Marketing Changes in the Creator Economy and AI Innovations in Producing Content with Descript CEO Andrew Mason

WEBVTT

00:00:00.059 --> 00:00:05.442
Today we are discussing the massive shift in marketing towards the creator economy and how.

00:00:05.442 --> 00:00:11.310
There's over 200 million creators out there creating videos for YouTube and social media.

00:00:11.310 --> 00:00:19.291
There's 51 million YouTube channels, over 5 million podcasts, and people are clearly addicted to Instagram and TikTok.

00:00:19.291 --> 00:00:25.042
So things are shifting massively for marketers as this is becoming very mainstream and we're gonna talk today about this trend that's been happening over the last couple of years.

00:00:25.042 --> 00:00:32.311
So things are shifting massively for marketers as this is becoming very mainstream and we're going to talk today about this trend that's been happening over the last couple of years and how it's really taking off.

00:00:32.311 --> 00:00:37.237
And we have a great guest to help us talk through it Andrew Mason, the CEO of Descript.

00:00:37.237 --> 00:00:37.959
Welcome to the show.

00:00:37.959 --> 00:00:43.060
Thanks for having me.

00:00:43.060 --> 00:00:43.962
So I'm a huge fan of Descript.

00:00:43.962 --> 00:00:47.509
I use it almost every day, many hours every week.

00:00:47.509 --> 00:00:54.593
I love it, but maybe for all the listeners, you could share a little bit about who you are and what Descript is.

00:00:54.960 --> 00:00:56.948
Sure, so I'm the CEO of Descript.

00:00:56.948 --> 00:01:06.081
It's a audio and video editor that's been around for about seven years seven years.

00:01:06.081 --> 00:01:18.421
The really quick backstory on the company is we started out as another startup that was making audio tours like a mobile, location-based audio tours, but basically glorified podcasts and the idea was to build this platform that would let anyone break into this new form of storytelling.

00:01:18.421 --> 00:01:24.673
And what we realized really quickly was that the tools were getting in the way.

00:01:24.673 --> 00:01:38.966
There were all these great storytellers out there who could write, who could talk, but they weren't trained to use these professional tools for working with audio.

00:01:38.966 --> 00:01:42.132
And even the people that we got that were trained.

00:01:42.132 --> 00:01:47.152
It was a very slow process and really stood in the way of the medium breaking out.

00:01:47.980 --> 00:01:58.587
So this was at about the time that automatic transcription was getting really good and we thought wouldn't it be cool if somebody could build an audio editor that looked and felt like a word processor?

00:01:58.587 --> 00:02:04.730
We found someone who was getting their PhD in exactly that to join the company.

00:02:04.730 --> 00:02:12.362
We built it out, showed it to people and it immediately struck a chord.

00:02:12.362 --> 00:02:14.968
And we took it further and said can we figure out how to extend this to video?

00:02:14.968 --> 00:02:20.671
And when we got confident with that, we realized it could be a really big thing and broke it out into its own company.

00:02:20.671 --> 00:02:36.830
And we've just been chipping away at that same vision since then of trying to reimagine the audio and video creation tools to feel more familiar, so that it can be something that's just in the communication toolkit of everyone and the way the docs and slides are.

00:02:38.900 --> 00:02:47.914
And so what made you passionate about spending the last seven years on this with your team of over 100 people?

00:02:47.914 --> 00:02:53.853
What really made this important for you to be the focus for the last seven years?

00:02:55.680 --> 00:03:00.254
I have been messing around with these tools since I was a kid.

00:03:00.254 --> 00:03:16.889
My degrees in music technology before I got into building tech companies and usually sold tools for a long time and felt like we were just in the right place at the right time, at what felt like an inevitable kind of shift in the way that these tools worked.

00:03:16.889 --> 00:03:37.723
Words just felt like a better abstraction of media than waveforms and timelines, and so we felt lucky to be there and it was a big problem that could have a big impact on enabling this, bringing this medium within reach to a lot, a lot of people, and that just felt like a exciting thing to help bring about.

00:03:40.151 --> 00:03:40.451
Awesome.

00:03:40.451 --> 00:03:45.504
Yeah, I think that you can edit like a pro even if you're not one.

00:03:45.504 --> 00:03:50.572
I mean, that's one of the slogans ChatGPT recommended for Descript when I wasn't doing my research for the episode.

00:03:50.572 --> 00:03:55.490
It also had a really good joke about Descript.

00:03:55.490 --> 00:04:00.269
It said it's better to use Descript than to be non-Descript.

00:04:00.269 --> 00:04:03.234
Pretty good, pretty good.

00:04:03.234 --> 00:04:24.877
So let me ask you guys have raised a lot of money, you've gotten a lot of adoption, pretty good things you guys have done in terms of marketing to grow the company over the last few years.

00:04:32.500 --> 00:04:43.324
I would say the best thing we can do is just try to build a good product, and I don't think this is true for every business, but we have a lot of customers like you where Descript is kind of a central tool in their workflow.

00:04:43.324 --> 00:05:02.673
They spend many hours per day or many hours per week in Descript and are passionate about it, and when we do our jobs well and the product works really well, our customers love to talk to each other about it and that's been the thing that's, I think, driven our growth more than anything else.

00:05:04.141 --> 00:05:14.901
So product-led growth has been huge for you guys, right yeah, exactly people can go in there and figure it out without getting a sales demo of it right there.

00:05:14.901 --> 00:05:22.274
I think you guys have a free trial, but people can go in there and largely figure it out, even if they're not a video person, a video professional.

00:05:22.274 --> 00:05:25.644
They can go in there and just figure it out and start editing things right.

00:05:26.266 --> 00:05:36.000
Oh yeah% of our business starts that way, with people just downloading the app rather than like a big fortune 500 company calling and getting a demo from sales.

00:05:36.000 --> 00:05:39.069
It's definitely people just using it and getting drawn in that way.

00:05:41.040 --> 00:05:41.740
One of the interesting things.

00:05:41.740 --> 00:05:43.682
By trade, I'm a chief marketing officer.

00:05:43.682 --> 00:05:59.954
I've been for tech companies for the last 20 years and I want to tell just a short story about when I was a CMO 10 years ago and I was working for this public company and we said you know what we need to get into this?

00:05:59.954 --> 00:06:01.595
Video stuff, video stuff.

00:06:01.595 --> 00:06:04.898
And so I started talking to my creative team Well, what do we need to do this?

00:06:04.898 --> 00:06:12.326
And like we need to buy this really expensive Adobe software to edit it.

00:06:12.326 --> 00:06:18.209
You need to buy a $5,000 computer that's the highest processor you can possibly imagine and then you need to buy about $5,000 more of camera equipment.

00:06:18.209 --> 00:06:22.702
And then you need to hire a full-time video editing person who can do it.

00:06:22.702 --> 00:06:27.625
And I was like that seems nuts to make a bunch of two minute videos for YouTube.

00:06:27.625 --> 00:06:29.733
But I was like, let's do it.

00:06:29.733 --> 00:06:32.442
And so we did it and we were really successful with it.

00:06:32.482 --> 00:06:44.127
Ultimately we got for the videos we produced like a hundred videos that were all one or two minutes, but the cost of those videos was we probably figured out to a couple thousand dollars per minute.

00:06:44.127 --> 00:06:48.163
It was insane and people were like what are you doing?

00:06:48.163 --> 00:06:51.161
And the company they're like why are you spending all this money in this time?

00:06:51.161 --> 00:06:53.565
It was so time consuming and so expensive.

00:06:53.565 --> 00:07:06.591
But we did get 20 million video views from those videos, so it was pretty high impact and ultimately people who came in for an interview to the company would say I've seen your videos and customers would say I've seen your videos.

00:07:06.720 --> 00:07:15.699
And so 10 years ago this was starting but it was so hard, like you're saying, as a company who genuinely wanted to do it Like.

00:07:15.699 --> 00:07:22.694
I used to walk in to the video guy's office and he had this like massive PC and he's just grinding away.

00:07:22.694 --> 00:07:27.346
The two minute clip you need will be ready in two days and I was like, really it takes that long.

00:07:27.346 --> 00:07:30.264
I didn't believe it until I stood there and watched his computer just grind.

00:07:30.264 --> 00:07:38.447
But we're in a much different day and age, just 10 years later, where I can produce a two minute video in 15 minutes.

00:07:38.447 --> 00:07:43.723
It's amazing to think how far we've come, just like in the last five or 10 years.

00:07:43.723 --> 00:07:44.244
I'm very old.

00:07:44.244 --> 00:07:45.908
That's why I draw this contrast.

00:07:47.271 --> 00:07:48.252
Yeah, similar.

00:07:48.252 --> 00:07:50.949
I've been using those tools for a long time.

00:07:50.949 --> 00:07:56.353
Content has changed dramatically since those tools first came about.

00:07:56.353 --> 00:08:12.072
They were designed with making film and television and commercials in mind, not really the daily turnaround that's expected by creators and businesses to keep up with the insatiable appetite that their customers have for content.

00:08:12.072 --> 00:08:22.807
And this is just one of those moments in time where there's an opportunity to kind of reimagine what the tool is from the ground up.

00:08:22.807 --> 00:08:27.415
And you know, what our belief is is the way that you.

00:08:27.899 --> 00:08:32.730
The most powerful way to use AI is to reimagine the tools.

00:08:32.730 --> 00:08:44.000
Use it to reimagine the tools so that it feels better in the fingers of humans, rather than just asking the AI to automatically do everything.

00:08:44.000 --> 00:08:50.065
That can be very powerful and, as a layer on top of a tool, that that feels natural to people.

00:08:50.065 --> 00:09:00.041
We've just seen, when you look at the types of stuff that businesses and creators make, there's still a lot of craft that goes into it and they want a level of control.

00:09:00.041 --> 00:09:15.256
So that's what we're trying to do is like have our cake and eat it too, where we can use AI to improve things but still give people the control that has kept them using the kind of conventional timeline editors for all this time.

00:09:16.921 --> 00:09:26.433
One of the things I love about Descript is that it is a amazing use case for AI, and OpenAI is an investor in Descript right.

00:09:27.274 --> 00:09:27.655
That's right.

00:09:29.020 --> 00:09:42.509
And one of the favorite things that I've watched from the outside that you guys have done in terms of marketing has been your launch around the AI capabilities, which you guys brilliantly branded Underlord.

00:09:42.509 --> 00:09:47.587
Can you talk a little bit about that and how you did that and why you did it in that way?

00:09:47.607 --> 00:09:52.211
Sure, yeah, we launched Underlord towards the end of last year.

00:09:52.211 --> 00:09:52.772
It was kind of.

00:09:52.772 --> 00:10:09.898
You may remember or may have noticed that most companies were branding their AI features with a sparkle icon and you kind of get it because AI feels magic, but worry about what it might mean for creativity, worry about turned into paperclips at some point.

00:10:09.898 --> 00:10:10.899
A lot of worries.

00:10:10.899 --> 00:11:00.125
A name that nodded to the complicated relationship that people feel with under, with ai, and and showed where that that we were very like, decidedly putting it below the, below the human, in the way that we were implementing it in in the product I loved it because I'm a huge fan of ChatGPT but it's not quite there yet is the way, the honest way that people who use it all the time feel about it.

00:11:00.309 --> 00:11:04.181
And the Underlord name is fun because it leans into that a little bit.

00:11:04.181 --> 00:11:11.192
I think your guys' messaging was it's like your intern it's not quite there yet, but it still does a lot for you.

00:11:11.192 --> 00:11:17.177
It's not quite there yet, but it still does a lot for you, and I just thought it was really a fun way to introduce it.

00:11:17.177 --> 00:11:26.423
In a morass of people just using buzzwords around AI, you can't even really understand what they're talking about at the time.

00:11:26.423 --> 00:11:30.827
So I really just loved the product marketing you guys did with that launch.

00:11:39.789 --> 00:11:41.514
And I think I've heard from a lot of other people that they felt similarly.

00:11:41.514 --> 00:11:42.918
What is your favorite ai feature in underlord an oldie, but a goodie.

00:11:42.918 --> 00:11:50.559
One, one that we've had for a while is is studio sound, which will take uh, which comes in handy so often, whether it's for me and I and I see it up level people's content all the time.

00:11:50.559 --> 00:11:52.510
What it what it does.

00:11:52.510 --> 00:12:01.071
It's not just like removing wind noise or background noise or something like that from your recording, but it takes a good recording and makes it great.

00:12:01.071 --> 00:12:20.859
So it takes a recording that was made in your kitchen with your laptop microphone the kind of thing where maybe you don't have the right environment or even the trained ears in order to get it to the next level and just makes it so you don't even need to think about that stuff.

00:12:20.859 --> 00:12:41.083
Let people, let people communicate, let people tell their stories and and take the skill and tedium and time required to look and sound good out of the equation.

00:12:42.490 --> 00:12:56.235
Studio Sound has been a lifesaver for me when I've had guests that came on my podcast and they didn't have a microphone and it sounded like they were in a hollow room and it's made them sound really good just by the click of a button.

00:12:56.235 --> 00:12:58.442
So it's pretty amazing.

00:12:58.442 --> 00:13:00.851
I agree, one of my favorite underlord features.

00:13:00.851 --> 00:13:20.583
The first thing I do after every episode is I go in and it will automatically count up and flag if you click on the transition words that you use and flag all of those and then with one click you can just remove all of those and I say like way too much.

00:13:20.583 --> 00:13:24.701
It's interesting when I look at a lot of episodes that are 20, 30 minutes long.

00:13:24.701 --> 00:13:55.405
The count is often 150 to 200 words between me and the guests, and so when you think how much dramatically better it sounds without all those transition words, it's really an amazing use case for AI in my opinion, because previous to using Descript, I had hired an audio editor which would sit there and he would manually go through and do that 150 times, one by one, and I was like that's nuts, 150 times one by one, and I was like that's nuts.

00:13:55.405 --> 00:13:57.383
That was all manual and now that's how quick it is.

00:13:57.383 --> 00:13:57.789
It's just one click.

00:13:57.789 --> 00:14:01.600
It's really a brilliant use case for AI.

00:14:01.730 --> 00:14:09.899
And the other thing is the show notes to create really good show notes and chapter markers and come up with suggested headlines.

00:14:09.899 --> 00:14:18.878
It always doesn't nail it 100%, but it does it like 80 or 90% and in a lot of cases what I'll do is the other AI tools I have.

00:14:18.878 --> 00:14:35.918
I will do a bake-off between the AI tool and Descript and the other tools, and Descript's AI output for show notes, youtube descriptions is just much better, and that's the sort of stuff that just took a lot of time to do manually for people when you have to do those things.

00:14:35.918 --> 00:14:39.340
So many great features there in Underlord.

00:14:39.340 --> 00:14:43.256
Are there more features coming in Underlord like that in the near future?

00:14:44.058 --> 00:14:50.255
Oh yeah, for sure you can count on there being a pretty continuous drumbeat of stuff we'll be adding there.

00:14:51.190 --> 00:14:57.130
Because AI is moving so fast, just so fast, and it's fun to watch.

00:14:57.130 --> 00:15:05.169
One of the latest things you guys just released is Rooms, your remote recording studio, correct?

00:15:05.870 --> 00:15:08.152
That's right, yeah, in which we're recording right now.

00:15:09.192 --> 00:15:10.332
Yes, I'm very excited.

00:15:10.332 --> 00:15:56.940
This is my first Rooms podcast recording and I us in the video and it automatically does it and that's a sort of professional video editing technique that a lot of podcasts would never get to because they just don't have a professional person doing the editing and the other things about being able to put on a green screen and have a virtual background the world is our oyster and how much fun we can have with that.

00:15:56.940 --> 00:16:09.700
I was thinking perhaps something like it starts like snowing in the background and accumulates at the bottom of the screen, or perhaps something more professional but a lot of times getting the right place to record and the right backgrounds.

00:16:10.370 --> 00:16:13.277
Yeah, we're really excited to be launching Rooms.

00:16:13.277 --> 00:16:59.638
We've been interested in remote recording for a long time and have done various types of integrations with Zoom and other tools, and then about a year ago, we bought a company called Squadcast, which is one of the original remote recording suites and had built some really incredible technology to deliver a stable, high quality podcast recording experience is it's a pretty seamless, simple experience that does what it needs to do, which is give you a high quality recording and a great guest experience.

00:16:59.638 --> 00:17:13.636
And then, when you're done recording, it takes you straight into Descript, where you have your conversation transcribed with speaker labels, and then, like you said, it's really easy to apply some of the Underlord actions that we've built.

00:17:13.636 --> 00:17:23.692
That will generate a multi-cam composition for you that switches between speaker views and kind of a zoomed out gallery view.

00:17:23.692 --> 00:17:35.220
It does some really smart stuff in how it's switching cameras and you can also apply your branded layouts to the multi-cam action.

00:17:35.220 --> 00:17:38.884
That just gives you a level of professionalism that would be hard to do otherwise.

00:17:40.151 --> 00:17:47.924
Yeah, it's really amazing that it does it automagically and you don't have to be a professional video editor to get that sort of functionality.

00:17:47.924 --> 00:17:52.862
That's the thing that blows my mind is it's like magic, like you said.

00:17:52.862 --> 00:18:01.884
Thing that blows my mind is it's like magic, like you said, and I think really it's going to enable creators to take their podcasts, their YouTube channels, to the next level.

00:18:01.884 --> 00:18:15.432
I find it disheartening when I go to YouTube and I see people's channels and they've uploaded a hundred videos and they don't have barely any traction, and part of it is because it's just hard.

00:18:15.432 --> 00:18:22.056
It's really hard to produce good videos before software like Descript and then people don't want to watch them.

00:18:22.056 --> 00:18:23.757
So people spend a lot of time and effort.

00:18:23.757 --> 00:18:41.327
They'll upload 100 videos to YouTube, and just because you upload 100 videos doesn't mean you get anything, and so I think this is really going to be helpful in letting people live the dream, because a lot of the audio podcasters have historically just said podcasting is for audio and they love audio.

00:18:42.849 --> 00:19:06.295
I get it, but I think at the podcast movement conference that we were both at last month, the big topic was doing your podcast both for for audio and for YouTube, because YouTube has 2 billion people watching it per month, and they have a podcast playlist right now that you can use, and I think a lot of people want that sort of distribution and reach.

00:19:06.295 --> 00:19:10.628
They want to tap into it, and a lot of the podcasts that I listen to that are on YouTube.

00:19:10.628 --> 00:19:33.198
They want the ability to show things visually, not just through the audio, and so I'm finding it very fascinating that for the last 20 years, podcasting has been really focused on audio, but as part of this creator economy shift, a lot of the podcasts are now doing both, which could be twice as much work if you didn't have great tools like Descript.

00:19:33.198 --> 00:19:41.112
So I think that is really going to fuel the fire, if you will, of people having successful YouTube channels.

00:19:41.172 --> 00:19:41.593
We hope so.

00:19:43.117 --> 00:19:57.519
And the other thing that is really popular is creating clips for Instagram Reels or TikTok or even YouTube Shorts, and prior to Underlord, I was trying to do that.

00:19:57.519 --> 00:20:05.375
I couldn't do it, it was too hard for me, and since then I've been creating all kinds of clips.

00:20:05.375 --> 00:20:06.778
It's super easy.

00:20:06.778 --> 00:20:17.684
I can do it in just minutes and you can do all the simple things like apply a headline and even put in some media like a picture that automatically gets generated from underlord.

00:20:17.684 --> 00:20:20.553
You can make some really great clips in just a few minutes.

00:20:20.553 --> 00:20:21.214
It's not hard.

00:20:21.214 --> 00:20:24.148
You don't have to be a video genius.

00:20:24.148 --> 00:20:28.558
Have you guys seen that that people are really digging in on the clips feature?

00:20:29.807 --> 00:20:32.112
yes, it's been quite popular it's.

00:20:32.112 --> 00:20:48.258
It's nice because people will edit their podcast in descript and then just having a one-click way to create clips that you can use your branded layouts on it just couldn't be faster to do and it's a quite popular feature.

00:20:49.986 --> 00:20:52.113
So where do you think things are going?

00:20:52.113 --> 00:20:53.195
Where do we go from here?

00:20:53.195 --> 00:20:55.726
What is going to happen?

00:20:55.726 --> 00:21:02.109
As the creator economy has really been surging, how do you think this unfold role in the shaping of content?

00:21:24.964 --> 00:21:33.644
And I guess, to some degree, we're counting on it, because if you don't need humans, then you don't need Descript.

00:21:33.644 --> 00:21:43.808
You don't need Descript, but AI is going to make it a lot easier to make bad video as well.

00:21:43.808 --> 00:21:57.119
We'll get to a point where the marginal cost of creating content is very, very low and, as that happens, it stands to reason that the differentiator for content creators is still going to be craft.

00:21:57.119 --> 00:21:59.828
That's going to be the thing that gives them a competitive advantage.

00:21:59.828 --> 00:22:08.964
So we are as excited as we are about building tools like underlord that can automate the most tedious parts of editing.

00:22:08.964 --> 00:22:15.436
We still are primarily focused on enabling craft.

00:22:16.598 --> 00:22:31.535
Yeah, I think about AI broadly and using it to create audio and video is we're now more than ever, as humans, directors, and what I've noticed is people who are really good directors.

00:22:31.535 --> 00:22:38.114
They're good at directing employees and people, but they're also really good at directing AI.

00:22:38.114 --> 00:22:38.694
What to do?

00:22:38.694 --> 00:22:48.085
You can be specific, you can be prescriptive, you can walk it through a number of drafts to get to the right answers rather than just taking what it gives you for the first time.

00:22:48.204 --> 00:23:09.424
And so I think about this, because I'm a huge movies fan and I think about how hard all the great movies I've loved over the years must have been to make, how much the directors had to do, how many different things they had to think about and really direct from the script to the actors, to the lighting, to the video, to the effects.

00:23:09.424 --> 00:23:11.613
All of it is making it all come together.

00:23:11.613 --> 00:23:19.979
That's really what humans have to do with AI is you have to bring it all together, because it has all the information that has been trained on.

00:23:19.979 --> 00:23:23.714
But bringing it together is actually the hardest part of it, right?

00:23:25.766 --> 00:23:26.691
Totally yeah.

00:23:27.726 --> 00:23:32.917
So what advice do you have for creators to be successful?

00:23:32.917 --> 00:23:48.390
What do you counsel customers and people who talk to you and the team at descript on how to live the dream, if you will there are people who are much better qualified to give advice on becoming a creator than I am.

00:23:48.650 --> 00:23:49.551
We just make the tool.

00:23:49.551 --> 00:23:53.606
I can tell you like getting reps in and practice.

00:23:53.606 --> 00:23:56.314
I mean, this is what everybody says and it's definitely true.

00:23:56.314 --> 00:23:59.320
Just the more you do it, the better, the better you get at it.

00:23:59.320 --> 00:24:06.938
So trying to trying to develop a ritual where you're forcing yourself to to get the reps is is the best way to improve.

00:24:08.925 --> 00:24:10.128
Yeah, that makes sense.

00:24:10.128 --> 00:24:16.126
I definitely am better at podcasting and interviewing and producing YouTube videos.

00:24:16.126 --> 00:24:18.855
After 160 episodes was after the first 10.

00:24:18.855 --> 00:24:21.531
Practice makes perfect, for sure.

00:24:21.531 --> 00:24:29.603
What haven't I asked that you can share about the journey you've been on there at Descript and what you guys are doing?

00:24:30.306 --> 00:24:50.271
Some stuff we're shipping later this year will be focused on visual presentation and making it easier to work with templates and layouts and just making that feel like a more tightly integrated part of the experience, kind of like a more first-class part of the experience.

00:24:50.271 --> 00:24:57.855
We're quite excited about that and the way that it is going to make using Descript feel so something to look out for.

00:24:59.178 --> 00:24:59.739
That's exciting.

00:24:59.739 --> 00:25:03.353
Is that mainly what the next season is about, as you guys call it?

00:25:04.335 --> 00:25:04.977
I think so.

00:25:04.977 --> 00:25:14.772
Yeah, seasons are kind of our way of just cobbling together the things we've been working on over a three to six month period, and that may very well be the thematic focus.

00:25:16.698 --> 00:25:17.018
Awesome.

00:25:17.018 --> 00:25:21.946
Andrew, thank you so much for joining us today.

00:25:21.946 --> 00:25:28.729
I appreciate you and the team letting me be one of the first people in the world to use Rooms and put out a podcast recorded on Rooms.

00:25:28.729 --> 00:25:32.237
This is quite the momentous episode from that perspective.

00:25:32.237 --> 00:25:33.905
I'm loving it.

00:25:33.905 --> 00:25:36.267
I think all your customers are going to love it as well.

00:25:36.267 --> 00:25:41.308
So thank you for all the great work you and the team have been doing and appreciate you being on the show today.

00:25:41.930 --> 00:25:43.089
Thank you, eric, it was a lot of fun.