Ex-Ramp operator Joowon Kim raises $25M to build an AI creative suite for brands and agencies

Speaker 2: We got stocks, options, bond, crypto, treasuries, and more with great customer service. Up next, have Joowon Kim, Melius, the cofounders.

Speaker 1: You know him.

Speaker 2: You know him from his He may

Speaker 1: have post triggered you. Climb for the first time.

Speaker 2: He used to work at Ramp. Now he's building his own company. Welcome to

Speaker 1: the

Speaker 2: How are doing?

Speaker 4: How's it going, guys? Yo, congrats on the success and becoming rich.

Speaker 2: You too. Thank you

Speaker 1: and Always that's odd. Crude in your ways. Yes. You're always and always consistent.

Speaker 2: But you got some funding. Speaking of money, how much did you get? How much did you raise?

Speaker 4: Yeah. We raised 25,000,000. We actually raised this...

Speaker 1: I can't hear you. I can't hear you over the sound of the gong. Congratulations. No. I appreciate you.

Speaker 2: General Catalyst, CRV, back to back series.

Speaker 1: Genius. Genius. Zen, Adam, the whole gang. But... Great lineup. Yeah. Great lineup. What are you building? And and... Go for it.

Speaker 6: Yeah. What's up?

Speaker 1: No. I was gonna say, like, you've you've... It feels like you've been in... Have you been, like, building for a couple years now? Is it more like a year? No.

Speaker 2: Dude, I

Speaker 4: left Ramp back in September of last year, which is crazy to think. Yeah.

Speaker 2: Feels like...

Speaker 4: It's been a wild ride. Yeah.

Speaker 2: It's been

Speaker 4: a wild ride. We raised, you know, from General Kallis back in November. We actually started on a different idea in like performance marketing and then we decided to pivot away from that back in March. And then we landed on building the agents' lab for creative work in April. And then we raised our series a in July.

Speaker 1: Awesome. What does that mean?

Speaker 4: Yeah. So I believe... Well, one, I'll give a little bit of backstory. I used to want to be like Niga Iga back in the day, if you guys know that Asian YouTuber that hit the for the first time. And then I kept telling myself, like, oh, I need to do this. And my dad used to be like a wedding photographer or like a videographer. So I wanted to do... I just... I wanted to learn Final Cut Pro, and I just couldn't. My team is, like, going to... Over the computer. But, anyways, yeah, I just figured, you know, with the era of generative AI and people being able to build software like they've never been able to before, like, that shouldn't be the case for creative work either in creative assets, videos, the coolest motion graphics, animations, bring your imagination to life. And, you know, with general media models making it much cheaper than ever to create, like, a... But it hasn't been easier. And, you know, Melius is gonna be the harness to actually close that gap by building a suite, a creative suite of tools like the node based canvas, the agentic timeline editors, even one off generations all the way to being first to the MCPN CLI.

Speaker 1: Very cool. So so who are you guys... What kind of companies are you working with today? What's the focus? Are you focused on certain types of assets? Like, it could be landing pages?

Speaker 4: Yeah, we're primarily catering to a lot of businesses like in fashion, CPG, e commerce for sure. Some notable customers are doors.com, Chon Media, a shout out to the marketing agency in New York, and Skinny Dipped as well. We also provide services for companies doing even more than $5,000,000,000 in ARR. So super grateful to be working with the group of customers that we are today.

Speaker 1: More than 5,000,000,000. It's time to start guessing, John. No kidding.

Speaker 2: I want to know about 25,000,000. I mean, huge congrats, but it feels like you're going a sort of inference light model. If you're going to be doing generations, you'll probably pass that cost through to the to the user and rely on other inference providers. Is that the thesis? Or do you see, yeah, at some point, we're gonna be training our own models? We're gonna be doing crazy stuff, building hardware 100%. Racking servers?

Speaker 4: I love that you asked that. I don't know if you guys had a chance to check out some of the press release stuff, but the theme of the fundraise is that we're building this so that anyone can make anything. Yeah. And we wanna kinda leave an open ended message there. Like, we're gonna start in general media for sure, but eventually, it's gonna be a play where, you know, like, I hope ten, twenty years down the line Mhmm. People have, like... This is a crazy thing to say, but maybe they have a microchip in their brain, and then they just think of this something that they wanna make and just materialize as an entire thing. Like, that is the kind of thing that we want to strive to build, and we're not gonna make any promises. But, like, you know, as long as we have a goal that says, you know, we're gonna build this company so that anyone can make anything, that is the right strategy, I believe. And this is just the first start. First start.

Speaker 2: I I mean, it seems like at this point, a lot of the a lot of the customer base is is, like, essentially, like, sales driven or you know landing page driven, news driven. Are you optimistic about any of the like foundation model plug in ecosystem driving like being the tool of choice for agents at some point being a growth vector? I've seen a couple companies sort of like go really, really vertical off of that. But of course, there's always like some trickiness to the relationship there. But how are you negotiating like AI discoverability or AI top of funnel?

Speaker 4: Yeah. I actually I actually think that we were pretty... I guess everyone would say this, but we were pretty early to that take, especially in the generative media space. If you if you go back to March, we were still in stealth. And our first post was that we built the first MCP to interact with generative media models on a canvas. And no one had done that before, and we positioned ourselves as the creative canvas for agents.

Speaker 2: Sure.

Speaker 4: And, you know, that had its own kind of, like, bipolar reaction from the market, I guess. But, yeah, I mean, I I totally agree that that is gonna be where the trend goes. Mhmm. That is definitely where, creative work is trending. Most of our customers are using Clod and ChatGPT plug ins with the m c... The MCPN Melius. Sure. In addition to that, even the self-service market, it's growing at a very rapid pace.

Speaker 2: What is demand for computer use like if somebody has successfully learned Final Cut and they want to stay there? Do you think it's important for you to offer an integration point there so they can sort of, like, watch the cursor do some of the work and then they can still jump in with the tool that they know and love. You can also edit the underlying files and then just bend them a project file. Yeah. That that's actually... What what's that? Do you wanna do it all or do you have a particular pattern that works?

Speaker 4: Yeah. We're never gonna be that walled garden that never lets people, like, do parts of things on different platforms. You know? I think plug ins are really great, especially in this era of, like, you know, multimodal computer use. You know, we're gonna build the plug ins for maybe After Effects or Adobe Premiere and whatever it may be. And we're also going to support all those features on our platform. It... The thing is the the reason to stay on Melius is that we've built business conducive and long term creator inducing workflows. That means, like, proactive agents, reactive agents, the timeline editor, the node based canvas, and the digital asset manager. These are all things that, like, maybe people don't have out of the box on their computer. But if they don't need it, that's fine.

Speaker 9: They can actually go

Speaker 4: and use our MCP coupled with their plug ins on their own computer when they are using agents with computer use. And so, you know, there... There's some value to using Melius on the platform, but we definitely are gonna capture a ton of usage off platform through MCPs, through plugins, and and and tools of that nature.

Speaker 1: Do you guys have a, like, an agent or a workflow that can do this yet, which is... I think it would be very powerful if a brand could could, you know, let's say, pick five to 10 companies that they that they look up to or they like their creative approach and say, anytime this company has a new ad in their meta ads library, make, you know, five variations of, you know, basically clone it for my brand, which is something...

Speaker 2: Slap it up.

Speaker 1: No. I don't I don't think it... This is exactly what happened.

Speaker 2: Yeah. That's why brands

Speaker 1: say that. Yeah. Like skinny dipped might see another company that, you know, in a different category with a... With new creative and and marketers today are just going and looking at what their competitors are or or I've seen it. Or other companies are are spending and just cloning it.

Speaker 2: Yeah. Yeah. That's actually a

Speaker 4: common workflow, especially with the Cloud MCP right now.

Speaker 2: Sure.

Speaker 4: What people are doing is they're having the TikTok TikTok MCP or TikTok ads library hooked up. And then every morning, they set up a Cloud project or an automation Yeah. To go look at the top trending ads, and they'll just go dump the script and the transcript and all the key frames in the malleus and then just, you know, reverse engineer everything. And so, yeah, I definitely think that we're... We are conducive to that workflow. It's kind of it's kind of scary what you can do with the tooling of today. But, yeah, a 100% we can support that right now.

Speaker 2: Interesting. Last question. What's your stance on taste?

Speaker 4: Taste? Oh, this is a hot take, actually. I I don't think there is a... There is taste when it comes to, like, UX. Mhmm. I think UX is, like, obviously, you have to make it conducive to how a user perceives, like, the screen and, like, what the user experience should look like. But I don't think there's such thing as slop in terms of creative content. You know, there's a reason why Tung Tung Tung Sahar has billions of views on the Internet. Yeah. Maybe it's considered AI slop, but people find it entertaining. And who's to say that you decide what good creatives are? If people find entertaining and they wanna watch it and they find joy in it, then, like, who who are we to stop that? And which is why we want to say, like, hey. We know we don't wanna bottleneck you on these, like, workflow constraints or these resources. If you have imagination, you think it's funny and you wanna make it, go do it on Melius.

Speaker 1: Love it. Yeah. Taste... Like, if something is tasteful or performant, like, they they can sometimes be the same thing, but very often, they're not.

Speaker 2: Yeah. Yeah. At the same time, some of these AI slop projects are tasteful in the sense that like the person that created them or or had the idea for a particular character, they expose their taste in coming up with the concept. And then, it is, you know, slop generative AI video that might have odd, you know, bugs and hallucinations. But they are expressing their taste and like, I think people want to see this cat dance or whatever.

Speaker 4: Yeah. Of course. Do you guys do guys do the hotel lobby trend a couple

Speaker 1: Offset weeks and Quavo Oh, yeah. Rapping.

Speaker 4: Yeah. I mean, do you think that was tasteful? I thought that was fucking hilarious.

Speaker 2: Yeah. Yeah. I I thought a lot of that stuff was was funny in that.

Speaker 1: Wait. What did you just say? Sorry.

Speaker 4: Sorry. I I asked you guys before. I was like in the email, was like, hey guys, is there something I shouldn't say? And I didn't

Speaker 6: Anyway.

Speaker 1: Congratulations to you and

Speaker 2: the Congratulations on the fundraise.

Speaker 4: Congrats to

Speaker 1: you guys too. Thank you. Only only, you know, only maybe 10 more raises to eclipse your former employer, but, you know...

Speaker 2: You're on the wrath. You're on

Speaker 1: the Just just two x every month forever and you'll you'll be there soon.

Speaker 2: There we go.

Speaker 4: Right. No. I respect them. They did an amazing job. I look up to

Speaker 3: them for sure.

Speaker 2: Awesome. Well, have a great rest of your day. Congratulations and we'll talk to you guys. Goodbye. Let me tell you about MongoDB. What's the only thing faster than the AI market? Your business on MongoDB. MongoDB, don't just build AI. Own the data platform that powers it. We have Rick Dungey from the National Christmas Tree Association, the executive director coming on TBPN.