97. How to Scale, Adapt, and Win in a Broken Market | Rio Caraeff

29:19And then I'll make a new derivative song. And due to these licensing formulas that are emerging now, all of those people will get paid, right? So there's not, there'll be - So if someone makes a request of that sort, those people would get paid? They get paid because I'm using the name and likeness, right? To create the new song, right? And then I'm also - That's actually happening, is it? These are, this is happening behind the scenes and you'll see this - And it's happening rather than coming soon. You'll see later this year, you'll see some of these things, you'll see them next year. So let me say it a different way.

29:51The music industry is no longer on its heels, like pretending this doesn't exist or trying to litigate as a strategy, right? They're now on their toes, bear hugging it, leaning in and saying, how do we take advantage of this? How do we harness it? What's the opportunity? How do we grow? How do we enable it? There's no more pretending like, let's stick our head in the sand and pretend this doesn't exist and hope it goes away. That's, of course, what always happens at the beginning. And the cycles are compressing. We've seen that before. We've seen that before. And that did happen a few years ago with AI.

30:26So basically, now the industry is licensing these services. And there's different ways to... With AI, there's three parts. There's the training. Like what happens? How is the model trained? and is it trained on licensed works and copyright? And what's the remuneration model for that? The second part is prompting. Like what happens? How does the user make something? Like make me a song like this or make me a song that sounds like this or with this artist. And then the third part is the output, right? What's the end result that's coming out at the end? And so who gets paid? What's the formula? How does that work?

31:08So hang on. So you're envisaging a world in which the training, the prompting, and the output are all potentially resulting in payments to people. I'm not envisioning a world. This is happening. This is actually happening. Yeah, yeah. But people are actually being paid, are they? Not yet. Well, that's what I mean. Yeah, yeah. Because none of these services are live yet. No, but they're in progress. Yeah, yeah. Okay, so we're on the cusp of this. There are services like Suno and Udio that are generative AI music services. Because I've got books behind you that were apparently fed into sort of AI training for which I've received no payment.

31:42Right. And then there are class actions about this. Yeah, yeah. Again, I'm not trying to paint a perpetually rosy picture. No, no, I'm interested in how this might be becoming more positive. Yeah, yeah. I mean, certainly the dust has not settled on, you know, is training fair use or not, right? Right. Courts have ruled so far in the US that training could be fair use, but not if the underlying source material was pirated. Right? Yeah. So if the books were purchased or if the musics was purchased, but if it was stolen or torrented or scraped from security there, then maybe that's not fair use. So there's that.

32:24Yeah. So I think the leisurely pace of the law will play out. That's one way of putting it. Yeah. Can't keep up. That's what we always call it in music is the leisurely pace of the law. The engineers will call it the latency of it. But basically, that'll get sorted out eventually. But business moves on with that regard. And so a lot of what the industry is focused on is how do you license training? How do you license the prompting? And how do you license the output? And because without that, there's no business around any of this. And again, I can't speak to books or any other industry. I'm just talking about music.

33:08No, it's interesting. So the industry is bear hugging it, actively licensing it, and you will see services out, I think as early as this year, you know, that embody these type of things where artists and songwriters get paid for all three, you know, parts of that chain. And so that's what's happening now. I think people will be pleased to hear that. Yeah, that's what's happening now. But I think the other thing I would say, there was a great interview with Strauss Zelnick, the head of Take-Two Interactive, the creators of Grand Theft Auto, you know, video game this week where he was asked about AI and, you know, why do you spend billions of dollars making a game and, you know, five years of production when somebody could just make it with AI.

33:50And I think his fundamental point was that creativity is by its very nature forward facing. Yes. You are informed. We're all informed by the music we've heard and the books we've read and the movies we've seen. But nothing new is ever created by staring backwards. New things and creativity are only enabled by looking forward, right? Right. And AI is fundamentally built on data sets and data sets are by their very nature backwards facing. So if you rely on AI only, you can only make derivative versions of the past. And if you are truly a creator, you're making something new that didn't exist before.

34:31Right. And so. So creativity is forwards facing. It's forward facing and AI is by its very nature backwards facing. Because it's relying on data. Because it's solely reliant on backward-facing data sets. Now, things will evolve and change. But I think that... So I think what is the artist of the future? I like to think about what is the ascendant generation? What is my three-year-old or 11-year-old going to do? What I want to do is irrelevant because basically dead. But the point is that they're going to... the creator of the future will look at these as tools and say, what could I do? What new things can I create using these powerful generative AI tools, right?

35:16That allow me to, you know, have unfettered creativity and powerful superpowers, you know, to make things. But yeah, like I said, the future is here. It's just unevenly distributed, right? And so a lot of people aren't even aware, you know, that the world has changed. So Music's Match recently launched Sentinel. Yeah. What problem is that trying to solve? Yeah. Sentinel is a new service we've launched for businesses that essentially allows them to detect what is being prompted and what is being uploaded to them because they do incur liability and risk if the copyrighted materials are uploaded to their service or prompted into their generative AI service.

36:02So an example, and this is not just music services, it could be, you know, Roblox or Canva or any company that has an upload button or a prompting box. And so if copyrighted works like Taylor Swift lyrics or, you know, other types of songs are uploaded in there and then it shows up somewhere in the service and it's not licensed, you are incurring risk and liability. Well, you're likely in the future to get a bill for that. Yeah, you could be an infringer, you could be sued, right? You could have claims made because those are copyrighted works. And so the first step is detecting what is being uploaded, what is being prompted, right?

36:42How can you figure out what's being put into your business? Well, if you're an organization with lots of people, you don't know what people are asking. Right. So that's what Sentinel is. So you integrate Sentinel into your service where you have a prompt or where you have an upload. And we essentially, in sub millisecond time, we deliver a confidence score to the business. We say, hey, I'm 99 % sure that's a copyrighted work. I'm 100 % sure it's public domain. I'm 100 % sure I've never seen that before and it looks original. We don't tell the business what to do. Business makes their own decisions about whether they want to let it fly or take it down, right?

37:20But the first step is detection, right? And that's what Sentinel is. The vision of Sentinel is actually not to take anything down. The vision is to leave it up because leaving it up is what I call the happy path, right? That's what the user wants. The user put it there. The user wants it to go out, right? And so I always look at YouTube content ID as the model for Sentinel because YouTube content ID is not about taking things down. it's about leaving it up, but you can't leave it up if you don't know what it is in the first place, because you can't, if you have to first figure out what is, what are people uploading?

37:56So that's the recognition part. Then you get to say, okay, now I know it's a Bob Marley song and now I'm going to license it. And then now that I know what it is and it's licensed, I can advertise and monetize it. Right. And then money flows. So YouTube content ID is a$58 billion business, right? And it's not, and none, it's not about taking things down. It's about leaving it up. So because we have 250 ,000 rights agreements, we could also issue instant licenses. Sorry, I'm hitting the table here. So we can say we're 100 % sure it's Taylor Swift love story. And now we could say, and here's a license for it so it stays up.

38:35And then that creates a second business model for us, which is the administration of that copyright and the annuity that that affords. that's dependent on someone who's posted this on youtube paying the license then is it or whatever business is using sentinel so yeah so it tends to be to businesses yeah so yes so sentinel is a service for any business right that has prompts or upload so business to business business to consumer yeah yeah and so um you know there is a a leading generative ai music service that has integrated sentinel today and we are everything that gets prompted everything that gets uploaded hits our sentinel service we deliver the score back to them in you know near real time and then they make decisions about licensing it or leaving it up or taking it down and it's essentially a way for them to manage and mitigate their risk so how does sentinel actually detect these copyrighted lyrics in music that may be ai generated or ai generated yeah and as you can imagine it's built upon our competency you know which is lyrics and lyric data yeah and and we have a the and we know who owns what right so we because we know who wrote it and who owns it and who and who where the splits are and who administers it we we can tell whether it's a copyrighted work or public domain or not because we have all of that information for our business so we that's ai checking it is it again ai is like at this point ai is like talking about oxygen right or air conditioning or you know ai is going to be in your toothbrush and your gmail and so it's a it's AIs in everything, right?

40:10And so we use AI, yes, in order to help it make it work well. But we're looking for strings and matches. Somebody may not upload the entire lyrics, but just an excerpt or a portion or parts of it. And so we're doing all of that pattern matching and identifying it. And we're saying, yeah, this looks like it implicates a copyrighted work. and so the business can take it down or the business can leave it up and we do that because we have all the lyrics and we do that because we have all the ownership data so we know what's copyrighted and what's not. The other point I would make is lyrics are very durable in terms of a signal.

40:49So for example, there are other content identification technologies but they don't focus on lyrics. They focus on the waveform or the way that sounds. Does it sound like a Taylor Swift song? Does it sound like a Bob Marley song? Like, let's examine the, you know, the bits, right? And look at the waveform. But those are easily defeated. You can speed it up and slow it down. You can put white noise in. You can flip it around and change it. You know, you can alter it. You can remix it in a different genre, you know. and all of a sudden the forensic fingerprinting technology says, you know, I don't know what this is anymore, right?

41:31And so lyrics, though, they have to be there. Lyrics are lyrics. Yeah, lyrics are lyrics. They can't be removed, right, because then it's not the song anymore. So they're surprisingly durable and not brittle, and we find that to be a good signal to base Sentinel on. So, Ria, I can hear that there are clearly a lot of opportunities emerging in the space with new technologies. What are the challenges you might be experiencing? Are there any sort of obvious things that you feel challenged by? Yeah, that's a great question. I think fundamentally the challenge we face is how we navigate all of the different parties and constituents in the music industry.

42:16You may have an idea or a service or a product or something you want to do, but there's a lot of cats to herd. There's a lot of people to compel and convince who may not be as far along in their thinking, or maybe more conservative or less progressive. And so a lot of the work is in moving... The hard part is moving everybody forward and moving the industry along and doing that through consensus. So I liken it to that Christopher Nolan movie, Inception, which is it's not, nothing really works if it's your idea. You have to make it everybody else's idea. So you have to be willing to give up ownership over your idea and essentially tell a story that implants the idea in everybody else.

43:05And then they take, then they dream about it and then it becomes their idea. And then all of a sudden they want to do it. And so I try to think about that. It's like, if I want to do something, it doesn't really matter what I want to do. what matters is how do I get everybody else to think that it's their idea so that they can also want to do it as well right and that that is sounds easy but it's really hard and and and I find that that's my biggest frustration often is how do I get things done and then I realize I have to do a lot of communicating and a lot of meetings and a lot of conversations and a lot of work you know to just to take one step you have to herd a lot of cats yeah that's a challenge and that's But that's what you have to do when you want to move the industry forward or do something new on behalf of everybody.

43:53And that's the nature of the fractured copyright or that's the nature of the structural elements of the music business. Again, no one person can do something. You have to get everybody to move in that direction. But this is what I had to do with Vivo. This is what I've had to do with subscription music. And so I'm used to it and I see a path to do it. But it's the hard part is I wish it could move faster. I wish I could break through walls easier. But that's the challenge every day. Right. So you've also launched Music Lens. Yeah. So what exactly is this and how does it work? Music Lens is interesting.

44:34It's the first application of AI that we undertook last year on our business. And the idea was, our thesis fundamentally is businesses who have proprietary data and who serve vertical industries are well positioned with this transformation of AI. If you provide horizontal services like generic databases or CRM or you don't have proprietary data, you're going to have a harder time, I think, in this new world. And so because we have proprietary data and we serve a specific industry, we said, what will happen if we take all of our data and we allow you to interrogate it with the power of a large language model?

45:18so for example now i can learn why are songs about love in a perpetual decline since 1950 you know in in america are they right and they are right and we learn that you know from uh from music lens right right or you can take a different tact and say because the lyrics are not because we're looking at the long arc of time we're looking at streaming volume and activity we're looking at this the meaning and sentiment of the song songs about love are in perpetual decline perpetual yeah for multi-decade decline yeah oh no songs why is that there's a lot of sad stories you know here you know songs about uh songs about politics and civil you know civil disobedience or unrest or protest are also in a perpetual decline right you know the so what are the new songs about they're about uh consumerism they're about autumn you know what are the top three luxury brands you know in songs or automotive brands or alcohol brands and so they're they tend to be they're more about but they're also about friendship and they're about you know love and you know some about love or about sex or other things and so um there it's not there's no it's not all hope is lost you know but but it is things that um you know i'm sorry about this loss of romance but yeah Yeah.

46:40Music used to be a way for people to talk about society and the government and the civil nature and everything from protest songs or protesting the war or these things. But there's not a lot of songs about the Ukraine war or about this in Palestine. There's not a lot of songs about that, unfortunately, there. And so, yeah, so Lens, getting back to Lens, it's essentially a powerful magnifying glass that you can use to interrogate and understand, you know, the corpus of music in whole or in part. And that is a powerful tool that you can use for different industries and purposes. so for example if you were a brand marketer or advertiser and you want to know what are the top you know fashion brands that are sung about in songs in the UK in the last 30 days you know you can use lens to to learn about that or automotive brands or you know consumer products or or so these brands are appearing in songs yeah yeah people sing about Gucci or they'll sing about you know Balenciaga or they'll sing about you know Mercedes or Bentley or different things like that And they show up in different songs in different ways, or they'll sing about different geography or different themes or places.

48:01So you can dissect the semantic meaning and understanding of the songs that way. But there are practical commercial examples. So in music, one of the big revenue streams for music companies and artists is called sync licensing. And sync licensing is where you license a song to be synchronized with a video or a moving picture. so like when you see a tv commercial and there's a song or a video game or a movie and there's a song that that is licensed and under what's called a sync license because it's synced with a moving picture and sync s-y-n-c-s-y-n-c not s-i-n-k not s-i-n-k yeah and so it's an it's in the in the realm of music publishing arcana you know or music you know it's an obscure thing but it's a big business.

48:48It's a multi-billion dollar business. And so my point is, so let's just pick an example. You're a big music publisher. You have 3 million copyrights in your database, and you have 11 people on your sync licensing team, right? Those 11 people don't have 3 million songs in their frontal lobe of their brain at all times. And they get the RFP from Nike or this McDonald's commercial. They're looking for a song to use. They can now drop drop that RFP in that proposal into MusicLens, which may be trained on just their 3 million songs, if they're a customer of it. And then we'll spit out, here are the top five songs in your catalog that are well suited for that proposal.

49:33Some of them may be obvious, some of them may be not so obvious, you may not have thought of them, here's why they're the best fit for them. That could lead to more sync licensing and more commercial opportunities for rights owners. Surfacing these sort of works. Another example could be financial forecasting, right? You're the CFO of a big music company. You're going to use MusicLens to figure out what are the top songs that are likely to generate income for me next quarter? Or what artists, who wrote those songs? And maybe I should re-up my contract with those people. Or maybe I should find other songwriters like that.

50:10What genres are likely to drive more income for me in the future, right? Right. How do I increase my market share, which gives me more pro rata income in a shared pool? So there's all kinds of financial forecasting benefits to Lens. So it's a powerful tool, and we haven't even begun to discover all of the possible commercial opportunities there. The Economist in London is using Lens to tell stories and find trends about the economy, about politics, about society. And then they want to express it through visual charts and graphs and using the power of music to do that. So they're using Lens. There's not much money in that, but it is a...

50:54It's quite an interesting cultural insight. Yeah, but we're happy to have them do it. So we're just uncovering the use cases of it. I mean, one thing that's on my mind is, obviously, there are huge back catalogs of music. Yeah. I mean, is AI and Lens and other tools like it, does it have the potential to surface more of that, bring it back to life, I suppose? Yeah. Even to my sync licensing example, there are some music publishers where their songs may be Sam Cooke from the 60s or Rolling Stones here. And the people who work in those teams maybe generationally come later. They may be 24-year-olds and they may not know everything about every song from the 50s or 60s.

51:41And so Lens could help them understand, hey, this is what's in your catalog. These are the little diamonds that you may have missed. And this is why they're good. And this is why they're useful for this purpose. Yeah. Yeah. So there is. So you sound, I want to sort of draw this up to a conclusion because we've sort of lifted the bonnet on the music industry in a really interesting way and had a look at the sort of engine beneath. And you've been incredibly insightful, which I'm grateful to you for, Rio. Do you feel optimistic about the future with these new technologies that are coming over the hill?

52:16I do. I do. I've never been more excited and more optimistic about it. Why is that? Because I feel like these are all enabling opportunities. They're powerful tools and they give you superpowers. So again, if you're an entrepreneur, I believe there's never been a better time to be an entrepreneur. There's never been a better time to start a company because you are, think of the leverage you have, right? You can now get things done as if you had a team of 10 people, as if you had the team of 100 people. You can now have powerful financial tools and powerful business tools and powerful marketing tools and powerful creation tools.

52:55And you don't need to say, like, how do I raise a bunch of money or how do I dilute my equity or how do I spend six months doing something that can now take six weeks or six days? This is very recent, isn't it? Very recent. You know, I have personally right now, I'm deep into my open claw, you know, which is an agentic AI framework. And I have, and I'm constantly having these mental breakthroughs and unlocks when I realize like what's possible. And I, and it's making, it's enabling me to do all kinds of things that I'd never would have been able to do before. Like, you know, manage my health or be more effective at work or be a better communicator with my 6 ,000 people in my address book and my network.

53:38And so I'm constantly giving my agent, like I used to think of my agent like an assistant, right? And now I think of it like a genie in a bottle, right? Where I just like whisper to the genie, you know, and things happen, right? And, you know, the first mental unlock for me was when I realized that software doesn't exist. It can be willed into existence, right? Right. I said the very first project I gave my AI was, I had 30 years of my address book was a mess. I got 6 ,000 people. It's a freaking mess. I don't have time to go deal with that. So I said, clean that up. De-duplicate it. Make sure all the most recent contact information is in there.

54:24And then put it in a CRM so I can actually stay in touch with these people on a more proactive basis. and and then in my head i'm thinking oh now i gotta license a crm and which crm and i don't have time for that and i gotta modify it and then and then my agent said so you want me to make you a crm right and i was like oh yeah you can do that right and then the next moment when i got my anthropic credit card bill you know from all of the tokens i was burning uh i was like oh this is expensive and it said oh you want me to figure out how to get that to zero right you can do all this, this, and this, and this locally.

55:02And I could set that up for you so you don't have to pay Anthropic to do that. Because I can tell on your network, you've got an NVIDIA 4090 card on this PC, or I can tell you've got this much unified memory and I can just do this, this, this, and make it work for you for free. And I said, oh yeah, do that. It's like, done. Now I don't pay for any of that. And so every day I have these breakthroughs where I can do more and more things. And it's really empowering because there's no limits anymore to my, I don't think, I don't have time for that, or I don't think I don't have the skills for that.

55:33And so it's a great time to start a business. It's a great time to be an entrepreneur. It's incredible leverage. It's a real, it's like, and again, the world hasn't, in the world of tech or San Francisco or London or New York, everybody seems to be like doing this, but you don't have to go too far out of those bubbles to realize that a lot of people are not even aware of this stuff yet. And that's a temporary, not a permanent advantage. It's a temporary advantage. But when you have a superpower and other people don't, it is an incredible advantage, right? And that window will close eventually.

56:08But it's like realizing the world has changed and other people haven't yet realized it's changed. You have a, maybe it's a six month advantage, maybe it's an 18 month advantage, but you've got an advantage over everybody else if you take advantage of these tools. You know, the alternative is to live in fear and to be paranoid and to be negative and to be, you know, here's going to, everybody's going to do this and everybody's going to do that. I tend to also think it's human nature to think in binary absolutes. Like everybody will lose their job and nobody will ever drive a car again and everybody will do this and nobody will do that.

56:43And that's normal human thinking, you know, to go to those extremes. But history shows the pendulum swings hard in directions, but it usually settles back somewhere in the middle and things are never that simple. And there's also a difference between what the technology could do versus what actually happens in the labor markets or these things. And so I tend to be more optimistic by nature because I like to build things. And I'm not an entrepreneur, but I have entrepreneurial tendencies. and you can never, I believe also, you can never be an entrepreneur and build things if you're a pessimist, if you're negative, right?

57:21Because being an entrepreneur and building a business or building a product is by its very nature irrational. It's illogical. It makes no sense. The odds are against you. It's probably not gonna work, right? And if you let that consume you and dominate your thinking, you'll never actually build a company or launch a product. You have to be able to believe in yourself and believe that you can do it. And even if you can't quite figure it out yet, you have to believe that you will figure it out. And so the best entrepreneurs are optimists, I think, and they believe in themselves and their team and their ability to pull it off, even if they don't know exactly how they're going to do it.

57:59Thanks, Rio. Thanks for that uplifting and inspiring conclusion to our conversation. I agree. I think it's important to be optimistic. And entrepreneurs, I think, have to be optimistic to be entrepreneurs. That's right. I agree. Otherwise, you wouldn't have a go. and it's so important to have a go. So thanks so much for coming in and sharing your incredible knowledge of the music industry and what's coming and also some of your own personal stories which I enjoyed very much. Yeah, thank you. I'm happy to be here. I don't think I've ever met anyone who was taught to read by Dolly Parton. So I love that.

58:30I'm a huge fan.

58:35I wanted to just finish as I always do with two questions that I ask all my guests. The first is at Reed. Here at Reed, we love Mondays. So my first question to you, Rio, is what gets you up on a Monday morning? On a Monday morning, what gets me up lately is the smile of my three-year-old daughter. You know, so I love having a daughter. I love having, you know, being able to see, you know, how she sees the world and loves discovering music and art and making things. And so it's really motivates me to think about, you know, my own health and my own, you know, my own ambition and my own responsibilities.

59:18So I, my family and my daughter. Does she start early or three or daughter? She likes to sleep in, you know. That's quite unusual. That's good. Unlike my grandson. Yeah, she does like to sleep, but eventually she gets up in the morning. Oh, cool. So you've got. Yeah. So that's, that's really why I look forward to every morning. Fantastic. And my last question, which is from my interview book, Why You, is where do you see yourself in five years' time? Hopefully retiring in Italy, you know. Headed for Italy. No, but I'm just joking. Just joking. Yeah. In five years' time, you know, I'd like to be, you know, I'd like to still be curious and building.

59:59And, you know, I find my passion and my purpose, you know, is really about supporting people who need to put their fingerprints on the world, right? Who, you know, artists, you know, they all have certain things in common, right? They need to make a living. They need to find an audience. Obscurity is the enemy of artistry. They need inspiration through tools, right? And creativity. And so my mission and my purpose is to solve those things. and that drives curiosity, that drives my ambition. And so to me, that's my purpose on life. I'm not a great musician, you know, but I like supporting musicians.

1:00:39I'm not a great artist, but I like supporting other artists and doing that at tremendous scale if I can. Well, I wish you complete success in that endeavor because I think that's really important. Yeah, thank you. And it gives people a lot of meaning and purpose in life. So thank you very much for sharing those thoughts. Thank you.

1:00:59Thank you, Rio, for joining me on All About Business. I'm your host, James Reid, Chairman and CEO of Reid, a family-run recruitment and philanthropy company. If you'd like to find out more about Music's Match or Reid, you'll find all the links in the show notes. Thank you for listening and see you next time.

1:01:29Thank you.