The trait a Silicon Valley AI investor rewards most in a management team has a name almost nobody uses: XQ. IQ brings the idea and EQ reads the people across the table. The execution quotient shows in something as small as how fast a team follows up after a meeting, and in her experience the people who score high on it have no patience for those who do not. Meanwhile the money in AI moves in waves, from models to tools to packaged products, and only some of those waves leave room for an early-stage fund. Ivy Nguyen is a Partner at GFT Ventures in Palo Alto, the early-stage firm founded by former NVIDIA and Samsung executives to invest in AI and frontier technology. She has looked at more than ten thousand AI pitches. In part two she explains her place on a board when investors and customers are angry at the same time, why market pull is a strong but incomplete signal, how she tells a feature from a platform, how a no from her firm is framed, how AI booms move from models to picks and shovels, why AI takes over tasks long before it replaces a role, and why care coordination is worth more in healthcare than AI note-taking. Part one is EP 184: [EP 184 LINK] WHAT YOU WILL LEARN What a good investor does when a company’s investors and customers are angry at the same time How IQ, EQ and XQ, the execution quotient, show up in a management team How to tell a feature with a short life from a platform that lasts Why AI booms move from models to picks and shovels to vertical applications Why AI removes tasks long before it replaces whole roles BEST QUOTES (00:04:30) “The CEO is very much the captain. I am merely a passenger, a very annoying backseat driver.” (00:09:25) “XQ is execution quotient. How quickly do they follow up after meetings?” (00:22:57) “I want to be wrong. Then come back and show me, and I will continue the conversation.” (00:33:29) “People who struck gold figuratively were the people who were selling picks and shovels.” (00:46:24) “You can’t automate, and you can’t replace, what you don’t understand.” TIMESTAMPS THE INVESTOR’S SEAT (00:00:00) Intro to part two (00:01:39) When investors and customers are both angry (00:02:31) Captain and backseat driver, the investor’s place HOW SHE JUDGES A COMPANY (00:07:42) Market pull and the execution quotient (00:12:28) Christian Soschner tests market pull against NVIDIA and Amazon (00:16:33) Feature or platform, the follow-up question she always asks (00:20:52) Direct and kind, how a no from GFT is framed THE AI MARKET TODAY (00:25:56) Inside GFT Ventures, a contrarian AI fund (00:30:37) How AI booms move from models to picks and shovels (00:36:06) Vertical AI and the data a competitor cannot buy (00:40:15) Why AI takes over tasks long before whole roles (00:49:54) Care coordination beats AI note-taking in healthcare WHAT COMES NEXT (00:55:44) Her ten-year prediction, the future arrives faster (00:59:45) How to get a busy investor to answer your email (01:01:00) Closing thoughts Ivy Nguyen at GFT Ventures: https://www.gft.vc/ivy-nguyen Ivy Nguyen on LinkedIn: https://www.linkedin.com/in/ivyknguyen/ Newsletter: https://lsg2g.substack.com/ Follow Beginner’s Mind for long-form conversations on leadership, capital, technology, and the people building what comes next. Ranked in the global top 10% of all podcasts, named #1 Deep Tech Podcast in 2025, and #8 Deep Tech Podcast in 2026 Send us Fan Mail Support the show Join the Podcast Newsletter: Link Your Next Move: Link
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
Christian SoschnerThis is part two of my conversation with Yvin Goyen, partner at GFD Ventures in Halo Alto. She has backed AI companies since 2017 when another investor told her AI was too small a niche to dedicate the fund to. In part one, she explained why Silicon Valley Money backs the vision before the plan, how a failed first company led her into venture capital, and the test her firm applies to every AI company.
Christian SoschnerA billion dollar exit inside a 10-year time window built on proprietary data. If you have not heard it, go back and start there. Everything in this half builds on that test. From here, the conversation moves from how she picks companies to how she judges the people who run them and to where the AI market goes next. One answer reaches back to the California gold rush.
Ivy NguyenThey weren't the miners who found who struck gold, literally, right?
Christian SoschnerPeople who struck gold figuratively were the people who Another goes straight at the biggest fear in the AI debate. In her words, my next question was what an investor does in that moment. I recorded this conversation as a live stream in October 2025, and this is the replay.
Christian SoschnerThis is an interesting point that you mentioned that uh an entrepreneur in the situation that the investors are not really happy. And uh, I think they show it, so it's not just uh this this uh British unhappiness where you don't realize uh are they happy or are they not happy, are they mad? Uh they're really mad. And then you have the customers, they are equally mad, they bait, they didn't get what they asked for. And this was a caddo.
Ivy NguyenUh yes, that was my cat.
Christian SoschnerUm the entrepreneur is in uh in between them. What do you see as your role as a VC as an investor when you have to deal with such a situation? Are you the capital allocator who says, okay, well, I write it off and uh screw them and uh walk to the next opportunity, or do you see more engagement from a VC side in that situation?
Ivy NguyenWell, um as much as possible, I tried, and and that was what led me to uh that um that firm where I was working with auto um auto automakers and and tier suppliers. Um, is that it's I I think it's my job to educate myself, right? To better understand buying behaviors and um my entrepreneurs' prospective customers, what their expectations are, um, you know, what uh what their pace of innovation ingestion is, um, and how they are measured so that I can make the um as much as possible the right decision, or at least avoid the the uh the ones that are doomed to fail right out the gate.
Ivy NguyenRight. But once we invest, um because so like much of my business is only so much of my work is only possible when two things align, when um when my limited partners, when our investors want to give me money, as well as when founders want to want to take my money, right? Otherwise, uh if like one if um if one of those two things fail, I'm out of a job. Once I commit to an entrepreneur, I have to see the course, right? Uh it's my responsibility to not only not only to my limited partners, the people who invest in my fund, but also the people who um, the entrepreneurs who I've asked to allow me um to come along for the ride, right, and to support them.
Ivy NguyenUm uh I might encourage them to uh to try to sell the company uh if it's clear that this opportunity is not going to get to further scale. Um, or I might encourage them to, if there's an opportunity to take one last big bet, right, um, to try to accelerate growth and to change the trajectory of this company. Um but um I I'd say that um at the end of the day, I have to remember that I'm not I'm not the captain.
Ivy NguyenThe CEO um is very much the captain. And I am merely a passenger, a very annoying backseat driver, but still very much a passenger along for the ride. Um and it's my job to try to see around the corner, um, to be an extension to the management team by um as much as possible sharing what I know and opening up my Rolodex to get them into the right conversations at the right point in time with prospective customers, and as much as possible sharing what I know, right?
Ivy NguyenAnd so that they can come into those uh meetings with uh an advantage perhaps over um a competitor. But but at the end of the end of the day, like uh I know that uh uh unfortunately, like as much as I'd like to think otherwise, um, it's not my show to run, and I'm not the one who's actually in charge.
Christian SoschnerThat's a good point. Um so you see your role as VC not only as a capital allocator and the cheerleader, but also uh to open your roller decks and help people get the right contacts. Yes. But you draw but you draw a line. So you are not uh stepping into the operator role and say, now I'm I negotiate for you. So you uh make contact, but then you let them run with it and work things out.
unknownYeah, yeah.
Ivy NguyenThe the entrepreneur is um is king, right? I think that that isn't yet another that's another major difference in Silicon Valley culture versus you know other other um like uh innovation markets, other innovation communities.
Christian SoschnerYeah, I think this is an important point that the entrepreneurs, I mean, uh otherwise you have employees and not a team of uh operators and leaders. Otherwise you are made with employees, and I don't think that it's possible for IPCs to manage.
Ivy NguyenIt's not going to stop me right from from telling them what I think and and like what I think they should do. But at the end of the day, you know, I I um I it's it's their business. And like I I have to respect how they want to run it because like they don't answer to just me, right? They, as you point out, they answer to other other investors, others, other board directors, uh and you know, and the employees and customers that they're responsible for.
Christian SoschnerYeah, I like this coaching attitude. At the end of the day, they have the whole information. I mean, they are in there every single day.
Ivy NguyenExactly, exactly. And I'm not, right? I'm just I flow in and out. And so I I have to recognize that I don't know the entire picture.
Christian SoschnerThat's a good point. But then um, when I look from the operator side on the VC side, I mean at the end of the day, it's all about scaling. And uh at one point in time, the team becomes proficient in selling and understands uh to sell the story well and also to the VCs. And you mentioned earlier the scaling part, uh doubling down and uh reinvesting in the team. Do you have developed any decision-making methodology to also separate here those where you think uh they can scale and it makes sense to invest more money and help them scale from those that uh say, yeah, it's just a matter of uh of money?
Christian SoschnerIf we had more money, we could do this and that. But uh your gut feeling says um no, they can't. It doesn't make any sense. So it's better not to reinvest in them. Do we did you develop any methods to to spot or to make this decision fast?
Ivy NguyenUm every every you know every baby is different, right? Every every company is different. Um, and and I want to make sure that I uh I try to approach every decision with the same uh with a blank slate uh in order to um to really give that challenge and that entrepreneur his um his or her do. Um but I found that a common through line to success is that um if if there is true product market fit, you don't have to work that or you don't have to work that hard to um to convince a market uh to adopt the solution.
Ivy NguyenYou might have to work to convince them to adopt your solution, right? Uh out of a of uh uh against a competitors, but um the the need is already there. Um is already there, then getting budget, getting buy-in should be fairly easy. Um and so market pull, like uh is it more of a pull motion, or is it is you are you do you have to are are you pulled by the market? Or are you having to push and kind of squish and squeeze right this the square peg into the round hole of the market?
Ivy NguyenUm one one thing that I try to assess. And um it's um when it works very well, it's quite obvious. When it's not not entirely working, um it's very hard for me to admit that like um that that it's not quite working. You never want to admit that that your baby's ugly, right? Or your baby's baby's ugly. Um uh like or might might uh benefit from a bit of plastic surgery. Uh that's one. But then two, I'd say that um the I I I'm stealing this from a mentor of mine, right?
Ivy NguyenBut I I think about um excellent teams as uh operating on um a a triumvirate of IQ, EQ, and XQ, right? XQ was a new one for me. Let me explain what that means. So IQ, I think we all know EQ is you know emotional intelligence, right? It's going back to what you and I talked about earlier about knowing your stakeholders, knowing your audience, right? Knowing how to how to code switch or how to sell them what they what they are actually looking for and deliver what they're actually looking for.
Ivy NguyenAnd XQ is execution quotient. So how quickly um do they follow up after meetings, right? How um like they they might know everything and like they might understand how to charm you and how to get you excited about opportunity. But if they just sit on their hands um afterward and wait for you to come to them, right? If they're not constantly chasing after opportunities, constantly following up, et cetera, et cetera. Time is precious, right? And like the 10 years might seem like a lot of time, but if you have very high XQ, if your feedback loop is very, very short, right, there's a lot that can be done in in 10 years.
Ivy NguyenAnd I find that people who are very high XQ have no patience for people who are also low XQ. And so the the management teams that are that index very high on XQ tend to move very, very quickly, right? Um, and to I think to your your other your earlier observation, um, is it true that most venture, like many Silicon Valley venture capitalists, would much rather um fail quickly and move on to try the next thing? Absolutely. And I'd say that that that is why high XQ founders are so attractive, right?
Ivy NguyenIs that uh they they are um more likely to hit the accelerator and go straight into that brick wall, hoping, um, hoping that they crash through to the other side or hoping that they they bounce back um and and can try again if um if they don't make it through.
Christian SoschnerI like this XQ term. It's totally new to me. Uh yeah, X Q. And makes absolute makes absolute sense.
Ivy NguyenYep. Yep. I love it too. I've been using it like constantly since uh since I've heard it.
Christian SoschnerUh is there any any book or literature to read about it, or is this uh such a news content?
Ivy NguyenNo, I um in case anyone is wondering, I I got this from Tony Palchik, uh, who leads innovation at Zebra Technologies. Uh so Tony, if you see this, thank you. Um, I don't know where Tony got it from, right? But there's lots of Tony isms that Tony Isms. Tony should write a book.
Christian SoschnerWrite a book about it. Write a book about it. It's good. I'll let him know. I I think it's good. It's a it's a huge differentiator. And when you look here at Europa, at the bureaucratic systems um that are anything but uh high on XQ, uh, they are high on politics, which is understandable in their situation. But as a startup founder who wants to work with VCs, execution is everything. This I think this is the core message.
SPEAKER_02Yes, I'm sorry. I'm sorry.
Christian SoschnerYou mentioned one point that I want would like to pick from what you said. You mentioned earlier that the market pool is a good indicator of a potential scale-up success. I would I would like to question that. Is it really the market pool that indicates that? Well, when I think about the NVIDIA story, for example, I mean in the 90s they almost failed as a graphic cards company, and then Jensen took this bat on CUDA in 2006. And when I look at the books, so uh books have their value uh for reading. Um the stories that are in the Nvidia way, for example, Day Kim and in other books, uh Sebastian Malabi.
Christian SoschnerThey always say that in 2006 his stakeholders were not really happy with the decision. He said, I mean, uh huh. You work with scientists. Um does that really matter? Um, they are not, I mean, universities are not the best payers, they have budget, but they are bureaucratic, they are long, uh, they look more at uh cheap innovation projects. Yes, and he doubled down and he went in and he sacrificed margin as a public company. And the Nvidia story before 2022 was not a huge success.
Christian SoschnerIt was always this uh graphic card company, and they have the cycles, and uh, yes, you need chips, but one dies the next comes. And it took off in 2022. And when I look as a public market investor, when I look at the NVIDIA story and would say I justify my investments with market pool, uh as I was not really there until 2022. Now we have this trillion dollar company, now it's the the most valuable company in the world, but before 2020, it was just up and down, up and down, up and down, and more up than down.
Christian SoschnerUm, is market pool really a good indicator?
Ivy NguyenYeah, it's it's not the only indicator, right? Um, and and I I think I think that's a great um that's a great example to bring up. I'd also um would love to suggest Amazon, right? As another very similar example where um the you know uh Jeff Bezos was famous for his shareholder letters, uh, where he was on the defensive for many, many years, right? Because Amazon just would not turn a profit. And you have Wall Street analysts sitting here scratching their heads, what the heck is this guy doing? Uh and where is this all headed?
Ivy NguyenAnd uh I think it took a very long time for Amazon also to get to the mass, so to go from an online bookseller to the online everything store, right? To the massive um like both uh like physical as well as digital conglomerate that it is today. So I think you're absolutely right, Christian. Um I think that that's that's absolutely fair. Um and uh I think it's also um this might be a good time to also point out that venture capital is really only the beginning of the story for for many um many massive legacy tech companies, right?
Ivy NguyenMy job um is to just get you started and hopefully uh make my investors a ton of money along the way. But many of these lasting platforms, um, they're really only on their first act, uh, like uh even 10 years in. And the truly massively transformative businesses, you know, on the scale of NVIDIA or an Amazon, um, as much as Silicon Valley likes to um you know, likes to idolize speed um and uh like the you know 10-year innovation cycles, they they like innovation could take like massive fundamental changes in innovation um to the point where we are rewiring our economies.
Ivy NguyenThat that still takes many decades, right? Um, as much as I I emphasize speed and and near-term value um in uh in this conversation.
Christian SoschnerYeah, I totally agree. You mentioned this term platform. I think this is important when you want to create value that at the end of the day you have a platform. And the question to you at the end then will be uh how do you how do you identify a platform? Um there's there's always this clubhouse story in my mind uh when I talk about feature and platforms. I mean clubhouse, market need, clubhouse was the thing in 2020. If I judged this company by the market need, I mean the conclusion was clear everybody uses it, this must be a success, and one day they will find a monetary monetization model.
Christian SoschnerAnd I think Clubhouse, I don't know if it's uh it still exists in 2025, uh, it was more a feature on the feature side. Yeah, um how do you distinguish when you look at Bitch's um feature potential from platform potential?
Ivy NguyenWell, I I think that um I I should point out that I see a lot of pitch sex, but I'm more of a consumer of pitch sex. I'm like if any startup founder comes to me and they say, Well, where should I take this? We are going to be in for a world of hurt and disappointment because I am not the like I'm not the the um the visionary, right? My I I will poke, I I am the detractor, I'm the person who pokes holes in your vision. But uh but that's something that um that I that that that actually uh is a major filter that I use uh to to um to go through my my opportunity funnel.
Ivy NguyenUm like the the best entrepreneurs have uh an interesting initial MVP um like um or a feature of some kind um that does something really interesting that gets them a foot in the door. But behind that MVP is a multi-year uh vision, right, as to how that allows, like you know, this initial wedge allows them to get to that that next workflow and that next workflow and that next step in the value chain. They have very strong opinions as to whether to go up or down, right?
Ivy NguyenAnd then how each of those incremental steps leads to greater contract value increase, uh, greater value capture, and also greater business value for their customer. And so behind that, that initial, you know, like wow, like Dazzle MVP, um, whether it's a feature or an initial like appliance or prototype, there is that um, there is that vision for a platform already in place. How we get there might be a bit fuzzy, right? Um, but um, but I think the the uh very there are very few um examples in Silicon Valley of companies that uh started out as a fun game, like a flappy bird, right?
Ivy NguyenAnd and might have, if anyone remembers that, um, that mega um like um near runaway success um that turned into um you know the next um mdzinga or EA games or so on. There's always like the um there's always underneath that the iceberg a massive vision for how do you get to um that that bigger opportunity. Uh and so, and and many times the best entrepreneurs already have that in mind um when they're when they're pitching to me.
Ivy NguyenOr if they don't if they don't make it explicit, one of the things that I might do um is in a follow-up email or in a meeting, just ask them that explicitly. Um, what is the the bigger point, right? What is the the bigger punchline here? Um, like how do you turn this from that feature or from that initial appliance um that will go away in a couple of years or will get commoditized into something truly lasting, like that is very difficult to rip out.
Christian SoschnerThat was an interesting point that uh you mentioned poking holes in the thesis. So you're not a visionary, you are the one poking holes. Um one of my first it reminded me one of my first jobs here in Austria was in a public company at MA in B2P in agriculture, and directness was uh one of their values. So everything was very direct and uh insulting, and criticism was really direct, but it moved people. When you look at your work, I mean there is uh a message, and then there is a way to rob the message into kind words.
Christian SoschnerUh Jansen Huang says he tortures into greatness, and management literature these days writes about empathy. Um how are you in your work? Are you more a fan of uh directness and torturing people into greatness or more on the empathy side when poking holes into thesis?
Ivy NguyenUm I I am a people pleaser by nature, right? And so I I uh try to be kind and I try to recognize that we're all human at the end of the day. Um I also recognize that if I'm too much of a jerk, right? I think like people won't want to talk to me anymore. Entrepreneurs don't want to talk to me, and then I'll be very soon out of a job. Um, and so um, and so I I try to deliver um the feedback with with kindness and with respect.
Ivy NguyenUm, but um because I have limited time and like the and I also respect the founders of televidents. I am direct, right? So I'm I'm not I'm not going to you know torture people um to get them to greatness. I don't have the time for that, especially when it's an entrepreneur that I I won't be able to invest in. But I'll at least try to give them feedback, which is um, hey, this is not like I can't move forward with this um because as a as a we make a decision, at least at GFT, as a collective, right?
Ivy NguyenWe're not lone wolves where if I wanted to make an investment, I will make an investment no matter what uh my my um my partner. to say. I have to get them on board too. So I I will frame it with my I might frame it to an entrepreneur as, hey, I I see what you're like, I see why what you're doing is exciting. If I have the time, I'll say like, and this is what I love about it. But at present, I don't see the bigger picture platform. Or I am worried that historically, companies that I've seen in this space tend to have a leaky bucket issue with customers, right?
Ivy NguyenFor every two customers you on board, you lose one and a half. And so I'm not going to bring this forward today because I don't want to take that risk. But I would love to see you come back and prove me wrong, uh, let's say a year from now, when you when you hit this number of customers or this level of revenue, as much as possible, I try to take at least a couple of minutes to think thoughtfully to give that some thought and say this is what I'm looking for, this specific metric. And this is what I'm going to use that metric for, right? It's not just voodoo.
Ivy NguyenAnd if you can if you can prove me wrong, right? And I'm open to being wrong, then um or I want to be wrong, then come back and show me, right? And then I will continue the conversation. So as much as possible, I try to frame it as it's a no for now. But this is the key if you're going to continue to pursue this idea to um to moving forward to the next step with um with my firm. Does that make sense?
Christian SoschnerYeah absolutely and um before we move to your fund um to your decision making criteria and how your fund is set up and what you do on the market one final question to this empathy and directness scale. When we put it on a scale and say one is directness and 10 is empathy how would you rate the Silicon Valley culture? Is it more on the directness side or on the empathy side these days?
Ivy NguyenThese days especially in 2025 I'd say it's more on the the direct like it's probably um like one it might even be 0.5 right in terms of directness. Well because things are just moving so quickly yeah and I think like it seems like a lot of the market um is uh is coalescing around around AI. There's you know Silicon Valley is very high in XQ right and if anything probably is it might be a bit lacking in EQ sometimes very high IQ very high XQ um sometimes uh need some work on EQ and and as a result I could see um I I've seen people perhaps be overly dismissive right or um sometimes um uh not necessarily as willing to be willing to collaborate um because they they were too fast to dismiss an opportunity or or an entrepreneur.
Christian SoschnerSo uh for everybody anybody who wants to move to Silicon Valley uh either as an investor or as a founder uh if you want to execute fast it's the place to be but have a fixed kit so it's uh it's not easy and that's right yeah have have a thick skin and keep showing up right uh keep showing up keep he you know I I go back to that brick wall metaphor I keep banging your head against that brick wall not the door may not like open or you may not loosen like that first couple of bricks but you will like um you will find uh or you will um you will eventually find a way in or as in the process of uh beating yourself up to a pulp you might uh you might remould yourself into um into uh something stronger that uh with a better idea um that that might have a better chance of getting in yeah it's like the Chinese Charlene temper uh as it was in the movies in the 70s uh you have to sit there in front of the door in winter and wait for six months and finally you'll get in.
Ivy NguyenLet's move to I like that metaphor yeah let's let's move to your fund uh how about you define your fund so uh GAT Ventures is a brand new venture capital firm and we aim to build a uh and hopefully if if um if we we do our work right and if if we have luck on our side we hope to build a multi-generational uh firm that invests in entrepreneurs leveraging frontier tech um and the massive tech shifts to transform sectors right and um and while we are very heavily prioritizing AI companies um in our first fund that's not all of what we do um but with this first fund um we my my partners Jay and Jeff raised it um in 20 at the beginning of 2021 so they were also one of the few people in the valley to have been early investors in AI right I'd say that one of the first true consumer AI companies that um the average person may have interacted with was Soundhound you know that that app that allowed you to uh listen to a song so to speak have your phone listen to a song and it can tell you very quickly you know what that song is.
Ivy NguyenAnd um and they were true believers in AI for the longest time. And when they when they started investing racing for this fund uh and marketing the fund to investors they were contrarians right this was in 2021 when everybody was chasing crypto um and uh I think um everyone was looking at enterprise SAS booms and um we had the amazing luck to have uh so they initially targeted about um I think a little bit under 100 million um because of their credibility um they were able to quickly raise that despite going against the grain on what the valley wanted to invest in.
Ivy NguyenAnd they had the incredible luck to have um uh scheduled their final close of their this first fund the week that chat GPC launched and when that launched oh my god right um like I I I I um I'm a much like less mature person than the my two partners I could I would have been so obnoxious but they I think they they um saw the opportunity and so they extended the fund a bit a bit more and so we ended up with a like a much larger fund size almost double what they initially set up to raise. So we're currently still investing out of that$140 million first fund.
Ivy NguyenAnd what we're prioritizing are uh entrepreneurs who uh are sector experts who have access to proprietary data or ways to generate proprietary data to solve very hard problems using using AI, right? And um in order to to massively transform sectors. So we're agnostic to go to market motion we're agnostic to revenue model we're agnostic to to sector um so but so long as uh we can see that opportunity to to build a clear data mode um and that to leverage AI to drive clear compounding performance improvements for our customers for your end customers um it's fair game and so if you look at our portfolio you'll see a very broad mix from uh one of our most recent investments is a company called Higgsfield and they are a video uh generation um kind of AI platform for uh influencers advertisers and kind of the prosumer user um and then we also have companies like Articite which is kind of the the physical intelligence and and um and kind of intel like uh overall interoperable intelligence layer for hospital assistants and healthcare providers right um very very different sectors very very different I think um demands of the product right uh but the underlying thesis is is that data mode um and uh is driven by AI it's an interesting story love love the story I mean uh chat GPT is in it and uh going from a contrarian position on the market uh to becoming uh being exactly at the right time in the right spot uh seeing the entire industry take up when you look at your fund in 2025 you're still investing so you have uh capital on the side you can invest in companies valuations are high i mean looking at NVIDIA on public markets I assume in the private markets it's not different uh you have a lot of uptake in everything that has AI somewhere in the story how how how do you how do you identify um uh bets still in this time um or do you sit back idle and say uh we wait until the hype cools off and uh come back then as an investor how do you identify opportunities in this market um well i I'd say that that's where the fact that uh I mean it I I was bragging about you know my um my uh advantage in having been to AI earlier right but Jay and Jeff uh have twice the um the head start in in a AI that I do right and I think that that's where um we have a massive advantage um in um in investing in AI is that we've we've seen uh actually mini AI booms and bus cycles before right prior to what we're seeing now when I got my start in AI there was there was a bit of a boom then too um although not uh too at all to the scale of what we see today and how how these AI um boom and bus cycles tend to evolve is initially um general usually there's some major um advancements in the efficiency and the performance of models right um this time it was around large language models last time it was around um a combination of um computer vision with you know uh advances in image net as well as natural language processing um and all um generally investors typically first uh general investors will pile money into investing in the models we're going to back the next company that will drive the uh the winning models um and like because models are fundamental to everything that is how we make the next bet right like this is going to be the the next Microsoft and then you realize that that is a very difficult game to play because how do you how do you continue to push the envelope and say the best in class um with with these models well you need to pay the all like you you need to it's a matter of hoarding all the best talents right in in that um in that speciality and so uh that that's why you see um yeah companies like meta offering uh hiring ai PhDs almost as if they're NBA players right signing uh you know 10 million hundred million dollar signing bonuses um and and then you realize that quickly um what started out as potentially a small scrappy startup game is probably going to be a game won by the massive hyperscalers right and and that's really not a um in a place that a hundred million dollar fund should be playing um especially when some of these rounds that are being raised are larger than our fund size and so uh and and I would say that that is true for GFT that is true for a lot of early stage venture capital firms so the next phase of evolution um tends to um tends to be uh more in the uh abstraction layer the the next layer up in the abstraction layer and so okay well uh most people in going back to the California history right um who came to California and made money in the gold rush they weren't the miners who found who struck gold literally right people who struck gold figuratively were the people who were selling picks and shovels um and you know Levi Strauss right the the denim company um was one of the the like um biggest legacy brands and like money makers sort of um uh emerged from that era of history right because they were selling denim overalls and and um and uh jeans to um to the miners that came out and so like the the the next wave of evolution is well if you can't win on the the the talent game um like then of the the buying talent game then um why don't we sell picks and shovels right to the the people who are going to implement ai and then what you what what becomes reality is that these practitioners of AI or of these uh the the people who are implementing these technologies for the broader market are still very scarce so like the the AI engineers who need those picks and shovels where are they they're not working at a Volkswagen or uh like a McKinsey etc they're all going to to you know meta to Amazon to Google.
Ivy NguyenThey have the money to like infinite like infinitely right to probably build their own tools. The rest of the market um outside of course Silicon Valley and the hyperscalers are still catching up and so there there may not yet be a massive market for those picks and shovels. So then we go to the next layer of abstraction what if we package an AI in a box right for uh various functional groups you know AI for sales teams AI for you know um like uh uh heads of finance AI for lawyers um and I I'd say that that's really where uh a lot of uh generalist venture firms have started to see like actually enduring success because once you get to that level of abstraction, the number of ready customers right that are able to consume what you have today out of the box is that much larger.
Ivy NguyenAnd so the market and the near term revenue generation opportunity um is I think much greater. And that but the challenge that you have is that like every company has some kind of um sales department or at least somebody doing sales every company has someone doing finance et cetera and so could you put is it is it easy to build a unique data set right um that drives performance and advanced like improvement. It turns out that it's it's not right because there's there's like variance on sales data everywhere there's variants on sales coaching data on you know translation data et cetera and so while it's easy to make money um and and to to generate revenue and to grow in these spaces there's very high commoditization risk uh because if if one company can do it, the the second fast follower can copy what you did in a more capital efficient way.
Ivy NguyenAnd then the third one will come in three years later with 200 million from SoftBank out the gate and then they will run laps around you. And so when where venture capital tends to evolve next with AI is in what people call vertical AI, right? Which is you look at specific sectors, specific workflows in each of the sectors where that data doesn't exist um and that data is only possible and it's only created once you start engaging with customers, right?
Ivy NguyenIt's when you uh it's when you you present uh when the model presents some kind of output, but it maybe has like 60% confidence level and then the end user, like that that subject matter expert whether it's a doctor or the hospital administrator or the technician, the electrical repair technician will say, okay, based on this recommendation, yes, I accept it. And that reinforces that confidence level uh and that improves the model. So the next time it outputs, it comes up with a similar output that um that confidence level is higher.
Ivy NguyenThe performance is that much better. And that's how you start building that data mode that I mentioned earlier and that data flywheel so that it makes it virtually impossible or at least very very difficult for fast followers and for well-funded second and third and new entrants in order to be able to catch up and to reach performance parity. So I know I've been talking for a long time so I'll pause for a breath here for you to comment but that's that's like the evolution that we tend to see um and um you asked well what did what did GFT do this whole time?
Ivy NguyenWe've been investing our job is to invest right our investors didn't give us money to then sit out on our hands to what to wager the cycle but we we find kind of pockets where um where we have high conviction that um those common modalities of failure right um for for startups in this as the market evolves uh in its understanding of of AI um uh where we can still find opportunities to invest in great companies because it's not linear and it doesn't all happen at the same time you know there might be markets or sectors where um those opportunities mature faster and those cycles move through faster if that makes sense.
Christian SoschnerAbsolutely it's uh uh was interesting listening to you uh and you're the experts in that field and the this is where it pays off now these days you can find the opportunities because you have the experience already for a few decades in that space so you can um differentiate between which part of this industry is hyped and uh which part still has opportunity and which part needs to grow. I mean what I find interesting when I look as an outsider on that space is uh the hardware investment that goes into that space is scary.
Christian SoschnerIt's so huge. It's so huge. Absolutely and when I look from the business side on the entire sector it feels like the entire business models are completely missing. So how how will how will this ever pay off? And uh if you understood you're right this is exactly the direction that you're working in that you identify opportunities that close the gap between these huge investments and uh the companies who need to monetize that at one point in time and the end consumer.
Christian SoschnerExactly right when we keep a little bit uh looking at this uh these dynamics on the market the bit I mean the the promises of AI were really big two years ago it was like uh everybody will be put out of business doctors lawyers accountants uh is this ever coming um this reality that uh ai can really take over complete job roles I mean currently I don't see it when I use uh chat GPD for example I mean it's it looks always smart but the answers very often in my areas of expertise are extremely flawed and it's really dangerous to put that in charge.
Ivy NguyenDo you really see a um a future where agents can take over entire roles I don't think so in the near term and I I think that the people who have that fear have a fundamental misunderstanding of of what a a role is right there's a um there's a distinction between uh a task to be done in a specific workflow and I think that there there are many um there are many tasks that um can be taken over by software not even necessarily AI right but there's just think of the number of um instances where you might have to copy and paste data from different fields from one platform to the next that might have been like that that could be solved and could be avoided if you just had like a direct integration, right?
Ivy NguyenIf one piece of software could talk to the other piece of software. That's not AI at all, right? That's just like building a software bridge and integration from one platform to another a lot of uh so those tasks I I think uh more of those tasks will go away um but um these roles are a composition of very many tasks very in complexity um and and um beyond the mechanical task of like inserting this thing here or like formatting information from one format to the next a lot of role a lot of work no matter what the role right is is in communication and is in coordination right and is in um in convincing somebody going back to um to IQ EQ XQ to change the behavior right or change a trajectory um to do something that I think is aligned with what you need to be done in order to do your job well.
Ivy NguyenAnd I'd say that we've all seen chat GPT AI automation fall on its face. Like I challenge you to find like like um to like to find an AI, humanoid or not, right, that that can can convince you know that most recalcitrant um stakeholder right or like a a customer who's having like who's upset with you um to um to give you another shot right like I I think that um a lot of I I could see a lot of our roles evolving to emphasize EQ more um and AI um helping uh maybe uh increase the bar on IQ across the board and I think accelerating um XQ um for people who might might be um intrinsically a bit more slow right in their clock speed uh but EQ is still very very hard to do um and and I I think um many of the tasks um like that comprise like um that comprise a little role right um that will not be touched by AI are very EQ oriented tasks.
Christian SoschnerYeah it's really um an interesting I mean it's AI generally speaking when we just uh generalize it it's really useful so I use chat GPD Gemini and Grog um on a daily basis um for work but what drives me crazy with chatGPT is this uh built-in people pleasing attitude so when I want to use AI I want to have an eye to poke holes in my blind spots and uh I want to get uh um a hint where I should look but uh I'm just blind in this area but chat every time it comes back is you're absolutely right whatever crap I put into chat GPT I get back you are so right that's right yes yeah and when I then look on social media and uh um read what technicians are posting I it feels sometimes uh like a flashback to the nineties when the internet was young there are now companies who Think about replacing humans, and this never worked out really well.
Christian SoschnerAnd in the internet area, when I was uh at the university in Graz, I had enough arguments why it cannot work out. But when you look at the eye, my gut feeling says replacing humans will also be a recipe for failure in that area. Um, criticizing Jensen Huang when he says, Okay, um, become um an electrician, um, this is the future, you can become a millionaire when humans for generations tried to get out of these jobs. Um, thinking back, Austria in the 70s was really hard and people didn't want to do it.
Christian SoschnerReplacing humans is currently on the market, and the future that I see is more integrating artificial intelligence with humans. How do you see the dynamic on a value basis in the industry? Are we are we really on this track that uh companies try to replace humans and we create a society where we try to put people out of business and have to learn it the hard way? Or do you see more an integration currently in the industry, humans with AI?
Ivy NguyenI think that there's like as is always the case in Silicon Valley history, there's a massive gap in what Silicon Valley thinks is how the rest of the world operates and how the rest of the world actually operates, right? And I think one of the fundamental challenges of the Valley is that you have people who are very, very good at building software who have no idea about how the rest of the like other sectors work, right? And so like they, if you if you have them build software to build more software, um, as you know, in the case of cursor and kind of some of the like uh runaway initial applications of AI um seem to be, but they're doing all right, right.
Ivy NguyenBut you're you're also starting to see rumbling that that um actually, you know, this code generated by AI and this five coding isn't that good at all. It's good for kind of the near-term appearance of work, but on net, um, it's actually creating a bunch of junk code and bloke code, um, if anything, causing more issues. Um, and I say that uh that same dynamic applies to every sector that is concerned about being um being replaced by AI. I think that there are definitely ways that AI will meaningfully change the work, but a lot of the operators in Silicon Valley who may have only professionally grown up, so to speak, right, in the context of Silicon Valley misunderstand is um like how these workflows even like are even organized, how these value chains are even organized.
Ivy NguyenYou can't you can't automate and you can't replace what you don't understand. And like, and and you know, with all due respect, right? What what like what in the fresh hell could a um like AI PhD um uh in um you know in large language models know about um about doctor-patient interactions? Like they can they can maybe very quickly understand the uh mechanical work of capturing information, but that's only the beginning, right? Um like um uh to uh how does that lead to a diagnosis?
Ivy NguyenAnd a diagnosis is one thing, but treatment requires participation from the patient. How do you convince them to come back for treatment? How do you convince them to stay on a like on the on that prescribed course of of um of medication? How do you convince them to continue to take their antibiotics, even though they're feeling better, right? And and they think they know better, right? That's not something that AI can do. And I think that that that level of depth of understanding the problem of how do we get from, let's say in this example, to continue this example, a patient visiting for a sore throat to actually curing them, right, in a way that like um that doesn't breed you know antibiotic resistant um like bacteria and so on.
Ivy NguyenUm that takes time to learn, right? And and that's not something that uh just shadowing a doctor, uh like the the better entrepreneurs will try to shadow the practitioners in whatever end uh sector it is that they're trying to disrupt with AI. But without true domain expertise, having lived that problem, how do you really know? Right, and that's why you know going back to GFT, we we place so much emphasis on the management teams that actually have real-world experience operating in not only tech, but that sector that that uh it is that they're trying to transform.
Christian SoschnerI see it, yeah. There's a lot of opportunity. I mean, when I when I go back to this chat GPT example from before, I tried to use chat GPT for negotiations because basically they have a pattern. And if I could automate in negotiations some mails with AI, it would take a lot of workload for me. But whenever I try with AI and I I use a mail that I get and think uh have this rudimentary prompt and combine it with AI, maybe something useful comes out.
Christian SoschnerI'm astonished how often AI just gives everything away for free. And this is uh this is this can kill a company in a negotiation. If you let uh AI automated negotiate deals where they agree to everything that a customer wants for free, that's right. Uh the roads to bankruptcy is fast. When I got when I when I used your example that you brought with doctor-patient uh integration, using AI to motivate a patient for adherence to a therapy um sounds easy, but uh when you just make a simple mistake and say it is time to take your one pill and it works for 10 days, and on the 11th day, the I says, and today you need 11 pills.
Christian SoschnerThis can be a dose. Uh do we understand you right that in these scenarios, Gary is where uh the industry now has a blind spot and uh needs new startups, new companies who rethink and their workflows and make that integrate well with the economy and the eye and connect these two worlds into useful business models. Did you understand you right?
Ivy NguyenThat's right. So so um what what I the the point that I I was hoping to make was that um oftentimes what actually the the levers that actually drive business value and actually meaningfully move the needle for for businesses in in these uh very specialized sectors um is not what you might initially think, right? And so a lot of Silicon Valley like supile into um into AI um like uh patient uh visit transcription and AI medical note-taking and turning that into billing encoding.
Ivy NguyenBut um it turns out that what actually makes a lot more money for doctors and what actually helps patients is in the coordination of information, right? How do you make sure that uh, let's say in a um in a cancer case, for example, um the the radiologists, the pharmacists, you know, the nutritionists, the doctors, like there are many, many different stakeholders. How do we make sure that everyone has all of the pertinent information um like up to date at the right point in time and can actually um uh deliver that, synthesize that together and work in a coordinated way uh in order to uh to care for the patient?
Ivy NguyenThat's really what moves the needle. That's really what drives the most money and perhaps the most profit. And that's really is what is win-win-win for the healthcare payer, the healthcare providers, as well as the patients at the end of the day. And while it might seem like the most annoying and like the most annoying thing to a doctor uh might be note-taking, that's just table stakes, right? And and that's not something that actually really, really drives true business value or even actual like um health value at the end of the day.
Ivy NguyenAnd and like and um from an outsider, um as somebody who might not be um a medical practitioner or have has had to um work in healthcare management, um, how could you possibly know, right? Like where where that that real um that real sticking point and that real point of inefficiency is.
Christian SoschnerYeah, no, it makes sense. I mean, this is a lot of for for consulting, consulting businesses who can close this gap and then create startups in that area. Yes, I mean in the past, in the 80s, 90s, um people also um solved the same problems, but um it was a disconnected world. So when a company solved that, it was basically uh their trade secret and nobody knew it. And um, there was a lot uh of uh um uh walls built around that to protect the secret, uh, why they can serve customers so good, especially in consulting.
Christian SoschnerBut when you think now in the AI word, I mean basically everything is a model, it's code, it's data, so everything is online, and the hard thing is always uh the first time to solve such a problem that you describe. But how can can companies protect that at the end of the day? I mean, it's easy to copy once someone solved it. Uh it's it's pretty obvious how to do it, and that they can uh do I miss something? How how will these future companies protect their IP, basically, their their special methods, so that uh other companies don't just copy them?
Ivy NguyenI I think that is the the going to be the question of the 21st century, right? Is what is it, what does it actually mean to protect IP? Um, and you know that's something that I think a lot of the automotive sector and just like the the the world, like the the titans of industry, right? Um like uh who um built their companies on the basis of selling widgets, physical assets, having IP on like tangible technologies, is trying to figure out is well, um when initially they were already like um facing threats from commoditization in Asia, right?
Ivy NguyenLike I I have this new product that I let I um with you know all the the classical IP protection that I've launched on the market. Um and then two weeks later I see like the um like the$10 version of that uh plot the market from Shenzhen, what do I do? Um and and now um they're starting to see a world where um the greater value is uh in the data, but um what what data matters, right? Because there's so much data um and what actually drives real value.
Ivy NguyenUm and if we could um if we could generate this data to your point, um, couldn't couldn't our competitors generate similar data? Probably, right? Uh and so what what um what does winning mean in this new world order? I don't think anyone really knows, Christian. I think that's something that we we still are figuring out. And you know, maybe uh in 21-25, when the economists and you know business writers look back then, they'll they'll have a through line. But I think a lot of the the playbooks are are being rewritten now as we speak.
Christian SoschnerAmazing times, definitely amazing times. We've been talking about RP. It's one of the the reasons why investors can invest because it's just protected at the end of the day, and when you take that layer away. Okay. Um then we will see. We will see, we will find out. Um let's move let's move a little bit forward and talk about the future before we wrap up the session at uh at the end. Where do you see the next 10 years going?
Christian SoschnerWhat's your prediction?
Ivy NguyenOh my gosh. I think my my job is like is um you know a million and one small predictions every day, right? And so like I like well but like I think you're asking me for one one big prediction. Um I think that the future is going to come at us faster and faster. Um and uh because the the tools of production are enabling greater speed than ever.
Ivy NguyenI think previously um I think the the the dot-com boom and cloud was so like and then cloud after that was so transformative because this dissemination of information and the cost of storing information basically went down to like um zero on an incremental basis. Um and now I think the the cost, what's really interesting about AI is the cost of testing different variants and the cost of um change is a lot lower, right?
Ivy NguyenAnd and that was like, in my opinion, the the next point of friction in an iteration cycle. Um so like when I think about iteration loops, I think about it as new idea, testing the idea, deciding whether or not to use that idea based on that the data from that test, and then implementing that idea and then testing it again, right? And what's fascinating about AI is that it is dramatically reducing the cost of idea generation. Then it might not be great ideas, right, for the most part, out of a hundred uh recommendations from Chat GPT, most of them will probably give the house away in the case of a negotiation or or cause you to blow up.
Ivy NguyenBut but there might be an idea that you previously had not thought about before. And um uh does that mean the AI invented something? No, right? But it it's an idea, it's another idea, another hypothesis generation. I think it's just that much faster. Um and uh given that um uh vibe coding makes it possible to test to like very easily put something together, it may not be very robust, but to is put something together to test in the market, right? The cost of testing is that much lower, at least on the software side.
Ivy NguyenUm, and then once you have something that is like potentially workable, um, like the MVP of something that you want to implement, uh it becomes that much faster just to to uh implement it and to to distribute it, right? Um it's only going to make the pace of change that already feels overwhelming today even faster. So I think that's the only reliable uh like um through line in the prediction that I can make is that things are only going to speed up from here. So buckle in.
Christian SoschnerYeah, amazing times, amazing times. Let's see where the speed takes us. Um, Evie, before we wrap up, um, my question to you is um amazing conversation. Did we miss anything? Is there any topic that you would like to tackle at the end?
Ivy NguyenUm we covered a lot over the last two hours, haven't we? Um I don't I'm not sure. Well, I um I'm always available and always happy to uh to um engage with with entrepreneurs and innovation leaders. And so um if um if any if this conversation sparked any ideas or any questions, I'm always happy to to connect um after the fact. But I I think we covered a ton of questions. Um I can't think of a single thing that that we missed.
Christian SoschnerSuper then uh my final questions are logistics, where can people find you? What's the best way to reach out to you?
Ivy NguyenUm I am available on LinkedIn. So if you look for IVNewin at GFT Ventures um on LinkedIn. And maybe you could share that um my LinkedIn um uh profile on um uh on the recording afterward, uh, or my email is simply ivy at gft.vc. So please reach out in in either platform. Um and and please be persistent because I do get a lot of emails. Uh and I I uh I look forward to to connecting.
Christian SoschnerAnd the final question: what's the one tip that you can give uh founders reaching out to you and VCs uh to make sure that they cut through the noise and that uh you read actually their email and their messages? What should I do?
Ivy NguyenHonestly, IQ, EQ, XQ, right? Um IQ is uh like um hopefully it's an interesting idea. Um, or at least like uh think about why the idea might be interesting. That's the EQ part. Um why should it be interesting in me? I I think I I um you you really grilled me, Christian, on like in the last two hours on well, how do I filter for come uh for opportunities and how do I make investment decisions, right? And so uh hopefully, hopefully it's not a secret as to what I'm looking for. And then the XQ part, right? Uh be relentless about following up.
Ivy NguyenUm I I try to look at every opportunity, I try to get back to everybody, but um I also I think in the in the course of our conversation have gotten probably maybe 150 emails. And so it's it's a it can be quite a bit. And so if you uh sometimes things do slip through the cracks. So be relentless, uh index high and XQ and uh and follow up. Um, and and that's how you get through.
Christian SoschnerThank you for this great conversation. I love this angle with uh IQ, EQ, and XQ, and be patient but persistent. Keep executing. This is really helpful.
Ivy NguyenGreat, thank you so much.
Christian SoschnerIvi, thank you. Have a great day, and uh let's stay in touch and talk soon.
Ivy NguyenAwesome. Thank you, Christian. Thanks, everyone. Take care.
Christian SoschnerThree things I take from this excellent conversation. First, cash decides. The most common cause of death for a startup is running out of money. Ivi learned it at her first fund and has kept that lesson through every firm since. Second, data decides who survives. A feature can be copied within three years by a competitor with more money. Proprietary data that improves the product with every customer cannot be bought even with a hundred billion dollars.
Christian SoschnerAnd that is the test her firm applies before it invests. Third, execution shows in the follow-up. IQ is the idea, emotional quotient is knowing why it matters to the person across the table. And XQ, the execution quotient is how fast you send the next mail after the meeting. Her own advice for reaching her follows the same rule. Follow up and keep following up. If it's page at GFD Ventures and her LinkedIn profile are in the show notes.
Christian SoschnerIf this conversation helped you, send it to one CEO or board member who is deciding on an AI investment this year. That single share does more for this show than any algorithm, and it brings the next conversation like this one closer to you. Follow the show to get the next episode, and I will see you there.
Invest in the wrong startup and you lose one times your money. Miss the next Google and you lose ten thousand times. Most boards are built to avoid the first mistake. Venture capitalists play by different rules. They...
In 2017 another investor told Ivy Nguyen that artificial intelligence was too small a niche to dedicate a fund to. More than ten thousand AI pitches later, every pitch deck says AI. Single funding rounds have grown...
A seed investor at Novo Holdings would rather read your rough idea than your finished pitch deck. Most founders spend a year building the deck, forming the team and waiting until the story holds together. By the time it...
What does it mean when the most respected adviser in Silicon Valley tells you that you are the best chief executive he works with, while your company is heading straight into a wall? Bill Campbell said exactly that to...
What do you do on the morning when there are no good moves left? In March 2001, Ben Horowitz took Loudcloud public in the middle of the dot-com collapse. The bankers cut the price from ten dollars a share to six in the...