you know, we, there are so many tools around the control of toxicity, the bioavailability, pharmacokinetics, all this stuff that we have in other areas now need to apply to new molecules. And designing drugs without some of those optimizers is in my opinion, over time, suboptimal. Speaker A: Yeah, you know, the biotech world can't stop talking about AI, but there's definitely like the folks who've embraced it in bigger ways than others, you know, where you're, you know, you're looking at from the other side, the computational side. In your mind, what's different about those biotech companies that have like truly tried to like get this in the bones versus the ones where it's more like a, you know, wallpaper on the outside of what they were already going to do? Speaker B: Yeah, I personally believe that those who are forward thinking have just become more sophisticated in their ability to use it. They're also more sophisticated in their ability to partner with those who are experts in it. And so I think they do have a significant advantage both in the ability to use the tools, but their in house knowledge base and then their ability to get value out of external partnerships of people who may be really pushing the envelope in these things. And it's feeding back into their drug discovery programs, into their clinical trial designs and everything. And so I think they have an edge. Speaker A: Yeah. So across AI, do you guys invest in any kind of AI company or are you staying within certain sectors of AI or. Speaker B: No, to us it's the innovation behind it. So the transformer was the innovation. Its application and spaces just have to be in areas where we understand the commercial reality. So we've invested all across the spectrum, from infrastructure to cybersecurity to applications in the enterprise, all the way to things like quantum and computational biology. So but what we think it is at the end of the day is software. So if it doesn't have a strong roi, a strong margin profile, if it really doesn't drive a step change in business performance for an end user, then we think it's less interesting. We also believe that there is a natural advantage that the large hyperscalers have in this space in terms of talent, the compute, and also the direct channel to customers. So we've been staying out of what we call the zone of commoditization, the areas that's on the roadmaps of those folks, because you'll end up getting some of that functionality either through search, or you'll get it through your cloud instance tools, or you'll get it through your enterprise suite. And so we've been staying more on the application side, the second and third order problems that we know they won't build into and where, where folks can truly enjoy the power of this stuff and doing something that isn't otherwise available. Speaker A: Yeah, well, maybe let's switch gears from the deployment of capital to getting it into the fund in the first place. You know, there's endless information online about how to pitch a VC and there's much, much less online about best practices working with LPs for a fund manager. What are a couple of your best practices that you feel like have helped you bring in such large amounts of money? Speaker B: Yeah, so we consider our LPs an expansion of our network. So we truly value our LPs both in terms of communicating with them, but also in terms of their purview because they get to see an awful lot of things from their perch that we don't always see. And some of them are from families doing incredible things or they're from institutional investors who truly understand the markets and other things. And I think having a two way partnership with your LPs is extremely important. I think it's sometimes forgotten that the value they bring into the relationship also. And so I think we build strong bonds with our LPs and that means they're always interested in coming back in our next fund. And then obviously we build a few more relationships each time. And so it's a methodical, thoughtful approach to bringing capital into our firm. And it's worked since our inception and the early days when Bill was setting us all up. Speaker A: Yeah, and I don't know if this stat was just for emerging managers, but this week I was listening to a podcast of a VC who interviews LPs and I heard the stat that fundraising for VC was down by 90% in the last year. Have you heard that? Speaker B: I have. I haven't seen that exact data, but I certainly know some have struggled. Yeah. Speaker A: When you think about, you know, a 90% decrease, what does that make you? I mean, for one, it seems like with less money going around, there's probably going to be better valuations if you have any cash to deploy. And it will scare a lot of people off from fundraising right now. But as you look at maybe the next couple of years, any thoughts for fund managers navigating a, you know, coming off, hopefully a bottom? Speaker B: Yeah. Yes, Well, I do. I mean, one would be build strong relationships with your LPs, because obviously, if they're there for you through the cyclicality, then you're going to be a much stronger, stronger firm. So that would be advice. Number one, if you're just starting out and you don't have those LP relationships, I think you really have to have an edge and a story that people truly understand, a pedigree and a set of relationships that, that are truly differentiated. And, you know, I think a lot of that differentiation involves, you know, who, you know, who's in your network where you can draw on for, not only for deals, but for expertise when it comes to doing due diligence and other things. And I think if you have a generic story, it's going to be a difficult story to tell. You have to find what's truly, you know, if you're going to set up and walk down this road, you really do have to have a story for why you're different, why you're helpful, and a whole bunch of entrepreneurs will back you up on those statements. Speaker A: Yeah, that makes sense. In your mind, what does relationship building with LPs look like compared to relationship building with portfolio CEOs or other colleagues in the industry? Speaker B: Yeah, I think a lot of it is probably the same. Spending some time together, educating folks on what you're seeing and why it's important. You know, I think sometimes, you know, the LPs want exposure to things, but they want more than exposure. They want to understand. Right. And they want to know why, you know, you would pick one investment strategy over another. And I think there is. I think it's two way. But there's an educational component on both sides where it's sort of your job to help them understand why you're doing what you're doing and why it's unique. And then inevitably, your track record proves that what you said was right and valuable, which is, at the end of the day, what matters. Speaker A: Yeah, you look at the space venture in general, there's so many people that are attracted to it, they want to be a part of it. They're hardworking, they're intelligent, they're friendly, they're connected. And, you know, statistically, almost none of them have achieved what you've achieved. What do you think is different about you? What do you do that maybe not everyone else does? Speaker B: Well, I think, you know, the grass is always greener when you leave, you know, operational roles and go over to vc. After you started in vc, then, then you did make that choice or have to make it. But we're all operators in our company. Like even Bill was SVP at Google, built the whole external facing sort of R and D organization and Google Ventures. Mike Pollini was the chairman of CEO, chairman of Foundation Medicine. Steve Kafka was involved in frive that sold to exactly Medicine. Andy Conrad built National Genetics and then it built Genomics and then built verily Life Sciences and I did as well. We're operators and you know, at the end of the day there's, there's a thrill to being an operator. I could certainly say it from my side. You know, it's stressful, it's hard work, but you get something done and all of a sudden you've moved the needle and then you get 10 more things done and all of a sudden you've changed the trajectory of a company. Now that's an easy thing to say. It's a hard thing to live, you know, and you learn a lot of things along the way when you're a builder and a scaler and you know, I think being able to speak to entrepreneurs from that direct experience set to be able to empirically look back at things that happen to you and help people navigate, navigate what they're doing, I think, I think that matters a lot. But the truth is, you know, venture capital, you go from being a player to sort of a coach or maybe a player coach in some circumstances. And you know, that's not always the best decision for everyone. Being a player is a wonderful thing to do. It's, as I said, it's long hours and hard work but, but it's really, it feels good. But being on this side, you know, I find it very rewarding helping and working with all the, you know, world class entrepreneurs that I get to work with and scientists and technologists. So it's a good fit for me. But I wouldn't say it's necessarily the road everyone needs to take. Speaker A: Yeah, talking maybe more to LPs, thinking about whether it's institutional institutions or individuals or family offices who have been considering getting into the asset class, but they haven't, they're kind of on the fence. You worked in pe, Distressed Credit. You've seen a lot in your mind. What are the Most compelling arguments for LPs making an allocation to venture. Speaker B: Yeah, I think in the earlier days of asset allocation and portfolio construction and all the alpha beta stuff and you know, all the quantitative work that was done on portfolio construction, you know, I think back then there was less of a track record in terms of understanding the performance of venture in a, in a portfolio. And obviously the big endowments and Ivy League endowments and others sort of pioneered bringing a lot of that exposure into a portfolio and realizing that it does drive a lot of alpha and it has performed over cycles that was lesser known 20 years ago than it is now. And so I think we see a lot of LPs now who have an active allocation to venture. What they're looking for within that allocation varies, but most haven't and then some are lagging into it. And I think now that the data is more strong that it's a little easier for people to hive off an allocation. But it usually is an allocation. It's not. I mean, some folks have pushed the envelope on it, but usually it's a chunk of the portfolio. Maybe 5, 10% can be much higher than that. But at the end of the day your job's the same when you're pitching your own firm. It's metrics like dpi, have you returned actual cash back to your investor and how much and over what timeframe and, and tvpi, the value of your portfolio and then inevitably your ability to perform over cycles and through fund vintages and at the end of the day that will, that will sell you to anyone. But it certainly is less common to find someone who has absolutely no exposure to venture unless they were a family that's in venture who has enough exposure on some other side of their, their life that they don't think they need anymore. Speaker A: Yeah. When you say their reasoning for being involved varies. What are a couple examples? Speaker B: Well, it's what they're. Is your question, what are they looking for in a manager? In a relationship? Yeah. Speaker A: I mean, no, even in the asset class. Like why, you know, what are the different reasons they might be in the asset class if they're not? All the same reason. Speaker B: Yeah. I mean typically it's performance. Right. That is the reason and what our responsibility is. Sometimes it's cross correlation or perceived lack of correlation with our assets. So then it fits into an asset allocation framework and then some are interested for their own purposes. If you're a medical institution, maybe you really need to understand what's going on in AI. If you're a large pension plan that that owns a bunch of airports that make most of their money off their garage, then you're really interested in the impact of autonomous vehicles. Right. So there's a variety of reasons why people want to partner with folks who are right on the edge of these developments. And it can range from anywhere from future proofing themselves in other parts of their portfolio to looking for ideas for their manufacturing businesses that will help them with productivity. You know, but at the, you know, at the end of the day, what it comes down to is, is performance. They want this sleeve to drive a performance gain, preferably a little less correlated because of the timelines and the amount of effort that goes into building enterprises. And that's what they're, that's why they're investing in. Speaker A: Yeah. Well, will you give people the website for the fund and then is LinkedIn or Twitter best for you or where should people be following you? Speaker B: Yeah, LinkedIn is probably the best for me. We're obviously on X also and other communication platforms. Yeah, Our website has a background on us and some of our investments. And obviously we want to speak to the world best entrepreneurs, the world's experts in various areas and the best investors who have an interest in us as a firm. And so we're open to all of that. Speaker A: Yeah. What do you want to leave people with today? Speaker B: I think the big takeaway is there's a lot of exciting stuff still yet to do in AI. I think we think we see it as a first cycle that exploded and you know, now maybe we're over investing and now, you know, maybe we're, you know, there's too much infrastructure being built. I think I would remind people what happened in the late 90s, early 2000s with dark fiber. And I don't remember what, everyone was pretty concerned that we put way too much fiber into the ground and it was, it was all dark and you know, no one was using it and the whole Internet had gotten ahead of itself. And you know, I think probably less than a decade later every single bit of that fiber was lit up and we were back to, to laying that fiber back in the ground and cables all over the earth. And so I think that the Internet or, sorry, the AI story is an early one and to look at early cyclicality and something where we haven't even scratched the surface of what we can apply these models to is a little short sighted. And so I'd be looking forward to next horizon where we're modeling not only text, but physical data, spatial data, biological data, mathematical reasoning, and moving into the more agentic stuff. As an aside, you know, we debate the concept of a singularity internally. And you know, a singularity is kind of thought of as a technical moment when, you know, AI reaches sentience or superhuman intelligence. But the actual 1965 definition, which was Irving Good, I believe, you know, he basically said it's an intelligent agent that enters a feedback loop of self improvement cycles such that it ultimately sort of surpasses what will be achieved as a, by a human being or human intelligence. And I think we're almost there. We're already seeing the agentic stuff taking action and entering into feedback loops where it's getting better and better. And so we are still very early in the development of AI, but you're already starting to see the ROI impacts and the productivity impacts and others. We just need the right humans now to step in and make sure we're doing it correctly, thoughtfully, in a way that increases productivity. Speaker A: I love it. I think that's a mic drop. Thanks for making time to do this. Speaker B: Yeah, great to meet. Great to meet you, Jess and I look forward to our next conversation. Speaker A: Okay, bye everyone.

Key Takeaways:

AI is still in early development, with significant potential beyond its current use in language models.
Successful AI investments require a balance of technical, operational, and product expertise.
Pedigree and proven results are crucial when selecting AI investments.
Relationships with LPs and understanding their goals can drive successful venture capital fundraising.
AI has transformative implications across industries, including biology and quantum applications.

Introduction:

Jess Larson hosts an insightful episode of Innovation Leadership featuring Andy Harrison. The episode dives into AI's expanding role and its impact on various industries like quantum computing and computational biology. Larson and Harrison explore what makes successful AI investments and strategy, as well as the dynamics of venture capital funding.

Outline:

Introduction and Guest Overview
AI's Current and Future Potential
Successful AI Investment Strategies
Funding Challenges and Relationship Building
Implications of AI in Different Industries
Concluding Thoughts on AI Advancements

Section Summaries:

Introduction and Guest Overview:

Jess Larson introduces Andy Harrison, highlighting his impressive track record in technology investment and company building. Harrison has a background with Google X and his current role at S32. He is known for his work in AI, healthcare, and quantum computing.

AI's Current and Future Potential:

Harrison expresses excitement over AI's potential to go beyond text. He discusses how AI models are evolving to include physics, quantum mechanics, and biological applications. Despite being advanced, AI is still in its infancy, with vast opportunities for new applications.

Successful AI Investment Strategies:

Harrison shares insights into making wise AI investments. Early investments focused solely on modelers, but now require teams with operational and product expertise. Understanding data, customer needs, and product viability are key to successful investments.

Funding Challenges and Relationship Building:

Despite the drop in venture capital fundraising, Harrison emphasizes building strong LP relationships for sustainability. He advises managers to differentiate themselves with clear strategies and leverage existing networks. Performance and established track records are vital for standing out.

Implications of AI in Different Industries:

AI's broad applications are transforming sectors like biotechnology and quantum computing. Examples include AI advancements in protein folding and novel navigation methods that could surpass existing systems. AI tools are recalibrating how industries address complex challenges.

Concluding Thoughts on AI Advancements:

Harrison highlights the transformative potential of AI's next phase. He argues that AI's current cycle is an early indicator of its broader possibilities. The increased integration of data types will fuel novel advancements, reshaping economic and industrial landscapes.

Quotes:

"AI's infancy means opportunities are vast beyond just language models."
"Following the pedigree is key in distinguishing AI's cutting-edge from its applications."
"Strong LP relationships are foundational for navigating market cycles."

Audience Insights:

This episode is invaluable to entrepreneurs and venture capitalists seeking insights into AI investment strategies. AI's transformative potential across fields offers guidance for industries looking to adapt. Prospective investors and industry insiders can benefit from understanding the nuanced dynamics of successfully selecting and funding AI companies amid ongoing innovation.

Resources:

AlphaFold: An AI program for modeling protein structures, widely used in biotechnology.
Sandbox AQ: A company known for quantum applications, including navigation technology.

Ask Bash:

How can industries outside technology leverage AI advancements for strategic advantages?
What are the emerging roles AI might play in reshaping global economic systems?
How can startups differentiate to attract venture capital in a crowded market?
What ethical considerations should guide AI deployment in sensitive sectors like healthcare?