How to Build Ethical AI for Artists: Kartini Ludwig Pitches Koup Music

0:02 Alan Jones: Founders scale faster on Deel. Set up payroll for any country in minutes. Hire anyone anywhere. Get visas handled fast and get back to building. Visit deel.com/dayone. That's D-E-E-L.com/dayone. So you wouldn't play a guitar without tuning it. So why would you use an AI model without tuning it too?

0:29 Kartini Ludwig: It's great to see something creative and musical out of the Australian tech startup scene.

0:35 Alan Jones: We kind of started last year with initial support from the Sydney Opera House, where we first tested the idea of Koo Music as a live generative remix tool for a performance project that was called Sonic Mutations.

0:47 Kartini Ludwig: Towards the end of your pitch, you bring up the fact that you're raising capital and you say about $1 million, which is totally fine. What are you going to use that $1 million to go and achieve? Welcome to Pick My Brain, the podcast where we help startup founders improve their pitches to better connect with customers, co-founders, and investors. My name's Allan Jones, and I'm an ex-startup founder myself, but now I'm an angel investor with decades of experience helping new businesses find their footing and achieve their goals. I'd like to acknowledge that this podcast is being recorded on Gadigal land, lands that were never ceded. I pay my respects to the leaders and innovators past, present, and emerging. On Pick My Brain, you'll hear real stories each week straight from founders as they pitch their startups, tackle the challenges we all face, and try to turn their ideas into a winning startup. Each episode, we'll see if I can help these founders take their startups another step forward with advice, ideas, and maybe a little tough love. Thanks for joining me. Let's get started.

1:46 Alan Jones: You're listening to a dayone.fm show.

1:48 Kartini Ludwig: As a startup founder, you're juggling multiple priorities from the expected, like finding product market fit,— to the unexpected, like customer requests for SOC 2 or ISO 27001 certification. But achieving compliance is time-consuming, and time spent on that is time away from the needs of your business. That's where Vanta comes in. Vanta is the all-in-one solution for startups to become compliant quickly and build a security foundation with ease. With a combination of automation, an extensive partner network, and a security marketplace, Vanta provides the necessary tools and expertise for startups to achieve compliance seamlessly. No matter how urgent your needs are, and at every phase of growth. Over 10,000 leading companies including Cipherstash, Handle, and Indetted trust Vanta to automate compliance so they can focus on growing their business. Startup customers like you get $1,000 off Vanta at dayone.fm/vanta/brain. That's Vanta, V-A-N-T-A, at dayone.fm/vanta/brain. On today's show, uh, we have Kartini. Kartini, hi.

3:02 Alan Jones: Hi, Allan. How you going?

3:03 Kartini Ludwig: I'm well, thank you. Tell me about your startup and where you got started.

3:07 Alan Jones: Yeah, I guess I'll just start a little bit about me. So yeah, my name is Kartini. It's a bit like martini with a K, which is a bit of a unique name. I have always been a big lover of community and music. Growing up, I was always singing in choirs. And even when I was a teenager, I was in a little indie pop band called ironically AI, which has been super serendipitous because now I'm working on an AI music tool called Koop Music.

3:37 Kartini Ludwig: How do I spell Koop?

3:39 Alan Jones: Koop is K-O-U-P. It's sort of a play on the word like a coup d'état. And at the time we were, well, I was watching a lot about Catherine the Great and some of her coup d'état. So I was very inspired by that at the time and the idea that this tool might be a thing to help revolutionize the music industry.

4:00 Kartini Ludwig: If I want to find out more about Koop, where should I go?

4:02 Alan Jones: You can go to koopmusic.com. We are actually currently doing some beta testing. So if you sign up, we'll give you a link to a staging site in order to ask for some feedback.

4:13 Kartini Ludwig: So that's K-O-U-P dot com.

4:17 Alan Jones: Yes.

4:18 Kartini Ludwig: Cool. Kartini, thanks very much for coming on the show. Let's take a look at your pitch and see what we fire.

4:26 Alan Jones: Yeah, sounds great. A little bit about, I guess, where we came from. Back before the ChatGPT 2 days is what we call them, I was working with a team to prototype a series of machine learning tools aimed at creatives. This sort of ranged from writers and musicians and even puppeteers at one point. But now we are working on KU Music, and it's a tool to create, control, and customize AI music models in order to amplify creativity and provide access to a whole new world of sounds. So this has really come from the idea that in a recent AI music report, well, we've had this idea for a while, but it's been super validating more recently where APRA, which is the Australian Performing Rights Association, have said that 82% of their participants said that they were concerned that AI could impact how they make a living from their work. But this sentiment has also been echoed very much globally where artists like Billie Eilish and Nicki Minaj, Stevie Wonder have all signed an open letter calling for responsible AI practices in music. But we know it's not all doom and gloom because we also have artists who can see the benefit with 54% saying that they think AI technology can assist the creative human process. So there is one thing that highlights this whole problem space really, and that it's that there is a universal arts demand for consent, fair compensation, and creative control when their work is being used alongside generative AI. Which is why we've been really compelled to help artists work with AI instead of against AI. We kind of started last year with initial support from the Sydney Opera House, where we first tested the idea of Koo Music as a live generative remix tool for a performance project that was called Sonic Mutations. This is actually online if you care to watch it on the Sydney Opera House website. And it was really developed in collaboration with local artists. One called Alexis Weaver and another called Rowan Savage. But working with these artists from the outset really helped us to understand that unlike other AI music tools that were being developed at the time that just relied on text-based prompts and were very ambiguously trained on these kind of foundation models where they've been scraped from the internet, what we've designed and Kuh Music is like a lean digital audio workstation or a DAW. A DAW. Which artists themselves can fine-tune a bit like you fine-tune your own instrument. So the way I like to kind of think about this is when you play the guitar, you need to tune the strings to the right key. So it doesn't really matter how skilled you are. It doesn't matter if you're Jimi Hendrix. If your guitar isn't properly tuned, it just won't sound that good. So you wouldn't play a guitar without tuning it. So why would you use an AI model without tuning it too? So one of our feature elements of Koon Music at the moment and what we're beta testing is this studio space. So you can take a little sonic idea like a vocal melody and you can transform that into something like a violin or a different genre of music like metal music or even that of another artist's own sonic idea. So you can sort of collaborate with those artists. Now I'm gonna play a short little demo, an excerpt of what that might sound like as well. And now it's generating. We're currently beta testing with this idea of a pay-as-you-go credit model so that we can start to test the market and tap into this kind of speculative idea of a $1.3 trillion generative AI market with this new creative stream. Our ultimate vision really though is to build a thriving online ecosystem where we can enable our users to use their own custom models and generate new revenue streams by adding them to a model marketplace and redefining and recreating a new kind of creator economy. So we are currently looking to raise about $1 million in seed investment to reach our community and continue iterating from our beta. And we're just hoping to empower people to claim that idea of their own sonic data in a world where they can create and control and customize their own AI music models. And that's it. Thank you.

9:07 Kartini Ludwig: Thanks, Kartini. Great. It's great to see something creative and musical out of the Australian tech startup scene. There've been a number of creative startups coming out of Australia that have done quite well. And the creative startups always have something of an advantage in that you can actually put something really engaging and interesting on a screen. So you can have artists and you can have people performing and, you know, if you are Canva, you can have beautiful graphics and photos and so on and so forth. It's hard for those of us building something that's all about algorithms in the background or, or fintech transactions taking place. It's hard to make those visually interesting and you have an opportunity to, to do that. I love the still photos of, of you and the team and the artists together. I guess one thing that would help, um, the audience that you pitch to is, is to delineate here in this group photo, which of these people are members of the team and which of these people are artist users of the product. Had you considered that?

10:14 Alan Jones: Yeah. In this particular, uh, image on our, our last kind of, um, hero image, we have a photo that was taken directly from the performance from the Sydney Opera House, and it includes some people from my team. I guess for a little bit of extra context, for the last 4 or so years, I have been running my own creative digital studio called Poppy Sue Studio. And so KU Music has been really born out of that. We've always been, you know, very much specializing in product design with this particular focus and interest in AI and creativity. So that's what has sort of landed us here. But in this photo, we have my team, which includes designers. Our collaborators Alexis and Rowan, and our technologist as well, who we worked with, who, um, uh, came from ANU Sports Hypenetics.

11:03 Kartini Ludwig: Cool. So we could maybe get this team photo and caption across the bottom perhaps, or across the top and just say, hey, here's who these people are and this is how big the KU Music team is. And these are the sorts of specialties everybody has.

11:19 Alan Jones: Yeah, for sure. The show.

11:20 Kartini Ludwig: Yeah, cool. Another thing that can be a really powerful part of a startup pitch from a creative or a musical industry background is to actually feature some of the artists using the product. Now I know we had screen recordings there of the AI generating new kinds of sounds based on that first vocal sound. But to maybe to split screen that. With the musical artist who's using the product at the same time. Might be a really fun way to show the emotional reaction from the artist when the new sound gets created.

11:58 Alan Jones: Yeah, I definitely hear you in that. We have been working quite, like, iteratively with specifically Alexis Glebo, who's been like a total champion for everything that we've been doing. And it's been really interesting as well to meet different artists and you know, understanding different ways that they might use this tool in future. But for Alexis and I, we actually recently did a TEDx talk together, and it was very collaborative in that, you know, I did my first 5 minutes that was sort of similar to my pitch, but the second 5 minutes of the TEDx talk was her actually doing a full demo and playing with the tool and then eventually creating an entire musical track out of it, which was a real fire hit as well. So yeah, I— it's, it's so obvious actually that maybe once we get that footage, we can find a way to put that in there as well.

12:46 Kartini Ludwig: Katrini, tell me about the commercial model for people's new sounds.

12:59 Alan Jones: Yeah, we've been exploring sort of 3 different tiers we have, or 3 different tiers or 3 phases really of how we want to release it. At least initially, we're looking at a credit a pay-as-you-go model. So when you, you know, log in, you can pay by credits, and there's sort of a few different tiers of credits. There's like an entry, we call it like warming up, or if you're in the groove, and every time you, you know, generate a different output, you get charged a credit. And then in the future, we are looking for a subscription model. I think right now it's really about focusing on how this pay-as-you-go model can inform the particular tiers of a future subscription model. But ultimately, where we want to get to is sort of a marketplace. And sometimes we speak about it a little bit like it's the Hugging Face, but for AI music models. And a world where artists might be able to go on, they can upload their own, train and upload their own AI music models. And you could subscribe to them in order to create new sounds. And then you're working within sort of an sort of microeconomy, I guess, of different types of sounds in the same way that you might go onto a platform like Splice to source your own samples. I think what we're trying to offer is more of like a collaborative model and a generative model for artists to create more diversity in music, but also collaboration and have the benefit of being able to generate their own new types of revenue from that marketplace as well.

14:29 Kartini Ludwig: How much does it cost to make a sound?

14:31 Alan Jones: From like a base cost or for the customer?

14:37 Kartini Ludwig: Me, the artist, if I wanted to make a sound, how many credits I'm going to use to make one new sound?

14:44 Alan Jones: Yeah, it's like, I think it's about like less than 10 cents per generation. I guess the idea though is because you're in more of a collaborative space and it's about sort of unlocking new ideas and amplifying your creativity through the process, we really don't want this to be like you go in and you find the thing instantaneously every time. I think that the, a lot, this is a lot what some of our artists who we've been collaborating with have spoken about is, you know, if they want something that's very specific, like a very specific 404 sample or something like that, they know exactly where to go get that. But if they wanna do something different, unlock creativity, and amplify their creativity, this is the place that you go and it might take you in new directions. And Alexis Weaver, one of our artists, speaks about this really beautifully in the way that she, as a musician, is sort of a more, I guess, like classically trained, and she does a lot of sort of ambient experimental music. But when she was using ku-music, it enabled her to write this kind of like awesome summer sort of banger that was a lot more upbeat and poppy. And it wasn't really her usual style, but she was having so much fun in the space, in the generative space of these samples that it really took her into a new space for her to explore.

16:00 Kartini Ludwig: Right. Okay. So it costs about 10 cents to make a new sound. And then do I own it?

16:07 Alan Jones: Do you own the sound?

16:09 Kartini Ludwig: Yeah. Or have I licensed it from KU?

16:12 Alan Jones: It depends really. It's something we're really looking to explore and unlock in this like next— it's such a new territory. So currently the idea is that you might be syncing For example, you might be using a little bit of another artist's model to generate your sound. So you want to be able to, you're essentially licensing a part of that particular model from that artist to generate your own sound. Or you as an artist itself might have your own model and might be leasing that out to other artists as well. So yeah, there isn't really a straightforward answer for that just yet because it is such new territory and we need to do a lot more research in that space, but we're hoping to really engage with the community as well. Themselves to help input a little bit on what that looks like. Because the attribution side of things is, I know, one of the biggest challenges in the music industry to figure out, especially in this new era of AI and where a lot of, you know, these larger AI music model companies, Masunos and Udios, are sort of under a lot of scrutiny with the way that they have been sort of ambiguously trained and potentially scraped from the internet, from, you know, people who haven't provided consent. So that's why we're trying to do this as transparently as possible and yeah, work with the community to figure it out.

17:31 Kartini Ludwig: Yeah. Cool. I think one of the ways that we can help investors and potentially customers become more comfortable with experimenting with Ku is by trying to be as simple as we can, you know, so you're stepping forward in the right way now, you know, you're open about haven't got all the answers to all this yet, but we're going to collaborate with our customers to figure out the best solution for everybody as we go along. That's going to feel liberating to some people and it's going to feel like a potential gotcha to other people, you know.

18:06 Alan Jones: Yeah.

18:06 Kartini Ludwig: Potential gotcha for some people who might remember the early days of streaming music when they felt like if anybody was going to make money from their tracks, it was going to be them. And then later on, it turned out to instead be Spotify and iTunes and so on and so forth and not really the artist. Um, so when we're encouraging people to, to, to step forward with us and, and collaborate with something, I think the only, only thing that we can do to help them feel psychologically more safe about experimenting with us is, is to try and be as simple as possible.

18:33 Alan Jones: Mm-hmm.

18:34 Kartini Ludwig: To say, try and make things as concrete and, and as black and white. Now we may review in the future our commercial model, but for now it's like this and it's like that. And any sounds you create while you're with us during this period, they will always be governed by the set of rights that we have for for you now. If we change our model later on, we're not going to drag all of your existing stuff into this, this new rights model. All the stuff you've created will be governed by, by the, the simple black and white model we have for you today.

19:04 Alan Jones: Yeah, I agree. It's, it's, um, it's interesting. It's like the— even just getting to this point, we've been working with, you know, some specific music tech lawyers to try and, uh, explore. And it— yeah, it's been interesting how even creating a beta testing agreement and, you know, our T&Cs for our site is completely new territory for a lot of people in music and technology, lawyers specifically. And exploring just, I think what we have often come back to is this, yeah, if we could be transparent that where we are in the process, then that is a step in the right direction. But I am also conscious that there are a lot of. people that I need to talk to in the music industry in Australia, like the APRA and AMPLES and those kind of people who are going to be very conscious of what the impact of AI and music will be in Australia, but also globally.

19:58 Kartini Ludwig: Cool. Towards the end of your pitch, you bring up the fact that you're raising capital and you say about a million dollars, which is totally fine. What are you going to use that million dollars to go and achieve?

20:10 Alan Jones: Yeah, so currently we have, other than our initial sort of what we felt was like our pre-seed investment from the Sydney Opera House in the form of a commission to build our national MVP, we have been bootstrapping to date with, you know, mostly as an existing creative studio as Kopisu. We've been bootstrapping along the way for that. But now we are really at a point where we want to put all our energy into Kū Music and use those funds to, you know, grow our team, iterate on our beta, continue our market validation. While we have been doing some of our own in-house beta testing, I think to be able to do that at scale, we need capital to do that. We want to do more Axel activations with artists themselves, mostly around market validation, growing the team, and yeah, iterating on the, on the product is what we need that capital for.

21:02 Kartini Ludwig: I'd be worried when we frame it this way to investors that maybe they might feel like they're paying for our salaries for a while and we're going to make further progress. And that may not feel like a safe exchange of value for them. If I came to you and said, "Kathleen, can I borrow 10 grand?" Your natural reaction would be, well, tell me what you want to use it for. And if I said, well, I'm not really sure, and I'm not really sure how long the 10 grand might last me, but I'm going to work, you know, really hard to pay it back. It's just I've got a bunch of things going on and life's not cheap. And you know, where I live, could you help me out? If instead we're able to go back to investors and say, here's some very measurable, accountable goals that me and the team think we can hit. Mm-hmm. With a million dollars by the end of Q1, or by 30th of June, or by the 1st of April, or in 2025, we're going to do this by here and that by there and this by then. That may not be the way that things work out, right? But if we can show investors this is what we're going to do with the capital you're giving us, it feels more like a commercial exchange of value. And I think that might really help you get into a constructive discussion with, with investors towards the end of that pitch.

22:29 Alan Jones: Yeah, that's very true actually. I mean, we, we have, um, as you can imagine, like a lot of people, early startups, we have like a thousand different versions of this pitch deck, um, that have continually iterated. And I think I've often been, you know, unsure of when to bring in, you know, you just want to keep your 5-minute pitch within like, this is sort of generally what we're asking for. And of course like, uh, there are other ways for me to kind of expand on different bits and pieces of that. I guess the longer— yeah, perhaps like the, the more granular version of that answer is that we have done some like financial projections around, you know, if we can reach 10,000 users in the first 3 months, then this would be like the, the amount of revenue we can generate from that and continue to do so. And I think ultimately we had a goal of, you know, how can we get to 100,000 users by the end of 2025, concede that we would be able to start actually generating revenue once we've unlocked that level of users.

23:33 Kartini Ludwig: Do you have some work that you've done to figure out how to get to 100,000 users?

23:38 Alan Jones: In terms of like transferring marketing into, or like activations into—

23:43 Kartini Ludwig: Yeah. How do you get 100,000 people to sign up?

23:45 Alan Jones: We've been working a little bit with someone who is more of a marketing specialist. Like we've had, we've had quite a lot of help from different people in the community and also from former, former life working at Google's Creative Lab. We've got lots of input from old colleagues. It's a roundabout way of saying I personally am not a marketing person. So that's something we keep looking for and we know we need that. And if there is anything I kind of put out a word for from this session as well is that we are looking, I am looking for a CMO and I am looking for a CTO, although we work with, you know, creative technologists and things like that. I know those are the two. missing pieces of the puzzle for this project. And until then, we find those people, we sort of just keep chipping along with the support we have. But in terms of marketing, I know that there is like some of that capital budget that we would need to spend in order to try to convert those kind of campaigns and that material into users, ideally. But also, we have a sense of like, we've started a Discord as we've doing our beta testing, and we're hoping that we can start to organically grow our users initially. And it is a very, like, as an idea, I think, because we are very community-focused, we do want to, uh, sort of take our time to an extent and build our community with very much, like, our values in mind and hope that, you know, it grows with word of mouth. But once we can get some capital to unlock more marketing, we'll do that as well.

25:15 Kartini Ludwig: Well, think about organic growth as something that's always going to provide, you know, the core of the community who are involved with your product and they're going to be the people who help you shape it and establish the character of the brand and so on and so forth. But not very many startups are able to grow purely through organic growth. And even those startups that can grow purely through organic growth can't really control where their organic growth takes them. You know, so there are examples of technologies and products and software products out there in the world that ended up. In the hands of a very different kind of customer than the original founders of the business first imagined, you know, and products get repurposed for different uses. And so a risk with organic growth is, first of all, there will be no organic growth because it sometimes turns out, you know, you build a product and you imagine people will love it and they do love it, but they need to be marketed to and marketing costs money to get them to start trying it.

26:12 Alan Jones: Mm-hmm.

26:12 Kartini Ludwig: But then the other challenge with organic growth is that you're not in charge. Right? So you're not in charge of who you acquire as customers. When you're using paid marketing, you can target people. You can say, I want this kind of artist in this kind of location making this kind of music, for instance. When it's organic growth, it can fall into the hands of people, I don't know, making awful toxic content on X, for instance. So try and use organic growth to kind of keep that core community close. But then start to experiment with paid marketing as soon as you can. Rather than saying to an investor, My goal is to get to 100,000 users in 2025. It's going to sound a lot better to them to say, my goal is to acquire 100,000 users and my starting cost for acquiring a user is, is this much, is, is $1.25. You know, and then we can do the math. We know how much it's going to cost us to acquire 100,000 users. And so then the investor can think about, okay, well, is, is a $1 million raise going to be enough to get to that $100,000 goal? So we don't have to have a full-time marketer on the team to start experimenting with paid marketing content and different distribution channels to see all the different kinds of places we might market this product and get people to sign up for the beta and start testing the kinds of calls to action and the kind of creative that we use in our marketing to get people to click on the link, interrupt what they were doing before and come and try our product.

27:33 Alan Jones: That's so interesting. So I guess it's just like we've never, yeah, I've just never paid for marketing before personally for anything. So, and we, and yet we as a, you know, creative digital studio have created so many, you know, we're always operating in the idea space and the research space of the initial product development and to get to, usually we hand it over to a marketing team, you know, to go and do the thing afterwards. So it's sort of a no-brainer in some ways because we can create all the material in-house. It's sort of an advantage of what we already have. I know other startups that I've, you know, come across, sometimes the biggest thing they need to unlock when they're in the early stage is to engage a studio like ourselves to create the products for them or design the things for them. But we actually have that—

28:22 Kartini Ludwig: For sure.

28:22 Alan Jones: Very easily accessible. So we could probably start to create material to then output in just some social campaigns, right?

28:31 Kartini Ludwig: Yeah. Over time you wanna grow a company that brings that core competency in-house, you know? So, so we, we delegate to agencies or partners for a period of time until we have the, the resources to be able to bring that capacity in-house. Because at the end of the day, most of the software startups out there in the world have, have two arms of the business, right? One arm creates the product and the technology and makes something that people want to use and pay to use. And then the other arm of the business figures out how to market it to people that haven't tried it before, and then how to convince them to start using it, and then how much we can charge them to keep them using it. Those two arms of the company have to be represented in the founding team before they can really work. So, you know, you touched on the idea that that you're not across, but you know that you have some gaps in your leadership team at the moment. And so it can be really helpful to spend a little bit of time wearing that hat a little bit yourself to start to experience what it might be like to go, okay, well, how does one go about putting an ad on Instagram, say, and what kind of creative messaging is gonna convert somebody when I serve them an ad?

29:46 Alan Jones: Good idea.

29:47 Kartini Ludwig: We don't have to have $100,000 to start experimenting with that. You know, we can buy a few ads at a time, try a few different channels at a time to spend a little bit each month. And then maybe 3 months from now when you're pitching to investor in 3 months, you might be able to give them, you know, I know I'm going to be able to drive down my cost to acquire a customer over time, but at the moment it's sitting at about $1.25 or whatever it tends to be. And then an investor can think, okay, I understand this is why this much of the raise needs to go towards acquiring customers. Kartini Ludwig, CEO and co-founder of Koo. Thank you so much for coming on this episode of Pick My Brain. Listeners who are interested, they can find out more about Koo at koop.com. That's koop.com. Thanks again, Kartini, for appearing on the show. Best of luck with everything and look forward to seeing where you're going. Thanks for joining me for this and every episode of Pick My Brains, the advice podcast for every startup founder. Never mind the "don't forget to like and subscribe" bullshit that every podcast host goes on about. Instead, please take a moment to think about someone you know who could use some of the advice I've shared and tell them that they should listen to it. I don't know, maybe they'll choose to like and subscribe. Anyway, I'm not a lawyer or an accountant, and what you've heard today is not intended as financial or legal advice. You should always seek that from a qualified professional before making the big decisions. And I'm not a superhero either. So don't forget that sometimes I'm fallible. And very occasionally, I might even be wrong. Please let me know when you think I might be so I can get better at this too. Just reach out to me on any of our social channels, or email the show at pickmybrain@startupfoundercoach.com. Speaking of startupfoundercoach.com, that's where you might sometimes find show notes, transcripts, and bonus bloopers of I have the time. The Pick My Brain podcast is produced, edited, and beamed directly to your ears by the hardworking and understaffed team at Day One, the podcast network for founders, operators, and investors. Find out more at dayone.fm. Thanks for listening.