Master AI Video Marketing w/ Magnific CEO Joaquín Cuenca Abela | AI Basics

Summary

Jason Calacanis sits down with Joaquín Cuenca Abela, co-founder and CEO of Magnific (formerly Freepik), for an AI Basics episode focused on how modern marketing teams can use AI video tools to do things that simply weren’t possible before. Joaquín demos a post-apocalyptic San Francisco ad for This Week in Startups that he built single-handedly in 24 hours — complete with storyboards, character generation, a rusted-out Golden Gate Bridge, a tattered TWiST logo, and a dystopian Jason Calacanis playing the worn-down guru that a young entrepreneur seeks out for advice on rebuilding civilization.

The conversation digs into the economics of AI video: roughly 10 cents per second of output (about a dollar per second when you account for retries), versus the $10K–$20K per minute traditional production costs. Joaquín explains that the real budget shifts away from the platform and onto creative talent. Magnific blends its own image models (upscaler, skin enhancer) with state-of-the-art third-party models from Google and others. Hollywood studios — including the production behind House of David — and small marketing teams are both adopting the tools, and Joaquín argues AI is creating more film jobs by greenlighting projects that previously couldn’t fit a budget.

The episode closes on creativity, individuality, and the future of hypertargeted ads. Joaquín believes AI will match the creativity of some humans but never replicate the individuality that makes a creator unique — which will drive more demand for great storytellers. Jason floats a Minority Report-style future where 100 versions of an ad are generated per city, and Joaquín explains why static localization works today but full video personalization still needs something like “Claude Code for video” — a harness that checks AI output and steers it back on track.

Highlights

”What a single person can do in 24 hours”

“I put together like a few restrictions. Okay, I wanted to show what a single person can do in 24 hours, and I’m at that point of making it myself. Okay, I’m not creative. My background is in tech, so there’re going to be some glitches in the demo.” — Joaquín Cuenca Abela, 3:19

yt-dlp --download-sections "*3:19-4:02" "https://www.youtube.com/watch?v=Mva8ccgaDOI" --force-keyframes-at-cuts --merge-output-format mp4 -o "single-person-24-hours.mp4"

”10 cents per second” — the new economics of video

“Per second is around 10 cents nowadays. Uh, to be honest, like you usually take like, you do multiple tries to get the right one. So you should assume like there is like one dollar per second. That’s nothing. Like the real budget here is going to be your creative team spending their time and doing something great. It’s not going to be anymore on the platform.” — Joaquín Cuenca Abela, 10:13

yt-dlp --download-sections "*10:13-11:00" "https://www.youtube.com/watch?v=Mva8ccgaDOI" --force-keyframes-at-cuts --merge-output-format mp4 -o "ten-cents-per-second.mp4"

”Raising the ceiling and raising the floor”

“The great thing about this is that it is raising the, the ceiling and it’s also raising the floor, both of them. So you have Hollywood studios, they are using it… the company that does uh, the House of David, they have used Freepik to make the production. It’s used by all the big studios in Hollywood nowadays and it’s used more and more.” — Joaquín Cuenca Abela, 12:04

yt-dlp --download-sections "*12:04-13:03" "https://www.youtube.com/watch?v=Mva8ccgaDOI" --force-keyframes-at-cuts --merge-output-format mp4 -o "raising-ceiling-and-floor.mp4"

Gal Gadot on AI: “We are now able to make films for one-third the price”

“I was talking to Gal Gadot, she, uh, is very famous for playing Wonder Woman… And I asked her hey, what’s your take on all this AI stuff? She says it’s incredible. We are now able to make films for one-third the price. We can, you know, film ourselves on a green screen or a grey screen, whatever it happens to be, and then let everybody else do the rest of the work.” — Jason Calacanis, 13:03

yt-dlp --download-sections "*13:03-14:27" "https://www.youtube.com/watch?v=Mva8ccgaDOI" --force-keyframes-at-cuts --merge-output-format mp4 -o "gal-gadot-one-third-price.mp4"

A film that wouldn’t have been greenlit — now employs 600 people

“I talked to some Hollywood directors and what they told me is that they had a project that it would have never been greenlighted because of the cost to make the project. But now as they managed to reduce the cost, they managed to get that film and employ 600 people doing the film, while otherwise… it would not have been enough budget to do it. So it’s actually creating more jobs.” — Joaquín Cuenca Abela, 14:27

yt-dlp --download-sections "*14:27-15:14" "https://www.youtube.com/watch?v=Mva8ccgaDOI" --force-keyframes-at-cuts --merge-output-format mp4 -o "600-jobs-from-ai.mp4"

”We still need kind of a Cloud Code for this”

“If you try it with a full video like this one, the problem is that at every step you have a decent probability of error, decent probability of things going wrong. So at the end you take like one minute and it completely loses the track, okay? It diverges. It’s not yet… I think we still need, and maybe we’ll provide that, we still need kind of a cloud code for this. Something that harness the AI, checks the output and kind of steers it back.” — Joaquín Cuenca Abela, 19:46

yt-dlp --download-sections "*19:46-20:40" "https://www.youtube.com/watch?v=Mva8ccgaDOI" --force-keyframes-at-cuts --merge-output-format mp4 -o "cloud-code-for-video.mp4"

Key Points

• AI is moving faster than any tech we’ve seen (0:00) - Jason notes law changes every 5-10 years, finance every couple years, but AI changes month to month
• Magnific intro (1:50) - Co-founded by Joaquín Cuenca Abela, based in Spain, 5 billion all-time downloads, 150M monthly users
• Doing things that are new, not just cheaper/faster (2:26) - The thesis: most teams use AI to do the same work cheaper; Magnific wants to enable what wasn’t possible before
• Demo: post-apocalyptic TWiST ad in 24 hours, solo (3:11) - Joaquín builds a full cinematic ad with characters, storyboard, rusted Golden Gate Bridge, dystopian Jason
• Storyboards as a primary workflow (4:22) - Create multiple images, combine them, then generate video from the storyboard
• Text prompts and image references mix (4:58) - All-text works, but pulling reference images from the internet helps when prompts aren’t precise enough
• The clip plays — Grandpa Jason, post-apocalyptic guru (6:30) - The actual generated ad rolls in the episode
• Economics: ~$0.10/sec, ~$1/sec with retries (10:13) - Real budget moves from platform to creative team time
• Magnific stack — own models + best third-party (10:57) - Owns upscaler and skin enhancer; uses state-of-the-art third-party models, including Google’s
• Hollywood adoption is exponential (12:04) - Studios behind House of David and others use Freepik/Magnific; lines blurring between writer, storyboard artist, director
• Gal Gadot at the Breakthrough Prize (13:03) - Hollywood went from “F these tools” to “we can make films for 1/3 the price” in two years
• AI creates new film jobs (14:27) - Projects that wouldn’t have been greenlit are now employing 600+ people
• Recruiting AI talent to Southern Spain (15:46) - Not San Francisco, but quality of life + reshaping creativity sells the mission
• Creativity ≠ replaceable individuality (17:25) - AI is a digital artifact you can copy; humans can’t be copied, so storytellers will be more in demand
• Hypertargeted ads — 100 city-specific versions (18:28) - Jason floats Minority Report-style targeting; works for static today, not yet full video
• The “Cloud Code for video” problem (19:46) - AI video diverges over a minute; needs a harness that checks output and steers it back
• Wrap and CTA — magnific.com (21:46) - Plus the Google Cloud “Future of AI: Perspectives on Generative Media for Startups” report

Mentions

Companies

• Magnific (1:50) - Joaquín’s company; AI creative platform for marketing teams and studios
• Freepik (1:50) - The original brand; 5B all-time downloads, 150M monthly users
• Google Cloud (1:14) - AI Basics series sponsor; “Future of AI: Perspectives on Generative Media for Startups” report
• This Week in Startups (0:00) - The podcast itself, featured as the subject of Joaquín’s demo ad
• Mistral (15:21) - Mentioned as evidence of strong European AI talent
• TikTok / CapCut (8:22) - Jason’s daughters use these for storytelling, but lack Magnific’s palette
• Instagram / YouTube Shorts (9:21) - Short-form platforms where startups now publish launch videos
• Tesla / Cybertruck (18:28) - Used as a hypothetical hypertargeted ad example
• Breakthrough Prize (13:03) - “The Oscars of science”; where Jason talked to Gal Gadot

Products & Technologies

• Magnific upscaler (10:57) - Magnific’s own image model technology
• Magnific skin enhancer (10:57) - Another in-house model
• Nano Banana (10:34) - Mentioned as one of Google’s amazing models making great progress
• Frontier models / open source models (10:34) - Referenced as the ecosystem Magnific draws on
• Cloud Code (19:46) - Joaquín’s analogy for the missing harness/orchestration layer in AI video
• goo.gle/futureofgenmedia (1:14) - Link to Google’s report on generative media for startups

People

• Joaquín Cuenca Abela (1:50) - Co-founder and CEO of Magnific (formerly Freepik), based in Southern Spain
• Gal Gadot (13:03) - Wonder Woman actress; told Jason at the Breakthrough Prize that AI lets Hollywood make films at 1/3 the cost
• Lon Harris (1:14) - TWiST co-host present throughout the conversation

Films / Shows referenced

• House of David (12:04) - Production used Freepik/Magnific
• The Walking Dead, Better Call Saul, Blade Runner (11:22) - Lon’s references when discussing Hollywood writer use cases and dystopian aesthetics
• Minority Report (21:00) - Lon’s reference for ads addressing you by name, served in real-time

Surprising Quotes

“Many people are focused on how they can do the same things that they were doing before but cheaper and faster. We are going to talk about the void when you do things faster, what can you do that is new.” — Joaquín Cuenca Abela, 2:26

“The real budget here is going to be your creative team spending their time and doing something great. It’s not going to be anymore on the platform.” — Joaquín Cuenca Abela, 10:13

“What I think is unique to us is that it’s our individuality… AI is very different because it’s a digital artifact. So you can copy it. Okay, and humans you cannot copy it.” — Joaquín Cuenca Abela, 17:25

“Sometimes people get scared, oh, people are going to lose their jobs, but we’re also going to see a Cambrian explosion in creativity.” — Jason Calacanis, 13:03

“Sometimes you want the opinion of the creative to drive the narrative. It’s like when you go to a restaurant — if they tell you you can get any dish that you want, here you have all the ingredients. That’s too much. I want your opinion on it.” — Joaquín Cuenca Abela, 21:13

Transcript

Jason Calacanis: 0:00 Hey everybody, welcome back to our basics series. I get asked all the time as an angel investor, as somebody who helps grow companies, what should I do in terms of customer motion and finding my ideal customer profile? I get asked, hey, how should I do my cap table, legal issues, finance issues? How should I do my revenue recognition? And increasingly, the last two years, how should I use AI to make my startup grow faster, produce better quality products, better, cheaper, faster, you know, that’s always the mantra. And you can see all of these great episodes at thisweekinstartups.com/basics. We started doing AI basics to help people understand the core things this technology’s allowing you to do. And the challenge is, this technology’s moving faster and faster, faster than anything we’ve ever seen. It’s easy when we’re doing legal because that changes every 5-10 years. Finance, hey, that changes also, like hey, every couple years you might get something out of the IRS to tweak. My Lord, when it comes to AI, it is changing month to month and we’ve got a great guest today to help us with creative.

Lon Harris: 1:14 And of course, this AI basic series is brought to you in partnership with my friends over at Google Cloud. Google Cloud really cares about founders, they’re constantly giving credits and tools and this very series to help you as a founder figure it out and just figure out how to scale your company and increase your chances of success. You can check out Google’s brand new future of AI perspectives on generative media for startups report. You can access that at goo.gle/futureofgenmedia, all one word, no spaces, no dashes, and we’ll put that in the show notes and the YouTube description.

Jason Calacanis: 1:50 Today we are lucky to have Joaquín Cuenca Abela. He is the co-founder and CEO of Magnific, Magnific I think would be the non-French way to say it, formerly known as Freepik. They’re based in Spain, I understand. Five billion all-time downloads, over 150 million monthly users for a reason. We’re going to talk today about how to use his creative tools, hey, to go help founders be more creative, but then turn their marketing into video, which is a big part of what you’re doing, Joaquín, yes?

Joaquín Cuenca Abela: 2:26 Yes Jason, thank you for having me. We are going to talk about how a marketing, a modern marketing department can do more with AI. Many people are focused on how they can do the same things that they were doing before but cheaper and faster. We are going to talk about the void when you do things faster, what can you do that is new.

Jason Calacanis: 2:44 And marketing is one of those spaces that has data, it’s got creative, and then there’s a whole bunch of best practices and alchemy between these two, yeah? You have the on one side you’ve got this great science of marketing and then on the other side… You’ve got the creative and in the middle, dare I say, alchemy. So let’s see what you’ve built and how it works. Maybe you could do a quick demo for our audience of founders here.

Joaquín Cuenca Abela: 3:11 Sure, of course. Let me share my screen. Okay, listen, for this demo, I took the liberty to make it a little bit personal for you.

Jason Calacanis: 3:18 Great.

Joaquín Cuenca Abela: 3:19 Okay, I put together like a few restrictions. Okay, I wanted to show what a single person can do in 24 hours, and I’m at that point of making it myself. Okay, I’m not creative. My background is in tech, so there’re going to be some glitches in the demo. I hope you will forget me. Okay.

Lon Harris: 3:35 You’re pre-forgiven. Yes, it could be some glitches here and there.

Joaquín Cuenca Abela: 3:40 Okay. So, the scenario here is I’m going to put together a little ad for your show for this week in startups, okay? And I’m going to put it in post-apocalyptic San Francisco just…

Jason Calacanis: 3:53 Love it.

Lon Harris: 3:55 Okay. I’m all about the Blade Runner, post-apocalyptic, Walking Dead, zombies. I’m there.

Joaquín Cuenca Abela: 4:02 The first thing that I did is I created a few characters here that I’m going to use on this little show, okay? So this one in particular, he’s looking for advice. This comes from a book that I read a while ago about science fiction that is from the 1915, very old book. So I decided to take a little bit this background and mix it with yourself, okay?

Jason Calacanis: 4:21 Love it.

Joaquín Cuenca Abela: 4:22 So, the first thing is I created a little storyboard for the opening. Okay, so you see how these three characters they’re walking through the Golden Gate. So everything is devastated.

Lon Harris: 4:38 Oh, it’s like storyboards.

Joaquín Cuenca Abela: 4:40 It’s storyboard. Okay, so you can create here like multiple images, you can combine them all together, and then you pick them up and you create a video based on that storyboard. Okay, so you see…

Lon Harris: 4:52 And now, do you do that with text prompts, or do you just find some images as inspiration and then add text prompts to it?

Joaquín Cuenca Abela: 4:58 Great question. There is a little bit of both. In this case, it was all text prompts, okay? Uh, but in some cases, I like to pick like some images from the internet, use them as a reference, especially when the text is not precise enough. It’s very good to use some existing images. But in this case, it was all generated from scratch from some text. Okay.

Jason Calacanis: 5:18 Great.

Joaquín Cuenca Abela: 5:19 And then I took your logo, This Week in Startups, I made it like rusty and old and all of that, okay?

Jason Calacanis: 5:26 Yes. Very dystopian. Yeah.

Joaquín Cuenca Abela: 5:29 Exactly. And then I introduced it in a video and I made the same with you, okay? So I took a few images from yourself and I made it post-apocalyptic, like you’re kind of, you know, worn down a little bit.

Jason Calacanis: 5:40 No haircut, I got a goatee. Love it.

Joaquín Cuenca Abela: 5:45 Yeah, exactly. So, you are going to be kind of the guru that those three people seek for advice into how to do like a big project. Again, it’s a bit messy because the creative process is messy, but here it shows how you can go from one video to… the next from one image blended all, use audio references. For example I used like this image from the internet as a shotgun that you are having.

Jason Calacanis: 6:11 Yes. And I’m a little dirty, I’m a little greasy, it’s been a rough post-apocalyptic era for J-Cal. I’m in my post-apocalyptic era.

Joaquín Cuenca Abela: 6:20 Exactly, let’s say it’s been a rough week, okay? So, we put it all together and at the end I got this little clip here. I’m going to play it.

Jason Calacanis: 6:30 I am excited to see this. Okay, so here we go, the bird screeching, we’re going over a rusted-out Golden Gate Bridge. There’s an old man walking down the bridge, he’s out down on his luck. There is no one, everything is desolate. Oh, there’s Fort Mason, you know, the various buildings on the north part of San Francisco. Yeah, here we go.

Joaquín Cuenca Abela: 6:57 Just see that the young boy, he’s happy because he’s still a boy, so there is a little bit of hope in the ambiance.

Jason Calacanis: 7:03 I mean, this looks amazing. Oh, and there it is, there’s the This Week in Startups tattered logo, like a billboard or something.

Joaquín Cuenca Abela: 7:11 Exactly. And they get to this whole building…

Jason Calacanis: 7:13 Maybe my studio? Maybe it’s a studio. That’s it, okay we’re going there, we’re going there. Here we go. Who’s that? He’s getting his laptop. Oh, he’s got a shotgun and his glasses on. This is some dark dystopian stuff here. It’s me, Jason. There he is. Grandpa? I thought you were dead. It’s my grandfather.

Lon Harris: 7:34 The kid needs your help.

Jason Calacanis: 7:36 I don’t do that anymore.

Lon Harris: 7:39 Who knows? I want to build. What do you know about building? What have you ever built? Do you even know how it feels to lose everything?

Jason Calacanis: 7:54 Yes.

Lon Harris: 7:56 I want to rebuild civilization.

Jason Calacanis: 8:07 Wow.

Lon Harris: 8:08 Wow.

Jason Calacanis: 8:09 Come on, follow me.

Joaquín Cuenca Abela: 8:18 So not bad for one day.

Jason Calacanis: 8:22 I mean, it’s so inspiring and fun, sincerely. Like, I could see, like it makes me want to use this this weekend with my daughters because they are into storytelling and they use CapCut and some of these, you know, like TikTok-type tools, but it doesn’t give you this palette. Now, when you were showing all of those storyboards, like sketches and whatnot, and they were getting attached together, did you draw those lines or did the AI draw the lines between the different scenes and collateral and inspiration you gave it?

Joaquín Cuenca Abela: 8:51 You can do both. Like, as soon as you are typing the prompt, it can suggest, ‘Oh, do you want to connect it with this one?’ So you start connecting, ‘Okay, I want to use this image as a reference, I want to use these voices, I want to use…’ …and you can put it all together super easy. So this thing was done just by me one day, imagine a true creative team like three, four people like to their best. You usually get like somebody that is pretty good at photography, somebody that is good at audio, you put them together and now you can tell much more compelling stories than before.

Jason Calacanis: 9:21 What’s really interesting about what you’ve built is you’re starting to see a movement of people doing launch videos now. Startups are all doing launch videos. Uh, there are some brands uh that are getting a little cheeky on TikTok, Instagram, uh YouTube has the shorts feature now uh where there’s reels, I mean all these different short features. You did letterbox there so it looks cinematic, but I assume we could have flipped that and made it in the short feature. And what’s great about this is uh this would have been 150,000 dollars to make, you know, maybe 10k a sec— 10k a minute, 20k a minute depending on the level of person you hire. So you do a five minute launch video, could be 50k, could be 100k. Here, you know, what do you charge for this software? What does it cost to do the output? Um, and how do you charge for the product?

Joaquín Cuenca Abela: 10:13 Per second is around 10 cents nowadays. Uh, to be honest, like you usually take like, you do multiple tries to get the right one. So you should assume like there is like one dollar per second. That’s nothing. Like the real budget here is going to be your creative team spending their time and doing something great. It’s not going to be anymore on the platform.

Jason Calacanis: 10:34 The models, um, you’re building all these tools, but there are many people building models, open source models. We have frontier models, obviously Google’s got all their amazing models that are making great progress, Nano Banana and everything. So what do you use as the back end for this? Is it headless or are you actually building language models as well or image models as well?

Joaquín Cuenca Abela: 10:57 So we have some image models, like the upscaler is our own technology. We have some skin enhancer that is our own technology, but we use many other like third-party models. We use state-of-the-art obviously, many from Google but also from some other uh platforms. We try to have like state-of-the-art whatever it is in the platform so that it is really easy for the user to achieve the best of the best work that they can do.

Lon Harris: 11:22 Who are the other customers for this product? Obviously, you know, the first person that comes to mind here on This Week in Startups is startups who want to do launch videos, but has this made its way towards uh Hollywood creatives? I would think they would like to use this, uh you know if you were a writer on a television show like The Walking Dead or, you know, any of these hit shows, Better Call Saul. It kind of would be cool to do a little cinematography yourself and, you know, do the dialogue and actually be able to use those actors’ voices in the creative process. So the line between a writer and a storyboard artist, a set designer…

Jason Calacanis: 12:00 As a designer and a director and a cinematographer, I would assume those lines are blurring, yeah?

Joaquín Cuenca Abela: 12:04 Oh yes. Uh, the great thing about this is that it is raising the, the ceiling and it’s also raising the floor, both of them. So you have Hollywood studios, they are using it, it for example like, uh, the company that does uh, the House of David, they have used Freepik to make the production. It’s used by all the big studios in Hollywood nowadays and it’s used more and more. The growth there is exponential, really many, many users in Hollywood. But also many, many marketing departments. So that’s the other side of the equation. It’s, you get used by companies that are 200, 300 people doing a film, but also by small studios that are 3, 4 people making an ad or making a short film. Okay, it’s, it’s used in both sides and they are both getting great value out of it. The usage is slightly different. I want to clarify, like if you are doing a film, typically you still record actors and you still do like all of that work, but now you can do much more with that footage.

Jason Calacanis: 13:03 It seems like Hollywood went from just two years ago saying hey, you know, pardon my French, F these tools or like these things are evil, they’re bad, to now two years later they’re like wait a second. Uh, and I was literally, uh, at this Breakthrough Prize this weekend, it’s a science, the Oscars of science, and I was talking to Gal Gadot, she, uh, is very famous for playing Wonder Woman and a bunch of other things. And I asked her hey, what’s your take on all this AI stuff? She says it’s incredible. We are now able to make films for one-third the price. We can, you know, film ourselves on a green screen or a grey screen, whatever it happens to be, and then let everybody else do the rest of the work. But we just focus on our performance. We’ll be able to make two, three, four, five times as much content and we’ll be able to service different audiences. So you know, sometimes people get scared, oh, people are going to lose their jobs, but we’re also going to see a Cambrian explosion in creativity where, you know, somebody who’s a screenplay writer or they want to be an ad executive, they can’t break in, so they just start making spec work. And spec work’s got a little bit of a, you know, people are like oh, I don’t want to just do spec work, but people write spec scripts all the time. You could make a spec storyboard here, bring it to somebody, and you’re just proving your idea out bit by bit, a little bit more, yeah? And it’s just so exciting.

Joaquín Cuenca Abela: 14:27 Yeah, yeah, exactly. You can listen, you can take like some sacrifices and put still what looks like a great story, and it’s going to be like a great story, uh, for people to see or you can go full on, on a bigger production. Like I talked to some Hollywood directors and what they told me is that they had a project that it would have never been greenlighted because of the cost to make the project. But now as they managed to reduce the cost, they managed to get that film and employ 600 people doing the film, while otherwise… it would not have been enough budget to do it. So it’s actually creating more jobs and I think you’re right, like Hollywood is coming to terms with technology because now they see it more and more as an ally rather than something that is going to challenge their jobs or I mean anything.

Jason Calacanis: 15:14 Yeah. And you’ve based the company in on the coast of Southern Spain, am I correct?

Joaquín Cuenca Abela: 15:19 That is right. Yes.

Jason Calacanis: 15:21 What is the AI talent like? I know it’s big in Europe, right? Uh, you have Mistral there, you had a lot of great PhDs etc. But when you pitch people on coming to the south of Spain, incredible seafood, incredible tapas, great lifestyle obviously. What’s that like to recruit AI talent and have that as in your in your back pocket?

Joaquín Cuenca Abela: 15:46 Listen, it is not San Francisco, that’s for sure. Uh, but at the same time it it offers like a great quality of life. It’s also a great place for people to live. So when you pitch, you know, the whole value proposition is you’re gonna build something that is exciting, you are going to reshape how creativity works in marketing environments, how films are made, how people tell stories. Uh, and it’s in a great place, you know, with a great company. People really buy.

Lon Harris: 16:17 Man, yeah, just amazing and first class. When do you think the line between being able to execute really well on the visuals here, the aesthetic, but then creative storytelling, this seems to be a blocker for some people. You ask any large language model, hey make me a script or make me a storyline. It’s good at giving you brainstorming ideas I find, like when I’m writing a blog post or I’m going to be on a podcast I might say, hey I’m going to talk about this, give me some bullet points, some history or some analogies in history, right? I’ll ask for things that are analogous or maybe some other people’s take on this topic. And it will do a good job brainstorming, but it’s not doesn’t feel creative. It doesn’t feel tight, um, or visionary in any way. It’s almost like asking a college student to, you know, give you some ideas and it’s good at giving you some ideas and looking something up like a like an intern might or a researcher might, but it’s not fully creative. So do you think we’re going to hit that point where the creativity of these models starts to match the creativity of humans or do you think it’s going to be a tool creative humans use for some time to come?

Joaquín Cuenca Abela: 17:25 I think it will match the creativity of some humans. What I think is unique to us is that it’s our individuality. So in a way, you are you and you are a result of your experiences and what is inside your brain. Uh, me it’s it’s impossible to replicate you or to replicate me. Okay? We are all very, very like… And AI is very different because it’s it’s a digital artifact. So you can copy it. Okay, and humans you cannot copy it. So the thing that is important about us is that whatever we do it’s unique to us. And I I think that’s important. I’m putting that in your work, I think that’s important and that’s meaningful, okay? So I think that those tools, they are going to drive more opportunities for people that are good at telling stories and are good at putting their own history into something that is meaningful for others. So I think this is going to drive up actually more demand for people that can tell stories, for people that can put together like a good film or a good story.

Jason Calacanis: 18:28 The interesting thing for me is as the compute gets better, I could see making a hundred versions of the film you just made in some dynamic way and then using targeting to share them. So we did one dystopian San Francisco, but we could have done Los Angeles, we could have done Sydney, we could have done Galicia or Madrid, you know, any number of cities. And then people from that region could actually experience the same concept, ‘Oh my God, everything died in this post-apocalyptic world except the Cybertruck and the Cybertruck made it through with its solar panels and you should buy a Cybertruck to prepare for the apocalypse.’ Are people doing that yet where they say, ‘Hey, you know, pick the iconic bridge in that city, pick the iconic restaurant or baseball field, the main street, and then make me a hundred versions by city’? Then you could use something like the Google ad network to do hypertargeting, regional targeting, or even other preferences, you know. It could have people who are my age, right? Or it could be more inclusive in different ways. Could have… I could have a Spanish accent, I could have a Japanese… I could speak Japanese. Like, there’s so many possibilities here. Have you had customers try that yet?

Joaquín Cuenca Abela: 19:46 Yes, but for simpler things, not yet for video. So what is happening is that for the previous generation of ads, when you have something static, that’s something that you can try, that’s something that we do. You can localize like 50 different languages, change the food in the ad, change like things like that. And that works pretty well. If you try it with a full video like this one, the problem is that at every step you have a decent probability of error, decent probability of things going wrong. So at the end you take like one minute and it completely loses the track, okay? It diverges. It’s not yet… I think we still need, and maybe we’ll provide that, we still need kind of a cloud code for this. Something that harness the AI, checks the output and kind of steers it back.

Jason Calacanis: 20:40 Yes. This feels like you know we’re making progress when we start talking about the harness, the fact-checking, the validation level because what it means is the tools are actually ready for production and we need the fit and the finish and the polish. Let alone doing this in real-time. I mean I’m talking about making a hundred versions but eventually this will be like…

Lon Harris: 21:00 Like Minority Report, where each ad will mention you by name, you know, and it’s… they’re going to be served dynamically in real-time, you know, and it’s… it’s going to be pretty crazy, the future.

Joaquín Cuenca Abela: 21:13 But… but I think we’re going to get both. We’re going to get like some productions that are very, very bespoke to the viewer, but some of… some of them you want the vision of the director. It’s like when you go to a restaurant, okay? If they… if they tell you you can get any dish that you want, here you have all the ingredients. That’s too much. I want you… I want your opinion on it. So sometimes you want the opinion of the creative to drive the narrative. That’s also pretty… pretty tough.

Jason Calacanis: 21:41 Yes. All right, listen, Joaquín, thank you so much for educating us. Where can people download the app and find more information?

Joaquín Cuenca Abela: 21:46 Go to magnific.com.

Jason Calacanis: 21:47 Oh, well, there it is folks, it couldn’t be easier than that. And of course, check out Google Cloud’s brand new report, Future of AI: Perspectives on Generative Media for Startups. It’s a great report that you can get at goo.gle/futureofgenmedia. It’s in the description, it’s in the show notes. Joaquín, you’re famous, you’re also featured in the report, yes? They got you in the report.

Joaquín Cuenca Abela: 22:15 Yes. I want to thank Google for that. Thank you.

Jason Calacanis: 22:17 Yes. And all the episodes of Startup Basics from marketing to legal, to accounting, and now AI basics are available at thisweekinstartups.com/basics. This is our gift to the community. Get those basics, lock them in as a founder and just keep building your knowledge and we’ll see you next time. Bye bye.