00:43:50 · Gorkem Yurtseven, Burkay Gür, Molly O'Shea
[00:43:50] S1: I like [00:43:50] S2: owls. You can do insects. I don't know. Nature, I think, would be my theme. [00:43:56] S0: Well, how did you come up with the name fal? [00:43:59] S2: Alright, so... [00:44:00] S0: Let's talk about this. [00:44:01] S1: Yeah, yeah. [00:44:01] S2: So, fal is... So, we have a longer name, Features and Labels. [00:44:07] S1: It's a double entrant. Yeah, [00:44:09] S2: it's the initials of Features and Labels. [00:44:12] S0: And it's not fal, it's fal. [00:44:14] S2: It's fal. [00:44:15] S0: Yeah, it's [00:44:16] S2: fal. Yeah, it's features and labels which if you're a real OG, you would know those two terms. It's like from the machine learning world. It's like very old, like pre AI. And so, it's the initials for that. But also, it's, I don't know, have you had Turkish coffee? [00:44:34] S0: Yeah, I have. [00:44:35] S2: Okay. So, have you been like told your fortune from like the coffee grounds? No. Alright. So, that's fal. It's like your fortune told by the coffee grounds. You should get it done. We'll do it Alright. [00:44:50] S0: Let's do [00:44:50] S2: it. You know, it's like your future prediction. [00:44:53] S1: Someone looks at when you finish the coffee, it [00:44:55] S0: looks Are you guys trained in that? Can you tell portions? [00:44:58] S1: You know, everyone's trained in Turkey. Everyone has done it before. [00:45:01] S2: Really? Yeah, yeah. It's like... [00:45:03] S0: Should we do it right now? We could Yeah, [00:45:05] S1: we'll get some coffee. I think they have Turkish coffee here.
00:45:05 · Gorkem Yurtseven, Molly O'Shea
[00:45:05] S1: we'll get some coffee. I think they have Turkish coffee here. [00:45:09] S0: Oh, gosh. That's fun. You think the open source, closed source debate is still going to be going on in two to four months, six months? How long do you think that's going to keep on going? [00:45:21] S1: I think so because anyone who's falling behind a little bit, open sourcing their model is an incredible marketing tactic, and the whole world starts talking about you if you open source your model. And there are real advantages. Like if you talk to anyone, any of these studios in Hollywood, they wanna be able to fine tune the model. They wanna be able to apply different, LORA's filters to change the lighting, to be able to use their unreleased characters, their unreleased styles on the models. And the only way to do that is is open source. So there are real technical advantages to it. Also, it's an incredible marketing tactic. So people are just going to keep doing it. Someone's going to create a frontier model, people who are falling behind is going to open source it, and the pendulum is going to keep swinging. We thought this was a bigger threat to our business in the beginning because we were running mostly open source, and then we started partnering up with closed source models and running inference for them as well. So it's a win win situation for us, but open source makes it a more interesting ecosystem and I think it's good for the industry in general. So we always welcome more open source models and we support as much as we can the community.
00:46:42 · Burkay Gür, Gorkem Yurtseven
[00:46:42] S2: I think the fact that people want to protect their IP and that's going to continue, right? Not just for media, but also, like, you want to protect your source code, you want to protect your customer list, whatever, right? All of that, that's always going to be a thing. So, I think, yeah, and if you have something local, you you don't have to worry about it. So, think it's always going to be there, this debate. [00:47:09] S1: Like, when we talk to our customers, if they know the model is run on our GPUs, they trust the whole process so much more. Open or closed, this works the same. But that has been a big differentiation for our business. And obviously when it's open, they know for sure that they can even sell for us, but they trust us because we can do it better than them. But trust is really important with video models especially because people don't want their unreleased content to leak, and it's usually an input to the video model when these creatives work with it. So we've been really pushing on that with the enterprise customers and it resonates.
00:47:52 · Molly O'Shea
[00:47:52] S0: How good is AI at real professional work? Mercor's AI Productivity Index or APEX tracks it with every major Frontier model release. Bill with partners like Harvey and Cognition, see where the Frontier actually stands stands at mercor.com/apex. Today's episode is sponsored by VCX by Fundrise, the public ticker for private tech, allowing investors of all sizes to invest in venture capital. Learn more at getvcx.com. Public is a multi asset investing platform where you can automate your investing with AI agents. Create your own custom agent with a simple prompt. Agents can monitor markets, execute trading strategies, and manage your cash across stocks, bonds, options, crypto, and more. Want to buy at market close and sell the next day's open? Just tell your agent. Automate your investing at public.com. 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 deal.com/sorcery. That's deel.com/sorcery. Talking macro, what do you
00:49:04 · Gorkem Yurtseven, Molly O'Shea, Burkay Gür
[00:49:04] S1: think the biggest question is that people aren't asking right now? Not exactly that, but if you listen to like any Hollywood podcaster or look at the news about Hollywood, they always have this angle of, oh, AI is going to kill Hollywood. And we are like really in this business, like we talk to these people all the time, so far I would say I had zero impact on anything in Hollywood. It's just like people are projecting to the future a little bit. The business we get from Hollywood has been increasing in the last six months. I think it reached to a meaningful percentage of our revenue, but I would say it had zero impact on the labor, it barely made any content that's worthwhile in Hollywood. It's going to happen, there are signs of it, but we've been talking about it for four years, and it's always been fear mongering and nothing real. I think that's the thing that people don't realize from a distance. [00:50:06] S0: Yeah, most of LA, like that's where It's this is really regulatory and what they're doing there with different laws. Exactly. And that's been [00:50:16] S1: the reason the adoption has been slow because it's not only the quality of the models, lawyers of the companies need to be okay with it as well. But finally, slowly, are seeing that everyone is going full force at it. [00:50:32] S0: What do you think people should be asking? What's going on out there? [00:50:36] S2: I think people are asking everything. [00:50:38] S0: They're asking too much.
00:50:38 · Molly O'Shea, Burkay Gür
[00:50:38] S0: They're asking too much. [00:50:40] S2: They're asking too much. Yeah, I don't know. There's a lot of skepticism, I think, about use cases. Again, like, just speaking more broadly about AI, right? There's this certain skepticism about, Oh, like, all of this CapEx investment, where is that gonna go? And, like, know, like, is AI actually useful? But I think, like, I'm more on the optimistic side. I don't think a lot of the people who are making these claims, I don't think they really actually understand, like, what's going on, how, like, AI is actually being used at companies. And, you know, I think gen media is still very early. If anything, it's just starting. It's just like on a I think ads, we're starting to see it. Hollywood, social, whatever, we're starting to see it. I think there's real value. Like, we were just estimating how much faster we're going. Like, our software engineering team is moving, and I estimated somewhere between three to 4x. Would need pre AI, we would literally need like four times the people to go at the same speed. And we're definitely paying less than that. Right? So, there's clear ROI for us. So, yeah, I think if you really, like, go deep into the ROI for the companies who are using AI, like, you really go deep and actually try to measure it, not read it, like, from a bunch of reports or, like, I don't know, nontechnical people, I think you'll see a different picture. So, I'm very in favor, like very long compute, as they say.