Humans Tell The Best Stories

In this episode, Descript CEO Laura Burkhauser talks about what it means to build a human-forward AI video company in a world flooded with synthetic content. Laura came to Descript by an unlikely path — German literature, then product management at companies like Salesforce — and became CEO because she believed in what Descript was building: tools that keep the story in human hands while AI handles everything underneath.

Hear about why Descript is investing heavily in AI but not aiming to be the best at AI avatars or fully generated content. How Laura thinks about the 'simplifier' role of a CEO. What she means by three layers of AI video. How business teams have become Descript's fastest-growing audience. And why getting your team AI-native isn't something you can mandate — you have to engineer the conditions for each person's own conversion moment. Finally, Laura also gives her candid take on the Beta version of the Descript API: they don't have a killer use case yet, and they're actively inviting partners to help find one.

Laura Burkhauser on LinkedIn
Descript

Transcript

Why Human Stories Still Win

Amy Jo Kim 0:18

Laura Burkhaser is the CEO of Descript, an AI video company. At Descript, they're betting that stories told by humans will matter even more in a world flooded with AI content.

Laura Burkhauser 0:29

We like AI, but we also really like humans. We think humans are better storytellers than AI. You can tell your story, and we can make it look amazing.

German Literature And The Wrong Train

Amy Jo Kim 0:37

I recently sat down with Laura to hear how she went from enthusiastic user to CEO and how she's using AI to empower people, not replace them. You've had an amazing journey. You started in German literature. What was that about studying German literature?

Laura Burkhauser 0:53

I've never been someone who has a grand strategy for her life. When I look back at how I've made decisions about what to do, it's been about following my interests in a really authentic way. And sometimes that's really paid off. And sometimes it hasn't immediately paid off, but I think it it's hard to regret. So when I was in college, I went to a liberal arts school and um University of Chicago, there's a core curriculum. So you basically don't have any choice over what you study your first year in college. And then in my second year, I took an elective where I read some plays by Heinrich von Kleist. And I just loved them. And I loved the professor and I loved the discussion. And I thought, well, this is fun, and I'm pretty good at German. I don't know, maybe I should be a German literature major. And I then started reading more German literature and I found Franz Kafka and thought that I was gonna go get a PhD in German literature. I got a Fulbright, I went to Berlin, I think the best city in the world. And I learned that there was just no way on earth that I was gonna get a PhD in German literature, that I just really did not want to do that at all. And I'm really grateful for that year because I hadn't applied and gotten into this seven-year program where I'd write a huge thesis. I learned very quickly, nope, that's wrong. If the first part of the way that I've lived my life is to follow my heart, the second kind of core piece of advice is when you realize you're on the wrong train, get off on the next stop. That's what I did then. I got off of that train and uh started on a new journey.

Consulting To Product Management

Amy Jo Kim 2:28

And then you got an MBA.

Leveling Up At Twitter

Laura Burkhauser 2:30

That's right. I went and I did what a bunch of people who don't know what they're doing with their lives do, and I became a consultant. Consultant is like a great job for a young person who just doesn't actually know how to do anything, but is pretty smart because you learn a lot about strategy, you learn a lot about business. And the third year that I spent in consulting, I spent doing something weird, which was building an online learning platform. My firm was a little bit ahead of the curve because this was now over 10 years ago. And the idea was we want to build a long-term top of funnel for prospects that may grow over time. So we need a lower cost option, way to reach these growing firms, like having content that these firms can read and learn about us from. So we should have this online learning portal. I had just moved to the Bay Area and was like, why don't I build this? I'm reading about lean startups. I can build a Squarespace, I can make all the YouTubes, and you can actually still find some of the early YouTube videos that I made. I had this experience where I was building this thing online and I realized, oh my God, this is what I want to do with my life. Like whatever this job is, where I'm like, there's a problem, and I'm building something to solve that problem, and I'm launching it quickly, and I'm getting feedback, and some people hate it, and what they hate about it is this, and I've got to go change that. Now, uh in the age of AI, that job would probably just be called founder because I would have all the skills that I would need to get to MVP all by myself. But 10 years ago, that job was called product management. That was the kind of discipline that you would put someone who's a smart generalist with a consulting background and like a creative streak, but no like hard technical skills when you become a product manager. And so that's what I did.

Amy Jo Kim 4:05

Wow. And then you had stints at Le Tote, Rent the Runway, Amazon, and Twitter. You started in fashion. Was that just happenstance? Or are you into fashion? Were you curious about it? Did you just have contacts?

Podcasting Leads To Descript

Laura Burkhauser 4:21

I'm glad you like it because it's a very similar story for how I got the job at Descript. So after Le Tote, I went and I worked at Rent the Runway. They were winning the category at the time. And I worked with them for a year and really loved that company. But I found that I just wasn't as successful in that role as I had been at Le Tote. And I didn't know why. Now I know why, right? Like later in your career, you can look back and be like, aha, that is the problem that I was having. But at the time I just felt frustrated. Like, I know I'm really good at this job. Why don't I feel successful as an org leader at this company? And I finally got to the conclusion that while I didn't know exactly what was going wrong, the reason why I couldn't get better fast enough was that I needed to see how other people were doing things. Whether I like it or not, I got to go someplace bigger. I got to learn from other people. I got to see how they do this. And so I went to Twitter and I got to see how all the other directors of product did stuff at Twitter. And I learned so much from them. That's how I learned how to be a product leader, not just a product manager. But then after I left Twitter, I didn't really know what I wanted to do. And I definitely knew that my passion was not in social media anymore. I was interviewing at some social media companies, and I just felt like all of the interesting stuff in social media had been figured out. We know what this is. It's a feed, it's got a like button, maybe some other reaction buttons, it's got some comments. You try to get people to spend as much time looking at the feed as possible. Like that's the game in social. And I was just not enjoying playing it anymore. That's a real soul-searching moment when you're like, okay, I work all day on a product that I will not let my children near. So I knew I didn't want to do that, but I didn't know what I wanted to do. So I did what a lot of people do when they don't know what they want to do in their lives in San Francisco and started a podcast. And when I was editing that podcast, that's how I found Descript. And so it's the same story over and over again that I did with Le Tote, where I love using this product. Descript was really a transformative experience. It made me feel creative. And I loved that feeling of creativity. It reminded me of when I was making videos back in the day for the online learning platform and thinking like, oh my God, this is so much better to use this tool. I wonder if they're hiring product people. I should just knock on their door and see if they'll have me. And they did.

CEO Work: Simplify And Repeat

Amy Jo Kim 6:36

So you mentioned you learned a ton from the folks at Twitter. What are some of those lessons that stayed with you that you're bringing to this new role?

Laura Burkhauser 6:46

What a great question. I think that one of them is just how much of the job is being a simplifier and being a good communicator. So one of the gifts of being senior is that you are able to appreciate things at an altitude that when you are two in the weeds, you just can't see the altitude because you are correctly focused on the three steps right in front of you and nailing them, right? But it keeps you from like kind of being up here and seeing the forest for the trees and understanding, like, oh, this is why this stuff is important. And so much of my job is looking around at all of my excellent people who are in the weeds, figuring things out and being a simplifier, pulling them out of those weeds every once in a while to just make things go faster, help them make some good decisions, and then jump back out. That's a lot of what I do day to day. I'm like on Slack monitoring, seeing when people are going down a rat hole and being like, hey, let me be a simplifier. What if we do with this? What if we do this? I perceive that as being quite helpful. You can ask my reports if they feel equally grateful for my interventions, but I think so. I think I that's a huge part of what you do. And then a lot of what you do is you make sure that everyone knows the story, right? This is the story of why we're doing things. Let me go tell this story. One of my mentors, a la tote, told me, like, the more senior you get, the more your job is walking into different rooms and saying the same three things all over again. And I do think that is a lot of the job.

Amy Jo Kim 8:16

So what is the hardest thing for you not to do?

Laura Burkhauser 8:20

Be a product manager. The hardest thing for me not to do is to say, you know what? I've got this one. I'm just gonna write this back. And now it's also like, I was gonna open cursor, I'm just gonna build it, right? And recognizing that although I find that extremely fun and fulfilling, it is low leverage for me to do that.

Three Layers Of AI Video Editing

Amy Jo Kim 8:42

That sounds like one of the hardest shifts of product leadership as your company scales. Like if you're a three-person team and you're starting something new, yeah, maybe you're building. So one of the things that drew us to DScript, and we love D Script for many years, is early smart practical use of AI. So now you're the CEO, AI is galloping ahead. We're all struggling to catch up. And you've got two things going on. You've got a product where you've got a lot of incremental change, and you're integrating AI, developing what's already there, all of that incremental stuff to improve on all the workflows. But then you're also doing new big swings, like Underlord. So as a CEO who's really leveling up into holding and communicating the strategy, right? How do you think about balancing those two things?

Laura Burkhauser 9:41

Well, I think you absolutely need both. The way that I think about uh the AI video editing space is you have kind of three layers. You have the generation layer, you have the application layer, and you have the orchestration layer. And you need to have a strategy across all those layers. So by the generation layer, these are the models. The problem we're trying to solve for customers is like it is very hard to look into a camera and say what's on your mind and make video. But that is the primary use case that we want to serve. We do value real humans telling real human stories and trying to make it as easy as possible to go on camera and tell your story. And yes, we have avatars, yes, we have AI voices, but that's not where we want to invest and win and have the absolute best experience. Because we think that the absolute best experience usually has a human messenger. And what we want to build for is how do we make human messages as easy to deliver as possible while keeping that production value high. And so our generation layer, all the models we're building are models that are all about taking whatever blather I'm blathering into this camera right now, and in post, making it look like I was poised and succinct to begin with. So, like a lot of the generative models that we're making are things like, okay, we will hide all of your jump cuts. So you can't tell that you just cut out six likes in that paragraph of content that you just delivered. We will fix your eye contact, we will fix your lighting, we'll do all of the things that make you feel like you're always ready for camera and like the thing that you said was brilliant. We do a lot of really good lip dubbing so that if I accidentally said the wrong word, you can put in the right word and it just looks like I said the right word all along. All of these things that make recorded media just feel really easy. So you don't need to do 17 takes, you don't need to buy $10,000 worth of lighting equipment. You can just look into your camera, say your story, which is the thing that you own, the thing that is uniquely yours. You can tell your story and we can make it look amazing. So that's where we're investing on the generation layer. On the orchestration layer, I think more and more, especially at scale, a lot of the work that people do to edit video, to clip video, to repurpose video, to translate video, to remix video will be done not within an application, but be via like an integration done with an API, the CLI, an MCP, in an abstracted layer where you're doing your work that's connected with a bunch of other stuff. And DScript wants to make it very, very easy for you to work and create video with knowledge of all of your other apps. And we use Underlord as a background agent to put that video together for you. That's really helpful at scale. But even then, we have a strong hypothesis that for a long time, the last 10 to 20 to sometimes even 30%, depending on how much artistic flair you have or what a perfectionist you are, you're going to need to do yourself in the application layer. Now, if it's transactional content like making clips, maybe all you're doing in the application layer is a quick review, and then you just say, yep, approve, send it. But something that feels higher production value or that you want to put more of yourself in, you may be going in and like changing a lot of things, right? It's just that we do all of the imports, we do the mise en plus, we do the rough cuts all via MCP, but then you come in and you really add your personal editing touch. You make sure that your story is the one that you want to tell. We believe that what you need is a dead simple application layer. So Descript wants to have interesting offerings across all three of those to help companies and people tell their stories.

Amy Jo Kim 13:31

So you recently opened up your API to do exactly this vision, yes?

API Beta And Video At Scale

Laura Burkhauser 13:36

We're in alpha slash beta.

Amy Jo Kim 13:39

So you're clearly someone who balances the realities of business with deep empathy for your customers. Tell me, how has your audience changed over the past say year?

Laura Burkhauser 13:56

20 years ago, YouTube launched, and there was a very small number of YouTubers. And then by 10 years ago, we had this emerging creator population. And so when Descript was founded, the first people who really used us and loved us were creators, podcasters, and YouTubers. But because what we're incredibly good at is if you can edit a doc, you can edit a video, right? We work with people who are more comfortable in docs and decks. And so who is that really attractive for? Sure, yes, creators and podcasters, but also a lot of people make video for work. And so the other thing that's really happened over the last 10 years is that every business now knows that they need to make video. Most companies have a podcast, most companies have a YouTube channel. If you look at Wistia's state of video, you'll see that like companies are making three times more video in 2025 than they were in 2024. In 2026, it's just going to keep growing. And so our customers have more and more turned from being creators to being small businesses, all the way up into big businesses using descript for marketing. And then podcasters and media teams, all the way from like, I have my Bachelorette podcast to like I am the New York Times making all of my podcasts and all of my video content. And so what we say is we serve marketers and media teams, whether that team is a one person team or a 100% team. That's the main way that our customer has changed. The other thing that is happening in video is we're all looking for ways to make more video faster. Democratization, which is what Descript does, is a huge way that you can make more video faster because you can empower more people on your team to make videos because you don't need to be a video editor to make a good YouTube video anymore. The other way that we are helping teams make more video faster is through this orchestration work, making it easier to automate the pieces that aren't creative. It's like get all of your files in, label them correctly, put them in folders, put them all into a composition, start a rough cut. All of that stuff you can now do with the API. So those are the two ways we help. There is also this third macro tenant trend of generative media, right? What if you can make media without getting someone on camera? Right now, I think that saves the most time in advertising. And so we're really seeing it there. I don't yet see it in like long-form storytelling yet, because the workflows feel really bad right now. And that's something that we're working on is can Descript have an application layer that can make creating generative content feel really simple? I do think that, like from a values perspective, Descript does have a preference for recorded media. When possible, it's great to have a human telling a story. If you're going to use generative media, use it because it's the best medium, not because it's easier or you're trying to cheap out. Because I think in a lot of cases, having a real human deliver a message and tell a story is just going to be more powerful. And is this more, I don't know, is more the future that we want to build for.

Amy Jo Kim 17:01

There's definitely two poles in our industry with AI, one of which is hire this AI person and fire your team. The other end is build tools that amplify humans. And the script is firmly in the build tools and amplify humans camp. We like humans.

Laura Burkhauser 17:21

We like AI, but we also really like humans. We think humans are better storytellers than AI. But man, AI is so fast at doing discrete actions. And like we really want to build a platform that leverages the magic of that so that we can tell more stories.

Amy Jo Kim 17:37

So, Laura, there's a big push in the industry that I'm sure you're seeing and feeling, which is everyone needs to be AI native. All teams need to be using AI or out the door, et cetera, et cetera. How are you managing making sure your teams stay up to date?

Laura Burkhauser 17:54

Every person needs to have their like Paul on the road to Damascus moment with AI, their conversion story. If the primary way that you use AI right now is to chat with a chat bot, you haven't had your moment yet. When Sonnet 3.7 came out, I tried Cursor again. I had tried it like the November before, and it was promising, but it was just like I was like, I still can't use this. And then in March after Sonnet 3.7 came out, I one-shotted a small feature that we actually released production. That was my moment, right? So the first thing that I wanted to do was get everyone else that moment. Now, for product and design, it's very easy to know what that moment is. It's just like you've got to download a script onto your local machine and you've got a vibe code a feature. Even if that's not the primary way that you're going to end up using AI, that is the thing that's gonna make you have the like, oh my God, this is real moment. Once someone has that moment, they're believers and they're highly incentivized to go out and figure the rest out themselves. And so the first job that you need to do is just make sure they have that moment. A hackathon's a good way to do it. By the time we had our hackathon last June, everyone on EPD was there. And there started to be pockets in all of the other functions of like someone who made something interesting enough that they were like, oh, okay, yeah, all right, I could actually automate like the way that right now I'm keeping track of all of the videos that we publish. I could just automate the creation of that database and I don't have to do the data entry anymore. But it's like you have to work with people to find out like, what is the use case? What is the problem they have that they could build something with AI for? Because they don't have the imagination to know that. Like people are experts on their problems. But because they don't have necessarily experience building tech solutions to those problems, they often need a product partner to help them be like, okay, so what if we actually create a performance review data collector that serves managers all of like the various things that they may want to look at when creating someone's performance review? And we just vibe quote a website that lets them do that. Or like, what about with recruiting? What if you could have a bot sit in on Zoom, interview the hiring manager about what they're looking for, and based on that hour-long interview, create the job description, the interview loop, all of the questions that you ask, the scoring rubric for those questions, and then serve them up to the hiring manager to go and review, right? It's things like that where it's like the recruiter is an expert in their existing workflow and in the problems with that workflow. But they maybe aren't going to be the person that's like, oh yeah, well, first you should you should have like a zoom intake, uh, whatever. What you need to do is get everyone their first experience of solving a big problem with AI and seeing that it's possible. And once they have that big experience, I think you're done. And then I think like you can monitor, but not goal. I monitor like the extent to which our engineers are using AI. And I do that in a few ways. I look at what percent of PRs are AI assisted, I look at cost, like how many credits are developers using every month. And that kind of is is good to let me know the overall health, but I would never set a goal. Because as soon as you turn like cost into a goal, it's super gameable and it's like just dumb. Instead, I just try to show people the truth and the truth will set them free.

Multi Agent Orchestration Reality Check

Amy Jo Kim 21:26

That's a great tactic. You talked about layers, we talked about orchestration. Right now, it's all about orchestrating multi-agent systems. I'm wondering what you could tell us about just how you're thinking about that at Descript and how that's playing out for how you help people solve users' problems.

Laura Burkhauser 21:48

We should talk about this again in a few months because, like right now, it's all theoretical. I love working with people, truly understanding their workflows and building. Building products that integrate deeply into their workflows to solve problems and create productivity. I feel less comfortable when I am sort of imagining use cases and saying, like, wouldn't it be great to be able to do this with your existing media library? I feel a little bit more like, what is it, a hammer in search of a nail? Oh, we have this background agent. Let me go like suggest a bunch of use cases to you. I feel like I am in the early phases of understanding some of the most powerful ways that people are using the background agent. The whole point of the beta is it should be work that you are doing very deeply with customers who are using your API in this case, really understanding what their interest is and how much you can help them and crafting the product to truly solve that problem. And then you get to go out in front of your mission accomplished banner and be like, let me tell you, Amy Joe, this is how DScript is solving all these problems for customers. Right now we have an alpha, we have a beta, but I'm not sure we have a killer use case yet. I think that's what we're looking for. And so if you out there think that you have a killer use case for video at scale, I have a few theories, but I I want to work closely with you through this beta profit.

Amy Jo Kim 23:12

Exciting, exciting. All right, have a great day, everyone. And I think we all got pretty darn inspired.

Laura Burkhauser 23:21

Thanks, everyone.

Closing

Announcer 23:22

Thanks for listening to Getting to Alpha with Amy Jo Kim, the shows that help you innovate faster and smarter. Be sure to check out our website, getting2alpha.com. That's getting number two alpha.com for more great resources and podcast episodes.