laura-edited-audio
===

[00:00:00] All right, Laura, welcome to the show. Thanks for coming on. Hey, Jeff. So happy to be here. Thanks for having me. I am super stoked about this, 'cause we have been using Descript for, I think, pretty much the entire life of all of our podcasts for all the video stuff. I love to hear that, and I always tell people I was a Descript customer well before I was the CEO, and kind of had the same experience.

Loved the product so much that I knocked on Andrew's door and asked if I could work on it with him. You've had a rapidly ascending career. You've grown really quickly and progressed really fast, and I think that's something that people see across board. Maybe, you know, speeds are always different across people, but we all run into similar problems at some point, where you kinda hit a point where you need mentorship, you need community.

So I wanna dig into that, and then also, across the country as we've been hosting events and meeting with product leaders, the big thing that hits for leaders is AI adoption and how [00:01:00] you are moving your teams forward with this, and I think you have great frameworks there. Before Descript, how did you get to where you are?

What does that look like? It's nice to hear one's career described as a rapid ascent. It's a nice way to see things 20 years into a career. I think when you're in the middle of it, it certainly doesn't feel like you're on a straight path, a rapid ascent. It's been serpentine, I think, for me. And I think that the first really big thing that I had to learn is just how do I wanna spend my days?

What excites me? What's my impact equation? And a lot of my 20s was figuring that out. And then really as I got into my 30s, it became, okay, here I am. I'm definitely a product manager. That's the thing that I wanna do. I wanna build products on the internet. And then it became about, what's the right company for me?

How do I succeed? How do I get better in my craft? How do I get better in my leadership? So 20s was just sort of like, "What the hell am I doing with my life?" 30s is like, how do I build a career in this thing that I've decided kind of works for me? And we'll see what the next chapter holds for me. I don't [00:02:00] know.

I think almost everything that looks like a rapid ascent or overnight success, what... I forget the soccer player who said it. One of the Premier League kinda level people put it, yeah, it took decades of hard work and preparation to be an overnight success. But in retrospect, we can give you the props you deserve here.

I had a similar experience where you kinda move quickly, and you find yourself in a role that maybe there's not the support system or, or something different about it where maybe you don't perform as well as you want. Yeah. I had that in a past role where I hit one of my first director roles. It was like, yeah, I know, you know, y- as a kid you're like, "I know everything."

Yeah. And then you get in, and you're like, "It's not going well. Oh, no." And I think we share a similar experience here. Yeah, so when I got out of business school, first of all, I was like, do I wanna work in big tech or not? I decided not. I really liked the way that working at a startup felt. And so I went and I worked at this startup, again, where I was a customer.

It was called Le Tote. Mm-hmm. And they did subscription fashion. And I did three years there and really learned the hell out of the subscription fashion business and learned how to be a really great [00:03:00] product manager. And because it was growing quickly, I then became, like, a, a people manager. And I was a pretty great product manager pretty much out of the box, just kind of self-taught.

And I was a pretty good people manager, just straight right out of the box, self-taught. A lot of being a good people manager is just authentically caring about your team and also being good at your craft. And so, like, that went really well. Then I went and I started working at Rent the Runway, and I thought that this was gonna be my dream job 'cause it was like, okay, this is still fashion subscription, this thing that I was really loving, and I got this director title, which was super shiny to me at the time.

I was like, "Oh God, I'm like product director at, like, this awesome company, just a couple years out of B school, just a couple years into, like, really doing the product manager job. Like, I'm killing it." And I went there, and I had one of those moments. There's, like, a book called What Got You Here Won't Get You There, and that was my what got you here won't get you there moment.

I'm like, "I'm an awesome product manager. I'm a really good people manager. Why [00:04:00] am I not a good product director, and why am I not getting better faster?" And the answer to the second question is I had done this thing that I think can happen to you if your career is moving really quickly, which is that I very quickly became one of the most senior product people at Rent the Runway without a lot of peers in the product world to show me.

I was reporting into a CTO, amazing guy, but came up out of engineering. So I'm now the eternal product level at the company at the time, so I have no one to really learn the craft of being a product director from. And I'm also at an East Coast tech company, we're in the water. There's not kind of this all-around knowledge in the company about what product does at higher and higher levels.

And so there was no one to kind of tell me, like, this is what a product director does. This is how you think about resource allocation. This is how you think about influencing the business strategy. This is how you start thinking about working on the business and not just working in the business. This is how you work with executives.

Like, all of these things that are very different than, like, [00:05:00] here's how you think of a roadmap, own that roadmap, and execute the hell out of it, and train junior PMs to do the job, right? It's just, like, a fundamentally different job, and I didn't have the support in company that I needed And I think if I were talking to Laura at that time and trying to advise her, I'd say like, "It's time to look outward.

Your peers aren't gonna be here. You need to go find them elsewhere if you're gonna keep learning at the rate that you wanna learn." I didn't do that, and so instead what I did is I realized I gotta go. I have to go. I've been avoiding Big Tech my whole life, but if I wanna actually learn how to do the product director role, I need to go to Big Tech where there are a bunch of different product directors.

I need to see what excellence looks like in action, and then I can kind of figure out what this looks like for me. Well, okay, now fast-forward, this is how one must always like, uh, don't make the same mistake twice. I found myself last August in this role where I'm literally the terminal position at the company and can't kind of look around and be like, "Am I doing a good [00:06:00] job?"

And so it's been critical for me in this moment to reach out and join a lot of communities to find peers, and not just any peers, but to find peers who inspire me, to find mentors and heroes and peers that I can kind of learn from and shape myself around, either because we have a direct relationship or because they're very good at being a thought leader and I resonate with what they write and I can kind of absorb what they're talking about.

And that is hard to do in a world of lunacy in the CEO influencer space I also think this is why you have so many CEO lunatics. It's like an epidemic because it's a lonely job, so you're like, "Okay, I gotta look outward. I gotta, like, find out how to do this job well." But if all you're doing is going on Twitter/X or LinkedIn and reading your feed to learn how to be a founder CEO, God help you, because there is not a position extreme enough that I could take satirically that hasn't been expressed earnestly online.

We always joke about the theater of AI happening on LinkedIn and maybe some other influencers who have other platforms, but [00:07:00] I think the metaphor applies to almost anything where any kind of algorithmically driven content distribution is never going to skew towards moderate, really intelligent, nuanced arguments.

It's always going to be, "Well, I got rid of 40 marketers with this skill," or, "Here's why this industry is gonna vaporize in the next three and a half minutes." Yeah. But, like, the pace has to keep going, man. You know, it used to be a year, but that's too slow. It's three and a half minutes now is... That's it.

That's the cycle. I saw, like, career-wise coming up, I moved very quickly through several roles, and then I feel like director is one of those roles. You're gonna trip on director if you're not careful. That's where things flip, and that book is the perfect one to reference there. When you're at a certain seniority level, you don't have peers in your company.

Yeah. You can't. We do a big product leader dinner set around the country. It's all invite only. We go through and we deeply vet everyone. It's been really popular because people come back and say, "Nowhere else can you go where it's my peers." And, and the big thing we pitch when we're there is, [00:08:00] "Your boss is in here.

We're talking about AI adoption and how people are fried and can't keep up. You're not gonna go tell your CEO this, right? Like, you're gonna get fired. Don't, and don't, you know- Yeah, you better not. You tell me you love AI. You tell me you go home and you build AI apps for your kids. That's what you dream about when you get home, and you understand everything.

You've read every post by every influencer in the world Yeah, there's really an emotional component to that too. I think, like, that's one of the things that's tough is, like, for me, one of the reasons I loved being ... This is very much a what got you here won't get you there. One of the reasons I loved being in ICPM, like, maybe my favorite job that I've ever had, is I never played sports in high school, really more of a theater kid, if you can't tell.

But when I was working with engineers and designers and we were shipping things, I'm like, "This must be how it feels to play sports," where you're just, like ... You're a team, you're all good at different things, you're leaving it all in the field, you're working hard every day, you're shipping together, you're winning together, you're pouring Gatorade on each other.

Like, it just, like, feels [00:09:00] good, the sense of belonging and community that happens when you build with other people. And as a coach, you feel, like, pride, you know? But it's not the same. Like, you weren't on the field, and you're not in the community in the same way. And so I think there's, like, an emotional growing up that needs to happen as well.

So that was tough for me. I think it's an important conversation to have publicly, 'cause this is one of those areas ... And maybe it's changing a little bit, but I feel like this is one of those things people will talk about behind closed doors. They'll say it privately, "Hey, you know what would help me is this," but it's rare to have someone kinda come out and just say this.

But you need this, right? You don't need to succeed on your own. It's gonna be really, really hard. You should have people around you, just when you're really, really junior, those people around you can be at your company. The ring needs to grow as well, I guess, or the circle needs to grow. Yeah. As someone who kinda comes from a product background in a chief executive role, I have to imagine the pressure is there of, like, you need to adopt these tools.

You know, the magic of AI, whether we can joke about the theater of AI on LinkedIn and, and all the BS that surrounds it and everything like that, and it is, it is a lot of BS, but it is a real tool. It's gonna have real benefits. It already has for a lot of [00:10:00] companies. How have you looked at that end of it from a reality standpoint?

You've got the weapons now to do this. You've networked with peers. You've grown as a leader, but you still have to go and bring in the benefits that were being promised from, you know, all these frontier models and capabilities. But you can't take forever to do it, even though some of these things are probably a little bit longer term to understand.

Like, how have you looked at this problem internally? Well, the first thing I'd say is I- Love AI. I love building with AI. And that's important because, like you said, it feels hard to have a nuanced opinion about AI right now because first of all, no one's gonna watch this episode if I, you know, I'm just like, "It's good for some things and not others.

Sorry." Unfortunately, I have a nuanced opinion about AI that is neither stop AI nor I'm gonna fire everyone in my company, and we're just gonna have self-learning agents running every single function. I don't believe in either of those things. Wow, what a surprise. But I love AI, and I've loved AI. I mean, like, Descript has been an AI [00:11:00] product since infancy, right?

The original idea of Descript is that you can edit video just like you edit a doc. And so that means, like, you're editing text to edit the video. It also means it's collaborative. Everyone can come in and edit at the same time. And so when LLMs came out, first of all, we kind of got multimodal editing for free because if you think about video as a document, what are LLMs incredibly good at doing?

They're really good at editing text, and guess what? When you edit text, you edit the video. So that was, like, our next big breakthrough with buttons like, you know, edit for clarity or removing retakes, really, like, gets you a lot of wonderful editing buttons that feel really deterministic to customers for free, provided by LLMs.

And then we immediately started working on Underlord, your AI editing companion slash Underlord, you know, servant. And I laugh about servant, but I think, like, when we created Underlord, we created a Descript, like, I created in the product team a couple of, like, grounding beliefs about AI and the [00:12:00] way that we should work with it, um, that I think have aged like wine, if I say so myself.

And, I mean, we'll see, TBD, but so far in a world of a lot of change. And the first is that with my big headline that I would put on a sign, "Struggle with your art, not with your tools." Mm-hmm. The purpose of AI should be to take the struggle out of the tools that you use and not out of you thinking about what the hell you wanna say, thinking about what your vision is visually for the thing that you're creating, finding your voice, your distinct voice, which is a thing that takes every human, most humans, except for, like, wunderkind writers.

It takes most humans decades to do. If you ever find it, if you're lucky enough to have a unique voice, it'll probably take you 30 years to find out what it is. And so all of these things, this is your art, and guess what? You're gonna struggle with it. The act of creation is a difficult process. But when you get it, the [00:13:00] act of cutting the tape in your timeline should not be the thing that's causing you to curse late into the evening, right?

And so thinking about AI as a way of easing the friction in tool use and not of replacing human creativity is one of the ways that we kind of grounded Underlord from the very beginning. Another way that I think we've been pretty disciplined from the beginning is we're excited about the idea of chatting with an agent, and I actually believe that voice is the future here.

Mm. And that you're gonna wanna direct Underlord to just do things for you. Some of those things will be high level, like, "Can you make this look a little bit creepier?" And some of them will be low level, like after you do that initial, "Can this look creepier?" Like, "Hey, no, you need to turn the green sub light up a little bit.

Can you do that?" Some of these things you're gonna wanna express through text But that humans are still going to need a lot of knobs and dials. You're gonna wanna put your hands in the clay to get things to be perfect. And that's where it's like, yes, you wanna make it [00:14:00] easier to work with your tools, but this obsession we have with one-shotting things, I think is not our vision.

I don't think that you should want to one-shot video editing, right? It's an iterative process where part of what you're building then teaches you what your vision actually is, and iterating is actually critical to finding that voice and that aesthetic. And so I wanna make it really easy to iterate, but, like, I don't think you should just one shot in post, like, probably.

Right. Part of the point is that you're making something that reflects you, and if it's just a one shot through Underlord or whatever, whoever, not even Descript necessarily, everyone's thing kinda starts to be the same. Yeah. If it's all based on the same average. And sure, maybe we can give it a ton of context first, but, like, you kind of have the same seed stuff, and that's where a lot of the stuff can enhance.

It can help you prioritize, it can help you parse, it can help you... The cutting, the nuance. You wanna cut all the ands and uhs. At some point, though, it does go too far. I think that's a really good, you know, back to nuanced opinion. The art is human intrinsically. [00:15:00] Yeah. And you can't remove yourself all the way, or else, like, no one's gonna listen.

AI is definitionally a derivative technology, right? That's how it works, is by taking a mean of things that everyone... Like, a statistical average of what is most likely to happen next, which is, like, the nature. Like, that is what derivative means literally. And, and so without a lot of input from you, without a vision, without a director who matters, you're gonna get derivative content.

Now, maybe derivative content is okay for some use cases, which is why I think a lot of media is going to get swallowed up by one-shotting, because, like, a lot of media kind of, like, is slop by nature. But all the media that matters, I think, yep, we'll use different tools. Yep, creating good stuff is still gonna be hard as hell, but for different reasons.

And hopefully it makes it easier for more people. Like, what always gets me excited about Descript is 15 years ago, I just couldn't make a video. Maybe if I really tried hard and I took a class, like, but you'd never get to hear my story, right? And I think that's true of a lot of people who use Descript.

[00:16:00] We would just never get to hear their story because they're never gonna learn those tools. And what I like about AI is the democratization aspect of it. If you have a vision, you don't need to know what a layer is, right? If you have a vision and you have patience, you can get to something, and you can start to develop as an artistic person, as a creative person, as a storyteller in a way that you maybe never could have before.

And that really excites me. One thing I wanna add on to that, looking at our journey on media creation, we've always been a very media-first company. We've always been very focused on education and on how do we create things that are going to help people learn and do their jobs better, 'cause I think that's a better branding mechanism and, you know, way to build brand than to just have an army of SDRs call until people die.

Don't get me wrong, we have some SDRs, don't get me wrong, and they're fantastic, but, you know, we found there's much more durable ways is education. But we actually did video for a while on some of our stuff we did, and we actually had to stop 'cause it was too expensive for the pro level, all the editing.

Too expensive. And Descript allowed us to do it again because we could have great, really, [00:17:00] really talented technical people and, and people who knew the subject matter, they could get in and they could do it themselves. And that's where the value laid. It wasn't in one-shotting a video. It allowed the people who were the experts to just do the thing, and we never could have done it otherwise, and I think that goes forward for other uses of AI as well.

When I've tried to one shot writing, it's garbage. Yes. When I sit down and do the process, it helps synthesize my thoughts and I move faster. So that takes it to, like, are you an AI native company or whatever the hell that means. But there is, in reality, there are ways we can be using this to help people move faster and get down to the things they're really, really great at and do more of those things.

I mean, I'd love to dig into how you guys are doing it there, 'cause you've helped us do it here. Yeah. I think that how confident I feel that Descript is on the cutting edge depends on who I talk to, right? When you live in San Francisco, you never feel AI native enough. And I think, like, the other thing is, like, we're still trying to figure this out, right?

And some of it's ideological. And so I think, like if you're honest with yourself, most of us are still in experimentation mode. So if the [00:18:00] first kind of framework that I would offer up that's been very helpful in assessing my teams and, like, assessing where we are in the scale is this tech acceptance framework, which starts with outright hostility Like, I don't wanna use this, I won't use this, I think this is bad, to skepticism, which is like, I'm open to using this, I think it's gonna suck.

The last time I tried two years ago, it sucked, but fine, you're making me, like, all right, I will skeptically try. To converted, where you're like, "Look, I believe. I'm a believer. I know that this will work, and sometimes I remember to use it, and I've got a couple of systems down," to, like, rewired. And when you're rewired, you just think of AI first.

Whenever a new job comes across your desk or whenever your role expands, your first thought is just like, "How do I use AI to organize this system?" And so at Descript, starting last year when we had, like, that hackathon, I think that now everyone considers best practice, where you give everyone a hackathon and you say, "We're gonna build with AI.

We're gonna [00:19:00] rebuild, like, a system that you're working on with AI." We did that a year ago, and that was because we had-- we didn't have too many outright hostile people, right? We're in SF, and we've been an AI company since the beginning. If you have a problem with AI, you probably don't self-select into being a Descript employee.

But we had a bunch of people in skeptical, and the majority of us were in, like, converted but not rewired. And there I think what was really useful was to step back and have people think about their function and think about the systems that are part of their function. So, like, in the people function, you have recruiting as a system, you have performance and comp as a system, benefits as a system, answering questions, uh, that employees have about various things as a system.

And you just think about, like, what would it look like if I were rebuilding the company from the very beginning and I didn't have any employees? What would it look like to build this on AI and to make AI the way that we do performance and comp? Now, I specifically picked people because it should make you, [00:20:00] as I say, have AI do performance and comp.

Some part of you should be like, "Oh," right? Like, "Ooh, that sounds kinda weird." And that's where I get into this point of, like, we're all in experimentation mode. I think that we should be very open to reimagining, to asking the question: what would it look like to reimagine these systems if they were AI first?

And to forbid people from answering that question with, "We would buy a tool that says it's AI native." You're not allowed to buy any tools. How would you reorganize the manual work that you do with the contracts you already have? You use Greenhouse? Great. Use Claude and the, and the Greenhouse MCP. Like, no tool changes yet.

How would you rebuild your workflow to be more automated or to be AI native? And I think at the time we were still thinking very human in the loop, and actually I think human in the loop is gonna endure for a really long time, but the latest hotness is what if you actually had an agent managing that system and they were like a self-learning [00:21:00] agent.

I have a lot of concerns there, but we're experimenting because the point is, right now, we should be experimenting with all of this stuff and seeing how far we should push it. But it's not clear to me that AI should be running all of these systems or should be the manager overlooking them, not just for ethical concerns and not just for quality concerns, but also for boring reasons like cost concerns, which I think now a lot of CFOs are waking up and realizing, "What does it mean that the CEO just said we were all gonna do AI native?

Is this really gonna be my bill every month because our margin is gone," right? We're all kind of experimenting to find out what is a good candidate, what's the framework for like this is a really good candidate for like an AI native system that should be like self-running and self-learning. And like this stuff, no, no, no, human in the loop all the way or let's just do it manually.

The other pitfall that we fell into actually with SDR is don't automate a system that's not working. [00:22:00] Automating a bad SDR strategy, it's just gonna give you bad results that are much more expensive for you to get. That's just very true across almost everything with AI. If you give it a capable person or system, you can make that capable person or system, you can magnify, right?

It kind of magnifies everything, so you give it to, you know, something working or in a way that makes sense or you give it to a very capable person, they're gonna get more effective. I do think you've gotten to one of the fundamental limitations to scaling AI within an organization right now, which is right now where we are as a whole, it's very much been let a thousand flowers bloom, which is give everyone AI and see who's using it well.

And we see these like beacons of, holy crap, this person is ten X or twenty X more productive than they've ever been, they're thriving. And then you're just like, "Can I take those best practices and move them around the company?" And it's like, not necessarily. Yeah. Not necessarily across functions, not necessarily across people.

[00:23:00] And also, if that person left, none of that is durable. None of that endures. And I think this is both like a software problem with Claude or whatever your AI stuff is, and also like an interesting organizational systems problem, which is right now if I bring in a fantastic head of product and she starts doing things like having product review, setting up a good design review, setting up these human processes And then she's like, "Oh no, I have to go have a baby.

Like I'm out for five months." Those systems endure and those processes endure. Not forever, but usually for long enough that if she leaves and comes back or if she leaves and you need to hire someone else, product review is gonna take you through most of that, of that time. The way a lot of companies are building AI right now, if the employee who created the AI system were to leave, the AI system just leaves with them.

It's gone. Because it's their own personal setup. I think just recently we've pushed everyone to get onto team plans and all that and shared GitHubs. A lot of these things are working in the business right now, [00:24:00] but you gotta kind of take that and zoom it back a notch to make sure that it's durable.

You're just building a lot of people who know what to do and as soon as they're gone, so is your capability. You know, once you have those... What is the term that everyone likes right now? AI pilled? Oh, yeah. Yeah. Once you have your AI-pilled people, how do you make that scale across the org so people who aren't can learn it, but also when that...

It's durable and just part of the DNA of the company now? Yeah. And so, like, we did the hackathon stuff, which was, I think, a good first step. Then, like, as an org leader, as a head of product, I actually did an offsite that was all around it, where we did. We looked at all of the product systems, so discovery, prioritization, PRD writing, running experiments, creating analytics dashboards.

We, like, listed them all out, and, and we basically created this two by two of, like- I love a two by two. Mm. People knock on them, but sometimes useful. Everything in the world can be turned into a two by two. I believe that in my soul. Yeah. I have yet to find a problem I can't two by two, Jeff. But the first question we asked [00:25:00] is just, like, what do you hate doing right now?

Yeah. Like, when you look at the last four weeks of work that you've done, what sucked? What do you just wish you could do less of, and what do you wish you could do more of? And this is something that I wish we did more when getting folks on board for AI is find the dream. Because, like, as a CEO, I know you're salivating at, like, two X productivity.

You know what two X productivity sounds like to me as, like, an ICPM? You're gonna do two times the work that you're normally doing, and it's cool. I'm super stoked about your vision for the future of AI in the workplace, but it sounds like I'm just doing a lot more work. I also will say I'm super uninspired.

I don't think it's gonna happen, but I'm also super uninspired by this dream of, like, in the future, AI will do all labor and humans won't need to labor. I'm like, I think labor is, like, a social good that connects us with other people. I love labor, but I love the idea of humans getting more choice about the kinds of labor they do, who they labor for, and I think people can get [00:26:00] excited about that.

And so when you're thinking about getting people to be AI-pilled, maybe have them separate their labor into, like, "I hate laboring on this stuff, and I wish I had more time to labor on this stuff." And it's like, okay, we've got our dream right there. We're gonna use AI to try to fast-forward our way through this stuff.

Not all of it, 'cause some of it's like, sorry, a human does need to do that. Uh, that's the other part of the two by two. Is AI, like, good at doing this yet? And so it's like, okay, let's take this stuff off our plate. Let's automate, let's systematize, let's find a way to accelerate through this. So that you can spend more time over here.

And by the way, you might still be using AI, but instead of talking about AI to speed things up or AI to replace, here on the stuff you love doing, let's talk about using AI to enhance your ability to do that, to connect you more deeply with that, right? You wish you had more time to do long-term visioning?

Great. How do we use AI to help you with [00:27:00] your long-term visioning? So that kind of stuff is how you start to craft the promise for your team about here's what AI is gonna give us. Here's the promised land. Like, here's why we're doing this big push. So that's my framework for that. I think we've done well over 100 episodes of this.

That is a top 10, if not top five framework we've heard yet. I love that. It hits on the human element, which is often the problem, because at heart, right, like software's nice. We make software 'cause we have to. It's the best way to solve a lot of problems. Software's not the goal. It's, it's solving people's problems and at heart, and something I think w- that, you know, you're seeing more events and stuff like that and live in person things, and more video too, 'cause it, it feels more human to have a person there than not, and that distills down to what do you actually like doing?

And then how can we use these tools to allow you to do more of that and spend more time on the things that bring you joy? It's the, what is it, Marie Kondo of work. Yeah. Laura, this has been fantastic. Thank you so much for coming on. I really appreciate it. This is enlightening and joyful all at the same time.

Jeff, it was so great to talk. I really enjoyed our conversation. Hopefully, uh, we can stay in touch, but yeah, I'll [00:28:00] let you know when we're in San Francisco for another event. But yeah, great to have you on. Thank you so much. Feature request- Yeah ... product feedback. We've done this. You have a direct line to me whenever you need me.

I love it. Awesome. Thank you so much, and we will take advantage of that. Thank you, Laura. Thanks, Jeff.