Cris Valenzuela: AI Creators vs Hollywood Writers; How We Grew Runway into a $1.5B Company | E1054

1. 0:00 – 7:06

   From Chile to Founder of $1.5B AI Company

   1. HSHarry Stebbings0:00

      Does it get easier over time?

   2. CVCris Valenzuela0:02

      It doesn't get easier. You get more used to the pain.

   3. HSHarry Stebbings0:06

      Chris, I am so excited for this. I heard so many great things from so many different people on your cap table, so thank you so much for putting up with my terrible British tones today.

   4. CVCris Valenzuela0:16

      Thank you so much for inviting me. I'm a huge fan of the podcast. Happy to be here.

   5. HSHarry Stebbings0:20

      That is very, very kind of you. Now listen, I spoke to Sunil at Amplify before we started, and he said to me that you and Runway is the perfect embodiment of founder/product fit. So just take me to the founding moment of the company and that a-ha moment for you. Why does he say that?

   6. CVCris Valenzuela0:36

      Yeah. Uh, Sunil is too kind. Runway's, uh, journey dates back to 2015, 2016, where I met my co-founders, Alejandro and Anastasios, in actually art school. Um, we went to- to a school, um, that's a very unique school that combines art and technology. It's been running for 40 years here in New York, um, ITP from- from NYU. And, um, the best way to think about it is art school for engineers and engineering school for artists. And while we were there, we were all coming from all sorts of different backgrounds and experiences from, uh, programming, engineering, business, and we started just tinkering and playing around with state-of-the-art AI at that time. I know- I know a lot of things that happen now, but try to imagine where, uh, models were and where research was, uh, 10 years ago or eight years ago. Um, and we just built a lot of different things. We built... I remember, um, Anastasios built this- this beautiful, uh, video semantic search tool that you could just type a word and it will, like, create a trailer for you, um, combining all sorts of different, uh, input videos, uh, analyzing, uh, sequences and scenes. Um, and so we started tinkering with this idea of taking AI models or research at the time and building creative tools. We built those three transformers. Uh, we- we built, using, um, LSTMs, this co-pilot for writing, and so you could... uh, that runs entirely on the browser using WebGPU, using, at the time, TensorFlow.js, which is- which is a framework that was built at the time. Um, and you could write and then, uh, and then a model was co-writing with you, uh, different sentences, trained on different authors that you liked. And so we were just building experiments and experiments over experiments, and realizing that there was something here, something special. And then we started delving into image and video generation. And again, PyTorch was, like, a year old. TensorFlow was, like, two years old. So things were very different from where we are today. Um, and I think we realized at some point that after building so many things, there was something special around both, um, where things are heading and how much passion we had to- for building these things, and for building an- an- an understanding how to take these models, uh, and apply them in a creative, artistic context. Um, and so I- I think the- the... we didn't found the company, I think the company founded us.

   7. HSHarry Stebbings3:00

      (laughs)

   8. CVCris Valenzuela3:00

      And it was the only way we could, uh, really continue building and- and doing what we're very passionate about, which is, uh, a new type of creative tools.

   9. HSHarry Stebbings3:08

      I love that in terms of the company finding you. Uh, the interesting thing I kind of spoke to many of your investors about was also your background, your not being in Silicon Valley, and you mentioned kind of artistry being at the center of both your education and also how you think today. They called you an outsider (laughs) , I'm sure in the loveliest way. Do you feel like an outsider? And how do you think being an outsider helps you be better?

   10. CVCris Valenzuela3:34

       Yeah. Um, I- I've- I've always felt a little bit like an outsider to be honest, but very proud of it. Um, I have- I have a background in econ and business, but when I was studying that, I was doing film on the side. So that was strange and I felt like an outsider in both of those worlds. And then I was doing art, but also learning programming, so also felt a bit of outsider in both of those. Um, and then I came to the US and started a company while being in art school. That also felt like a bit of an outsider. I think it's, um, it's been a common, uh, thread of, uh, my professional life, I would say, um, to really approach, uh, domains that perhaps you might not have experience with a beginner's mindset, and trying to learn and observe as much as possible. I think being a- an outsider is both a blessing and a curse in the sense that, um, it allows you to really reason from first principles. You discover everything and ask yourself why a lot. Um, at the same time, you take longer than other people because you- you're just experiencing for the first time. But the- I think the key aspect of being an outsider is that creativity really sparks from being able to merge different domains and different languages and different, um, uh, professions, um, in- in a coherent thing. And so Runway, I think it's the manifestation of that. It- it's a company that merges art and science in a- in a beautiful way by being able to be in between. Um, and so you, instead of being an outsider, you create your own world. That's much more powerful, I would say.

   11. HSHarry Stebbings5:03

       You know, in terms of creating your own world, you do that when you kind of grow up in the world as an adult.

   12. CVCris Valenzuela5:09

       (laughs)

   13. HSHarry Stebbings5:09

       I always think that we're all running from something, and that our paths shape us in many ways. When you think about what you're running from, what would you say you're running from, Chris?

   14. CVCris Valenzuela5:18

       Oh. Um, that's- that's a good question. I've... uh, I grew up in Chile. I spent my whole, uh, life mostly in Chile. Uh, and- and Chile is a... I don't know if you've been or not.

   15. HSHarry Stebbings5:29

       No.

   16. CVCris Valenzuela5:30

       Um, it's a very, uh, it's- it's a beautiful place. Uh, I'm a huge fan of the south of Chile, Patagonia. But culturally and- and- and- and, um, as a country, it's a very traditional structure, strict, um, tradition-based country. So things are very rigid and it's very hard to do anything that goes beyond the norm. Uh, and really I felt that, um... I don't- I don't feel particularly passionate about, uh, people telling me what to do and how should I, like, follow or what are the norms. I think I'm- I'm running away a little bit from- from that tradition and trying to figure out more of- of the things I'm passionate about, the things I'm curious about.Um, and that was for me, like, uh, really going to art school, it was, it was a curiosity-driven approach of just, I want to understand how much of, um, AI I can take and put in the art world. And, uh, and that's what I did, and I think that's, that's what I, I'm, I'm really passionate about.

   17. HSHarry Stebbings6:26

       Would you say you're fearful of mediocrity?

   18. CVCris Valenzuela6:30

       Uh, a little bit. Um, I think I'm, I'm, I'm fearful of, of, of people not pushing themselves, uh, both intellectually and, and creatively, um, and maintaining a status quo. I think that's the, that's where companies go to die and where creatives go to, to, to not really progress is when you, when you feel you're too comfortable, um, I get, I get a little bit anxious when I feel like that.

   19. HSHarry Stebbings6:54

       I, I totally agree. I think, yeah, uh, David Goggins, uh, says, uh, something brilliant, which is, "There's nothing more dangerous than a civilized person."

   20. CVCris Valenzuela7:01

       (laughs) Exactly.

   21. HSHarry Stebbings7:02

       Um, (laughs) but I, I do wanna, uh, on, on that, you know, it takes me to performance

2. 7:06 – 11:16

   How do you define high performance?

   1. HSHarry Stebbings7:06

      a little bit. As a leader today, how do you define high performance then?

   2. CVCris Valenzuela7:11

      Yeah, um, I don't think there's one singular thing that I look when I think about high performance, specifically in a, in, in a fast-moving field like we are right now. Uh, but I do think there are a few things that constantly, uh, surface when you think about high performance. I think the first of all, the first one for me is being a, a, a vision and a motto and a, and a value that we've, uh, embedded in the company since the very beginning, this idea of getting things done, just figure it out. Uh, it's become so much of a thing at Runway that I make a T-shirt that says, "Just figure it out." If you learn how to learn, you can figure out anything, and that's like a super power. Unlocking the ability to just do things without complaining, without understanding, or without necessarily f- um, uh, contextualizing things in the sense that you haven't done it before is really a super power, and for us at, at Runway was. I've, I've never started a company, um, I had to figure out how to stay in the, in, in, in the US to start the company. There's so many hurdles and challenges at the very beginning that we just had to figure out stuff. And-

   3. HSHarry Stebbings8:13

      Chris, can I just di- can I just dive in there and just ask?

   4. CVCris Valenzuela8:15

      Yeah, go ahead.

   5. HSHarry Stebbings8:16

      If you learn how to learn, there's nothing that you can't do. I love that. Can you help me understand, how do you approach learning? Do you have a process? Do you have a preference for the types of ways to learn? How do you approach it?

   6. CVCris Valenzuela8:30

      Um, very hands-on. If there's something I wanna do, um, for example, when I was, uh, I was diving into, uh, newer networks and, and really understanding the field, I was coming in from, um, a place, um, of, of, uh, beginner's mindset. And so really what I did before trying anything was I built, uh, from the ground up a neural network. Um, every single, uh, layer, every single, um, uh, function, activation function in between, um, I curated a dataset, I trained a model, and I built everything that I could get me a sense of how the inner parts of a system work, in this case, you know, networks. And so, um, I, for a long time, I didn't know what I was doing. You just try things, you try, you try, you hit walls, but you keep pushing yourself over and over and over and over. Um, I don't feel, uh, I d- I like breadth, so just go and read as many things as you can, and then find a project that you like, and then go as deep as you can on that. Um, I think being hands-on and being active on, on learning by doing something really helps make sense of the world, and that's something I've been trying to do in everything I'm trying to learn since then.

   7. HSHarry Stebbings9:37

      How do you not get dejected when the learning is slow or not happening? It always happens that it's not going in as fast, you're finding it hard to find the resources. How do you not get dejected in those moments?

   8. CVCris Valenzuela9:49

      Find brilliant people. Find people that, um, inspire you and can help you along the way. I think that, that for me has been critical, and that, that really is a, a core aspect of Runway. You, I go into a room and I feel, find myself, uh, with the most brightest, more smartest, and kind people in the world, and they're all there to help you, and you can help them as well. And so if you feel, um, stuck, if there's something that you can't really come across, it's too difficult, just ask someone that is gonna help you. And I think that's, that's become, uh, you need the right people around and be surrounded by, that's, kind people who are gonna be able to help. But, um, it's not that common to just rely on someone else.

   9. HSHarry Stebbings10:30

      What has been the most recent subject that you chose to really engage in a learning process, beginner's mindset on?

   10. CVCris Valenzuela10:37

       Um, I mean, everything, to be honest. Uh, from fundraising to building a company to hiring top talent, everything has been, uh, a journey of learning. And I think really not being afraid to be embarrassed or mistaken, uh, because that's the only way you're gonna do it. Um, more recently, I think we've, we've learned a lot with, with what we're doing with Gen-1 and Gen-2, which are our two latest video generation models, and there's so many assumptions that you, you might think about how these models actually work, but the moment you put them out, most of those assumptions are gonna be changed. And so, we've learning a lot and we're learning a lot about how people are using and how creatives are using some of

3. 11:16 – 18:45

   Why UX Doesn’t Matter

   1. CVCris Valenzuela11:16

      the research we're putting out.

   2. HSHarry Stebbings11:17

      What assumption has changed the most?

   3. CVCris Valenzuela11:20

      Um, around video or about creatives?

   4. HSHarry Stebbings11:23

      Yeah, from Gen- from Gen-1 to Gen-2.

   5. CVCris Valenzuela11:25

      Uh...

   6. HSHarry Stebbings11:25

      What was like, we, we assumed X and it was totally wrong?

   7. CVCris Valenzuela11:30

      I think there's a lot. Uh, one that I've been thinking a lot about is, um, UIs don't matter as much. Uh, I think, uh, uh, first approach to creative software comes from this idea that you have to be extremely thoughtful and extremely diligent on, like, how everything is gonna look and it's how everything's gonna work. But true is, um, since the field is moving so fast, um, the assumptions that you have around how models are gonna be used in a creative context might change rapidly. And so if you build a lot of assumptions on preconceptions on, on really how to get the best out of these models, um, you might be wrong really soon, and you just need to throw it out and start again. And so keeping the value, uh, at the center and less of what you think people might get out of those models is really, really critical.

   8. HSHarry Stebbings12:16

      How do you do that? And what I mean by that is, if you don't project out what they can get, it's kind of like a free-for-all. It's almost like product marketing to a horizontal audience.

   9. CVCris Valenzuela12:27

      (laughs) Yeah.

   10. HSHarry Stebbings12:27

       So if you don't guide the user, how do you give a great UX?

   11. CVCris Valenzuela12:31

       You, you test a lot. Um, like, if you think about it, the first way, the, the, uh, the best, the, the first thing people do when they come across this idea of video generation is they, they, uh, assume that everything they know about film and cinema and film will replicate itself to this new medium. And the thing is, this is a new medium. It's a new tool. It's a new set of narrative possibilities. And so it's very different from everything we've used in the past. The most challenging thing is uncovering those things, because we haven't had the chance of doing it in the past. And so it's, it's like you have yours- you find yourself in front of a camera, the first camera ever, and everything you've been using in the past are paintbrushes and paint tubes. D- a camera doesn't work in the same ways as a paintbrush, and the things that you can do with it are only gonna be uncover and tap the momer- the moment where people get their hands on it. That's how Cinema 4D g- was born. You have so many people trying this device and coming up with new forms that are beyond your comprehension or beyond what the creators even imagined. And I think that imparts a lot of what we do. We need to build this camera. We need to put this device into the hands of more people so we can collectively, as an artistic community, figure out all the primitives and UIs and narratives that will emerge from it.

   12. HSHarry Stebbings13:48

       What do you think is harder, distribution or education post-landing?

   13. CVCris Valenzuela13:53

       I think education is really, has been really challenging these days, uh, really helping people understand where things start, where things are heading. Uh, we've, we get a lot of comments around, for example, "Hey, Gen-2 is fantastic, but it can only generate four seconds. Um, and it, I can't use it because four seconds." And I'm like, "Yes, but remember, this is two months old, and it's gonna change." And today, w- we're recording this, we, you just changed it. You can now generate 18 seconds. And so all the assumptions that you had around how you couldn't use it because... It will just change, just completely, like now it's a new open set of possibilities. Um, and so educating, I would say-

   14. HSHarry Stebbings14:31

       Can I, can I, can I, can I ask you, I, I think straightaway to Dylan Field at Figma with this. And what I mean by that is, like, they have really heavy build-outs technologically, and you chose to kind of build in public where, bluntly, you know four seconds is short. But it's better to be out in the wild and have people use it. Dylan almost took the other side, which is, really didn't release for two to three years until it was much more perfect. Why did you decide to release knowing that maybe it was challenging for people to use in full?

   15. CVCris Valenzuela15:03

       Because, uh, I think every company is different and every product iteration, uh, needs to be tuned to the audience that you're targeting. In, in our case, this is something new. This has... The idea that you can generate video or create video with nothing more than words, for example, has never been tested out before. And so you can, for years, build assumptions on how people are gonna use it and the value and opportunities on the market. But the truth is that you might be wrong, and really wrong. And if you build with that assumption for too long, you're gonna miss really getting feedback and really understanding and learning from use cases and from people who are gonna be using it. In other cases where domains and perhaps the values have been more defined, you can spend a bit more time and building in, in private and then release. I think for us, it was pretty much the realization that the only way to move forward the field and the intersection of creativity and, and art- and artificial intelligence was, was we build by really closely, as close as we can, with the community that we're building towards, which is artists and filmmakers. And we're gonna gain a lot more from being there than being too silent or too private about what we're doing.

   16. HSHarry Stebbings16:11

       Do you think it's your job to master time to value? And what I mean by that is, do you need to show people greatness within the first few seconds, or is it a longer journey?

   17. CVCris Valenzuela16:22

       Oh, absolutely. I think being able to use a model, uh, like Gen-2 in just a few seconds is critical. We've, we've released, uh, only a couple of days ago this idea you can preview frames. And so when you generate videos, sometimes you're in the dark because you don't know exactly what might come out of it. It would be great if you can preview what's going on while the model is generating. And so we offer now customers a way of previewing specific frames and then generate. And that made a huge difference, because the time to see the result was so much faster than ever before that usage went up to the roof, because people understood that how the model was working and it was, uh, it was showing you a bit more of, of the inner workings of the model itself. Um, and so it's really important to... Now, I think, a couple of years ago it was more challenging because both infrastructure and inference and speed wouldn't allow you to do it at the speed of what we need. I think now we're, we're able to do that.

   18. HSHarry Stebbings17:21

       Can I ask, I, I'm a, y- paying user of Runway. You have options, and there's kind of a paywall page, and free is an option and you can click free and go through. Many don't give you free as an option, and it's quite a hard decision on the consumer side of pay or don't enter. What, why did you decide to let free be an option versus what others did of just pay or don't use?

   19. CVCris Valenzuela17:49

       Yeah, I think it's g- goes back to the, the, the idea of, um, helping people understand where things are and understanding that this is a creative tool. And a key concept in creativity is experimentation. Like, if you ask any artist in the world, uh, how they think about their practice, they speak a lot about experimenting. They're experimenting with techniques, with ideas, with processes, with art, with combining, uh, references. So when you're working with a creative tool, you need to experiment. You need to be able to try things multiple times.And sometimes, um, um, it might take longer than you might expect. And so, the free preview for us and just allowing people we- we, for example, have unlimited plans. You can generate as many videos as you want. You just go ahead and, like, explore anything you want. Just don't worry about being charged because you can generate as many as you want with no charge. Um, that is really a creative liberation. It's a way of experimenting. And I think that's a core value of- of the product itself. You need to be... This is not... These

4. 18:45 – 23:10

   Writers Strike over AI

   1. CVCris Valenzuela18:45

      are not facts. Uh, these are not... There's no only one solution. It's an infinite set of universes that you can create. So go ahead and experiment. And I think that's a key aspect of why we- we chose to do that.

   2. HSHarry Stebbings18:55

      You mentioned experimentation being key to the workflows of artists and creators. So many are scared that AI will replace them, and we're seeing the strikes. Uh, you always say that it's an enabler. Has the media got AI wrong in terms of replacement, not enabler? And is enabler not just an intermediary step to replacement?

   3. CVCris Valenzuela19:16

      (laughs) You know, I think a lot about this. Um, I think on the one hand, language models have dominated the discourse, the public discourse, and the whole field of AI has been reduced to this idea that AI is just language models. And I think part of it has been due to the success, of course, of ChatGPT, which has been on the minds of everyone. Yet, it's really important to be nuanced. Like, in everything in the world, it's about in-betweens or gradients and it's spectrum. And AI is not just chatbots. I think the reductive view of thinking about what AI is bringing to the table and how revolutionary and- and how much change it's going to do is just not an interface where you can chat with a system. It's more than that. Uh, and- and specifically in film and video, um, we have millions of users and we have so many people using the tool to make films, to make short films, to make award-winning films. And it's such a liberating tool for them when they realize they can iterate on their ideas faster because you go back to experimentation. And so, a lot of the discourse, I feel, sometimes comes from the place of- of, um, uh, I guess, extrapolating the current state of language models and the challenges they have, which- which- which are true, to other domains, which diverge and are very different from- from- from just building chatbots. Um, and so in the- in the one hand, I think the core realization for me and really, uh, what I've been trying to emphasize more is that we need better stories. We need better narratives. We only hear one story these days, which is horror stories. Um, and I think part of it has been heavily influenced by fiction, by movies we watch, by books we read. Um, but that's one story. That's one way of seeing the world. There's other stories and other narratives that can help us really untap another set of possibilities. And so with Runway, we're untapping those possibilities of, hey, not everything is a language model. It's more and more complex than that. Let's take a look and go deeper into that.

   4. HSHarry Stebbings21:11

      I agree. Who's gonna tell that story though? The story has to come from someone who's not incentivized. So it can't be you and it can't be investors. And so, who is it then?

   5. CVCris Valenzuela21:21

      I think... I agree. It hasn't... It can be one person. Stories, the best stories, and the best way of shaping our technology is by having people use the technology and have them shape the stories. Uh-

   6. HSHarry Stebbings21:32

      Can I be bl-... I'm so sorry.

   7. CVCris Valenzuela21:34

      Yeah.

   8. HSHarry Stebbings21:34

      I'm in the UK as well.

   9. CVCris Valenzuela21:35

      Go ahead.

   10. HSHarry Stebbings21:35

       Why- why do screenwriters, like, strike if it's an enabler? What do they not see?

   11. CVCris Valenzuela21:44

       Um, I think, uh, a lot of the discourse around replacements comes from, um, this rather unsophistic, uh, view of the world that you're gonna type in, um, something, just write me a script or write me a movie, click Enter, and you get entirely the exact same script that you did a month, a year ago. And for me, that comes from a place of maybe- maybe you've never used these tools, maybe you've never tried to actually experiment with it. Um, I see a lot of, uh, articles these days which are like, "I've- I've written this first paragraph with ChatGPT, and therefore, I predict the whole set of, like, consequences, X, Y, or Z." And it's like, I mean, sure, you wrote a paragraph, but, like, writing a book and writing a script is way more than just pressing Enter. It's a process. I go back to experimentation. It's about feedback, and it's about people, and it's about using these tools in the context of more a macro project. And so, um, I think I would invite more of those, more people who are very skeptical over these tools about to- to use them and to experiment with them and see where they are and see how they provide value to your creative process and understand that this is not a zero-sum game, which I think a lot of people think about it in that way, which is you're gonna type in, "Give me a movie," click Enter, and you're gonna get a movie out of it, a perfect movie, exactly the way peop- we've been making movies for the last 100 years. And of course, it doesn't work like that. (laughs) It's more, it's much more nuanced, it's much more complicated, and it involves a lot of- of

5. 23:10 – 27:43

   Data vs Model Size in Machine Learning

   1. CVCris Valenzuela23:10

      feedback processes within.

   2. HSHarry Stebbings23:11

      I totally get you in terms of the importance of those, the level of feedback processes. Can I as-... I'm always stuck here, Chris.

   3. CVCris Valenzuela23:18

      Go ahead.

   4. HSHarry Stebbings23:18

      What's more important, data size or model size? And how do you think about, like, bluntly, the importance of size of model?

   5. CVCris Valenzuela23:28

      I think size of model matters in the sense that, um, we've seen that larger models, uh, parameter-wise are gonna get better at doing more things, uh, in multi-modalities. But at the same time, it depends. It depends on what you're trying to do. Um, and there's opportunities to s- get smaller, get models to be smaller but has more specific to the type of thing that they're trying to accomplish. And we're really seeing approaches like this in the language domain. And where we start seeing... We'll start to see this as well in the image and video domain. And so, one model that's-

   6. HSHarry Stebbings24:00

      Do- do- do you think we'll see the verticalization of models or like Richard Socher with You.com very much pronounces the importance of a single model to rule them all, which is much more horizontal?

   7. CVCris Valenzuela24:11

      Yeah, I don't think there's gonna be a single model to rule them all. That's like saying that the internet would only have one e-commerce site. Um, there's many. There... Depends. You have something like Shopify where everyone can build their own website and their own e-commerce site and you can sell anything. Um, if you think this is the tool and it's a general purpose tool, and I say tool, AI in general, there's so many opportunities to build different type of models and different ways of working with those models. And it's still very early to be so specific to say, "Oh, we're gonna only use that thing or that other thing." I think we need to be more...... more humble in recognizing that it's still new and it's still changing. And so, I will be open to have a, a diff- a different perspective on that.

   8. HSHarry Stebbings24:52

      How do you think about model lifespan? Like for you guys at Runway, do you have to continuously update models? Is it on a yearly cadence, uh, like multi-year? How do you think about that?

   9. CVCris Valenzuela25:03

      (laughs) Uh, these days it's on a weekly cadence, um (laughs) , I think, uh, I've, I found myself, uh, hearing a lot about models, uh, as a mo- as, as the moat, and moat has been something that Silicon Valley has been discussing for, for some months now. I think models are not a moat, uh, models eventually don't matter. What matters most is the people building those models, and how fast can you change and learn from those models. And so I don't think... That's why I go back, there's no one singular model that's gonna rule them all. Um, it's much-

   10. HSHarry Stebbings25:36

       What impacts, what impacts how fast you can learn and change?

   11. CVCris Valenzuela25:39

       Uh, what we w- we were discussing before, getting to hands-on with people and understanding how it works, what are their values, uh, they're getting out of it, and pushing the boundaries of, of both training, research, um, and figuring out improvements. We, uh, four or five years ago, state-of-the-art in language was, was totally different from where we're right now. Same with images. The basic assumptions of how image generation worked, uh, just a couple of years ago was completely different from where we are today. So if you find yourself trying to optimize, and engineers love to, uh, optimize things, uh, that was the wrong thing to optimize for. And so be very humble, be very clear that whatever you have right now today might change, and just ship it as fast as you can, learn as much as you can, and continue

   12. HSHarry Stebbings26:25

       Can I ask you, you know, when we think about creativity, um, someone once described hallucinations to me as students who kind of are brilliant but go off their meds, which is a form of creativity in some respects.

   13. CVCris Valenzuela26:39

       (laughs)

   14. HSHarry Stebbings26:39

       How do you think about hallucinations as a feature or a bug? Especially given the artistry and creativity that's inherent in what Runway is.

   15. CVCris Valenzuela26:48

       Yeah, um, and this is, this is where, this is a great example of being a bit more nuanced, because hallucinations in language models are different from perhaps what we might understand as hallucination in video models.

   16. HSHarry Stebbings26:58

       Huh.

   17. CVCris Valenzuela26:58

       Um, if you're asking for a fact, if you're asking a model, uh, a language model to give you the capital of Chile, and the model gets it wrong and starts rambling and hallucinating other things, you're gonna be like, "Hey, that, that's not true. Th- I wanted a fact and you gave me a creative interpretation of, of the answer." Um, but in a video domain, in an image domain, in a creative context, maybe you want to be the model, to, to have the model ramble a little bit or go off the charts, or maybe you can control the temperature to see where you wanna go, because again, it's a creative process. And so it depends. I think in some cases, uh, it's actually a feature and it will actually help you uncover more, and in some other cases you, there's no, you can't afford to have a hallucination if you're,

6. 27:43 – 30:41

   Open vs Closed Models

   1. CVCris Valenzuela27:43

      you're, uh, you're getting and you need facts.

   2. HSHarry Stebbings27:45

      I, I totally get you. Can I ask on the artistry and the artisan element, I had Yann LeCun on the show, and he spoke about how unwaveringly this will be an open ecosystem, and the benefits that will come from that. With, with artisan, uh, developers, creators, do you agree with him on the open beating closed model? And how do you think about that?

   3. CVCris Valenzuela28:06

      I think open is always better in the sense that allows you to get to a broader audience faster. Um, and you also get different perspectives because people can build on top of, of models. And we've seen this with, with some of, uh, the, the models we've released, uh, that have created some sort of like creative explosion of sorts, where people are building things that we never thought of building, and that's great. Um, I think, again, it's more nuanced once you start defining where you wanna build products for people. And so open source versus closed source is only one part of that stack. You're not gonna build an entire company just by being open source or closed source. There's way more than that. Um, and so it really depends. Some companies, I think, are gonna go into open source more and more because it helps their value, uh, prop. Some others are gonna go closed source if you wanna build other types of products. I think we're, what we're striving to be in between. We've already opened sourced some of the most, uh, I, I would say, uh, relevant models in this space that have been game-changers. And at the same time, we have more proprietary models as, as well than we might never open source.

   4. HSHarry Stebbings29:09

      Chris, what is the biggest rate-limiting factor to the Runway product today? If I had a magic wand for your product team-

   5. CVCris Valenzuela29:16

      (laughs)

   6. HSHarry Stebbings29:16

      ... for your tech team, and could take away a problem, what would it be?

   7. CVCris Valenzuela29:20

      Um, you know, uh, I think speed. I mean, and I say this because we're moving very fast, incredibly fast, but speed really matters. And so the rate limit I would say sometimes for us is, um, uh, being able to move even faster. And those, those things come in place with, um, sometimes compute. There's a limitation of compute that, um, we, the industry has at the moment. And so sometimes you're constrained because of compute, uh, sometimes constrained about, um, use cases or deploying those models and making them efficient. I think it depends, but speed as a constant thing that you're thinking about when deploying models and training models is, for us, really critical.

   8. HSHarry Stebbings30:02

      Why is it difficult to retain speed over time?

   9. CVCris Valenzuela30:06

      Um, I think what's... I mean, on the organizational side of things, as companies continue to grow, you have more nodes of communication, more, uh, uh, coordination elements. And so we're, we're a small team, we're 55 people, so we move super fast, and that's, that really matters. That's a competitive advantage. If you're a 10,000-person organization, trying to move at the speed that we're moving, it's gonna be complicated. Um, but eventually you start moving more. And that's why we've been so conscious on really having a small lean team, a very high-performance team, rather than just hiring people,

7. 30:41 – 37:00

   How Runway Built an Incredible Team

   1. CVCris Valenzuela30:41

      uh, for the sake of hiring people. Um-

   2. HSHarry Stebbings30:43

      So many, so many on your cap table told me about the performant element of your small team.

   3. CVCris Valenzuela30:48

      Yeah.

   4. HSHarry Stebbings30:48

      And I wanted to ask, what have been your biggest lessons on hiring a performance-first, speed-first, but small team?

   5. CVCris Valenzuela30:58

      Focus on, focus on, again, what we were discussing before, people that can figure things out and get things done. I think a lot of the mistakes I've done in the past on hiring is that we, we focused too much on credentials or past experiences or people who came from all these lofty universities or companies and not on the folks who were hungry and who just wanted to do things. Um-

   6. HSHarry Stebbings31:21

      How do you ch- how do you test for that in the interviews?

   7. CVCris Valenzuela31:24

      (smacks lips) Uh, we've iterated a lot. We've learned a lot. (laughs) We've learned a lot. We have a interview process now that I think is ... It's, uh, one of, one of the best, of course I'm biased to this, but it, it bi- it, it measures your proactivity and your level of ... Of course, there's a baseline level for every position and domain, but at the same time, it measures you in ability to get things done. Um, and there is a bunch of experiments and, uh, exercise we run and interviews, and then we'll get together, the whole, uh, folks, the team that was interviewing, and try to make sure that the values of what makes Runway really special are there in that person.

   8. HSHarry Stebbings32:00

      When you look at your interview process today, what would you like to know as a founder listening that you could share with them? Like, what works, what doesn't? What can someone learn from your interview process?

   9. CVCris Valenzuela32:13

      Um, doing is more than talking. Um, I think there, there are a lot of great people who are great at interviewing and are great at saying what you wanna hear, and I'm always more ... I've been cautious about that now and, and really understanding, "Okay, I, I get it, but can you do this? Can you show me how to do it? Can you do it right now?" Uh, and so focus on, focus on, on actions more than, than words.

   10. HSHarry Stebbings32:38

       What would you say are your biggest mistakes in hiring, other than credentials?

   11. CVCris Valenzuela32:42

       Credentials and people that are not, uh, I think, humble. I think humbleness is a key component of, of a company. The rate of learning as a company... Building a company is about just pushing every single day. You wanna quit, you're, you really wanna quit every single day, you need to push it again and again and again. And, and it's, it's kind of like a mad process. You need to be extremely ... have a lot of tenacity to be able to do it every time, but at the same time recognize that some things you're gonna do are gonna be wrong. And it's fine. You might spend months working on something, and the next day, you're gonna throw it through the trash. And y- it's fine. Don't worry about it. Have a very stoic approach to it and continue moving on. There's people who have very definitive worlds, views of the world, perhaps (laughs) what I was running away from before you asked me, um, don't really match and work in Runway. Uh, people that feel like that's the way, that's the only way are not gonna be able to learn as much on a field that's ever-changing. And so humbleness, for us, has been a core value. Look for people who are asking questions, who are knowing and are, um, um, have always, uh, thought of embracing change, uh, which I feel is the only constant these days.

   12. HSHarry Stebbings33:52

       I f- I find people who are boring and capable. That's my two things that I love-

   13. CVCris Valenzuela33:56

       (laughs)

   14. HSHarry Stebbings33:56

       And you know why? Because people are never as boring as they come across in an interview process. They're always more fun. You would ... Obviously, your most boring self. But boring people are rarely arrogant. (laughs)

   15. CVCris Valenzuela34:08

       Yeah, yeah. And also, I feel like w- we always tend to ... There's always, like, the norm, the medium. You can hire folks that are just, like, good but not great and excellent, and they're gonna raise the bar for everyone. And I think that's, that's really what you need to look for.

   16. HSHarry Stebbings34:22

       Dude, what do you think are other sm- our, like, complete startup BSes? You're outside of the ecosystem. What do you think are some other, like, crap advice that you see given about building companies now?

   17. CVCris Valenzuela34:34

       Um, I'm very cautious about recipes and, and, and rules and, and algorithms in, in the sense of company building. I think there's no roles. There's no algorithm. You just have to do it. You just have to do it. I think it's ... If you've never been to a city, you can read everything about that city, and you can read everything everyone else wrote about that city. If you've never been to Paris, you can read everything about Paris, every writer that has ever been to Paris. You can, uh, learn how to cook French food. You can do everything you know and, and, and find and, and understand about Paris, but nothing, nothing will beat like being in Paris for 10 minutes. You're gonna learn so much by being there than by reading anything else. And so for me, this is similar. If you wanna know about something, uh, don't read recipes, don't, uh, read blog posts, don't absorb, don't, like, try to build on the, on the ideas of someone else. Just do it yourself, um, and you will figure ... You will build a much more coherent process and idea of how things work. And so I'm very skeptical about people who are, like, or are, are writing or suggesting or advising as like, "Here's how you do things." It's like, no, no, no. That's good. That's a good sign, but just if you wanna figure out stuff, just do it yourself.

   18. HSHarry Stebbings35:46

       What recipe did you follow that did not work out? (laughs)

   19. CVCris Valenzuela35:49

       So many. Um, we don't, um ... We structure teams, uh, at Runway in thing called Ensembles that we came up with. It's a way that we ... it works for, for our team. Doesn't come from any of other more traditional scrums or, or, or ways of organizing teams. Um, at some point, we thought, oh, we need, like, OKRs and systems because some investor told me once that that's the way companies do it. We tried, and it's like, this is ... No. (laughs) Why? This is the worst thing you can do right now. It's like, you don't need this. Uh, but I could see the appeal of it because it gives you a way of making sense of the world, and it's, it's, it's, it's easy and it's ... You don't have to think too much. You just follow a process, but you're not really getting a lot of it. You're mi- perhaps it'll work for someone else, but not for you right now. And so there's so many, but team organizations and finding objectives and ways of defining those objectives, I would be skeptical about those very, very, uh, thoughtful, like, frameworks.

   20. HSHarry Stebbings36:45

       I, I, I totally agree with you there. Um, I, I do wanna ask, you mentioned learning the fundraising process earlier. It's a weird-ass world, I have to admit, especially if you're not kind of trained in Silicon Valley casting school. Um, what been some of your biggest

8. 37:00 – 49:29

   Lessons from Fundraising at Runway

   1. HSHarry Stebbings37:00

      lessons in fundraising for Runway? That was from Zane.

   2. CVCris Valenzuela37:03

      Oh, um, I can ... There's so much there. Um-You asked me about being an outsider, and I- and I felt that my first, um, years of fundraising were totally... I felt totally like an outsider. I- I- I didn't knew anyone in Silicon Valley, I didn't knew any investor in Silicon Valley. I wasn't connected to anyone either, so it was a very, like, hard start, cold start. I had to figure my... I had to figure it out. And so that's where I go back as to that mentality of just, okay, you want to do something, just do it. One of my biggest learnings on fundraising has been, I would say, uh, focus on the right investors, and ask yourself... ask questions to the investors the same way they ask questions to you. Uh, you should be much more inquisitive in the type of things they wanna build and they wanna see in the company.

   3. HSHarry Stebbings37:48

      So what- what questions do you ask, for founders listening? 'Cause I totally agree with you, no founders ever ask questions, bluntly. Not- not 100%, but like 99% don't. What questions should they ask?

   4. CVCris Valenzuela37:59

      I think what you... For us, something that we're always trying to strive and find in my investors is really making sure that we're both on the same- on the same direction. Because an investor may come to understand your company from a completely different vantage point from the one you have, and that's not gonna work. Long-term it's not gonna work. You're gonna be with these people, with this team, with that investor for years, so you really need to be aligned. So have them pitch to you, the company, and see if they're figuring out and saying the same things. Um, and that's what I've realized sometimes, that like, I had a real... an investor that, uh, gave us a term sheet, um, and we had them describe what Runway was, and we're like, "That's not what we're doing." (laughs) "Just, I love your term sheet, beautiful valuation, everything else, but- but I don't wanna... That's not gonna work. Uh, I don't think we're aligned." And so I realized that just by having that conversation and asking them questions.

   5. HSHarry Stebbings38:56

      Many would say, "I don't mind, it's the best deal," my question to you is, do you think investors really add value in

   6. NANarrator39:03

      (laughs)

   7. CVCris Valenzuela39:04

      (laughs)

   8. HSHarry Stebbings39:04

      like, no offense, isn't it all just

   9. NANarrator39:05

      Yeah.

   10. HSHarry Stebbings39:05

       ... you anyway? I'm being honest.

   11. CVCris Valenzuela39:07

       I, um, I think it's the... I think no one is gonna care about your company, about what you do, more than you. And so if you try to think that an investor's gonna change everything you do or is gonna bring something that you never thought of doing, I think you're wrong. I think an in- the investor, the role in the investor is not to build a company, it's to trust that you will build a company and that you will build, uh, a generational product. And so I think they help sometimes, they- they might help you with intros and with other things, but it's not gonna make a big difference in the sense that the success of what you're building comes from grinding and the tenacity of waking up every day and doing it for multiple days in a row, even when things are very hard. And so don't rely on, like, some magical solution or helpfulness of s- just one person, because that's not gonna- that's not how things actually work.

   12. HSHarry Stebbings39:59

       What was the hardest round to raise, Chris? You've had multiple rounds now in quite quick succession, and the company's actually, you know, close to five years old, which is kind of a interesting data point. What was the hardest round to raise?

   13. CVCris Valenzuela40:12

       I think the hardest round for us was our Series A. That was, uh, so the company was started on late 20- 2018, 2019, um, and then we raised a seed, uh, and then we've collected ... since then, every 12 months or so, we've raised a different round. We're now, we- we've, we just raised, uh, uh, a Series C. Um, our Series A was- was definitely, I would say, the most challenging, in the sense that we were pitching... Again, this is 2019, 2020, we were pitching building a generative AI company. We were, we wanted to build, and we still are, obsessed with this idea that generative models are gonna change how we think about media, entertainment, how we think about content, how we think about creativity. And so we- we went to investors and we showed them the state-of-the-art research that we had at the time for image generation, and told them, "Hey, that thing that you see here is gonna be the future of media." And if you search for an image, um, that you could generate using our model in 2020, you would be like, "Chris, come on. That's like... Of course not." We're still far away. I mean, I have so many investor emails of people telling me, "Love your team," et cetera, "but generative AI is not a thing. Stop saying it's a thing, it's not a thing." Um, and so realizing that after so many rejections, because I... Hundreds of rejections, you're, like, still mad, or like still- still, sorry, still, um, obsessed with making it work, until it works. Um, and we're

   14. HSHarry Stebbings41:40

       But did you ever doubt it?

   15. CVCris Valenzuela41:43

       ... just being constant about it.

   16. HSHarry Stebbings41:43

       Did you ever doubt it? You know, Marc Andreessen says, "There's no such thing as a bad idea, only a bad time." You know, when- when 100 people tell you, "No, it's not the right time. No, it's too early," do you doubt it?

   17. CVCris Valenzuela41:54

       I think y- of course you doubt it at some point. You're like, "Well, am- am I seeing something perhaps not clearly enough, or am I mistaken?" At the same time, I think a lot of really good investors are the ones that are not looking for patterns, that are not looking for the norm, the average, the middle of the curve. You need to be able to- to be able to understand that there's risk, and things are gonna not play out in the way that, uh, you thought every company w- would work. And so when it comes to doubting, doubting is actually good. Like, you need to be able to be like, "Wait, is everyone... So everyone is thinking one way, and I'm thinking in another way, that's really interesting, because that's different. That's not patterns. Everyone's finding the same patterns. I'm thinking about something else. If it works, it's gonna be huge. If it doesn't, I'll learn something." Um, so I think actually doubting is sometimes- sometimes good.

   18. HSHarry Stebbings42:46

       Now, I- I do agree with you in terms of also the humility to doubt. When you look back over the rounds, you mentioned kind of raising every year from the seed in 28- well, late 2018, what would you have done differently, looking back on the rounds, knowing all that you do now?

   19. CVCris Valenzuela43:01

       Something I've- I've learned a lot more over time, um, has been always emphasizing the vision of the company. Um, I think products and models...... don't really matter in the sense that what really matters in a company is the people that you have, and the collective tissue of what brings them together. And some things for me that was an obvious thing. It's like, "Hey, we're artists and researchers and engineers building a wh- completely new set of tools." But I sometimes, like, didn't have to mention that because I thought everyone knew that. But of course they didn't. (laughs) And reminding myself of like, "Hey, the product that you have today, it's gonna change. It's gonna change because these people are changing it all the time. So here's the vision and here's what we're building it." Four years ago, we were like, "Hey, media and entertainment, every company will be using these models to create everything from movies to short films. It's gonna happen." They didn't spend enough time emphasizing that, because I thought it was obvious. And then you realize, of course, of course, it's not, not that obvious. So always emphasizing and reminding yourself of, like, the vision, for me, has been a good, a good learning.

   20. HSHarry Stebbings44:02

       Can I ask you a, a tough one? And I'm sorry for it, but I am interested. You know, your last round in Series C, I- it was a high price. Um, when you think about, like, scaling into valuations, many startups they have to scale into their valuation. From a business and a money standpoint, it's a scale into. Do you think you need to scale into your valuation?

   21. CVCris Valenzuela44:25

       I don't know. It's-

   22. HSHarry Stebbings44:25

       Or do you think it's underpriced? (laughs)

   23. CVCris Valenzuela44:27

       I think it's very early, I would say, for Runway and for the industry otherwise. And, and I don't tend to overemphasize or focus on valuations, to be honest. Um, I think valuations are, if you're optimizing for a valuation, you're optimizing for the wrong thing. A valuation is just the market giving, the market giving you a price based on multiple considerations and, and, uh, and elements and variables. But if you, if you think of optimizing to grow into the valuation, you're gonna miss so many things because the incentives are put in the wrong place. And you might do things that are gonna be great for the valuation, but not great for the company long-term. And so, uh, s- always striving between, I would say, short-term or long-term vision of, okay, valuation is a great sign in terms of where the market values what we're doing, but really, w- where we are right now, it's not even, like, close to our final shape and form in 10 years. Let's work towards that because the company's gonna be worth... If we do succeed at doing that, the company's gonna be worth 100 times more than what it is right now if we try to just grow into that small valuation.

   24. HSHarry Stebbings45:26

       I, I totally get you, and I agree with that. When you think about your views on being a venture-backed startup, uh, I think this was Sunil who asked this. How have your views on what it is to be a venture-backed startup, how have they changed since the founding of the company?

   25. CVCris Valenzuela45:41

       I think for us it was, it was slightly different in, in, in the sense that there wasn't really another option, to be honest. Um, we all, the three founders are immigrants. Uh, we didn't have friends or families. We didn't have savings. There was nowhere to start building a company. And so sometimes I feel it's a privilege in being able to do internal rounds and not raise to be able to build and hire an excellent team. I think for us, it was very radical and way different in the sense that we had four months to figure out how we're gonna like, stay in a country, build a company, hire a team, build a product, do everything we need to do. And really the best way of doing that is by working alongside some of the best investors. And raising ama- raising capital and being a-

   26. HSHarry Stebbings46:22

       (laughs)

   27. CVCris Valenzuela46:23

       ... ve- a venture-backed startup was really the only way of really achieving our goals and realizing our, our vision.

   28. HSHarry Stebbings46:30

       Chris, I'm gonna get my magic wand out again.

   29. CVCris Valenzuela46:32

       Go ahead.

   30. HSHarry Stebbings46:32

       You can add anyone to your cap table, anyone in the world. It could be a, you know, Christopher Nolan. It could be Julia Roberts. It could be, I don't know, you name it. Who would you add? Anyone?
9. 49:29 – 52:47

   Quick-Fire Round

   1. HSHarry Stebbings49:29

      helps a lot, in terms of children. So yes, I agree with you totally. Uh, listen, Chris, I wanna do a quick fire round. So I'm gonna say a short statement, you give me your immediate thoughts. Does that sound okay?

   2. CVCris Valenzuela49:38

      That sounds great.

   3. HSHarry Stebbings49:39

      So what do others not know that you know to be true?

   4. CVCris Valenzuela49:43

      My co-founder, Anastasija, has this beautiful quote. Uh, "The best movies are yet to be made and the best stories are yet to be told." I think that's, that's th- the way we're heading.

   5. HSHarry Stebbings49:52

      Do all AI founders need to be in the Valley?

   6. CVCris Valenzuela49:56

      No. I'm in New York and I've been into, in the South, like, just a couple of times.

   7. HSHarry Stebbings50:01

      I mean, I was slightly teeing myself up with that one.

   8. CVCris Valenzuela50:04

      (laughs)

   9. HSHarry Stebbings50:04

      Uh, what, what do people not understand about Runway that you wish they did?

   10. CVCris Valenzuela50:09

       It's very early. We're a five-year-old company, but I think we're, we're just a baby.

   11. HSHarry Stebbings50:14

       What single element would you most like to change about the AI community?

   12. CVCris Valenzuela50:19

       Stories. We need better stories. We need more nuanced stories. This is not about horror stories or fairy tales. It's about humans. And so, let's focus back on people.

   13. HSHarry Stebbings50:29

       What is the strongest belief you had which turned out to be wrong?

   14. CVCris Valenzuela50:34

       Fancy UIs don't matter. It doesn't matter if you think about building beautiful UIs. What matters is using these models to just do great things. Everything else will follow that.

   15. HSHarry Stebbings50:44

       What's the most painful lesson you've learned that you're pleased to have learned?

   16. CVCris Valenzuela50:48

       Uh, building anything is painful. (laughs) It's, uh, it's, it's hard and you wanna quit. Just don't.

   17. HSHarry Stebbings50:56

       Does it get easier over time?

   18. CVCris Valenzuela50:59

       You get... It doesn't get easier. You get more used to the pain.

   19. HSHarry Stebbings51:06

       How so?

   20. CVCris Valenzuela51:07

       You start building a skin. You understand, you tru- y- you build an intuition. And things, as you grow and scale, are gonna be different. You're gonna have different challenges to scale, but challenge is gonna be different. Uh, but if you keep a stoic mindset and just understand that there are things you can control and things you can't control, you can do great stuff.

   21. HSHarry Stebbings51:29

       You can have dinner with anyone, dead or alive. Who do you have dinner with and why?

   22. CVCris Valenzuela51:33

       Uh, my m- my mom. I haven't seen her in a while.

   23. HSHarry Stebbings51:37

       I love that answer. Uh, most people go for, like, LeBron James, so that's a lot more fulfilling. Uh, tell me, what role does AI play in society in 10 years? You said we can't comprehend it, we don't even know. How do you f- project, forward plan AI's role in society?

   24. CVCris Valenzuela51:54

       We're gonna stop referring to it as AI. We're just gonna think about it as tools. I think I'm really looking forward to the moment in time where there's no discussion around AI in the same way that today we don't discuss the internet as, as a thing. It's just so ingrained in everything we do. When you go to the bank and you do a wire transfer, you're not saying, "I'm gonna log into the internet to do a wire transfer." You just do it. You know that's the way you do it. And I think we're heading to a world where everything we do is gonna be built and powered by models. You just d- you, you forget about it and you just continue with your life.

   25. HSHarry Stebbings52:27

       Final one. 10 years' time, 2033, where's Runway then?

   26. CVCris Valenzuela52:33

       Making movies, the best movies.

   27. HSHarry Stebbings52:36

       (laughs) Chris, as I said, I did 17 reference schools on this.

   28. CVCris Valenzuela52:39

       (laughs)

   29. HSHarry Stebbings52:39

       It's been such a pleasure to do this with you, so thank you so much for joining me.

   30. CVCris Valenzuela52:43

       Thank you so much for having me here. It was a great conversation.

Episode duration: 52:47