The Next Medium: Why Real-Time Interactive Video Changes Everything — Ahmed Ahres, Reactor

>> Hi everyone. Welcome to the talk. First >> Hi everyone. Welcome to the talk. First of all, thank you all for making the of all, thank you all for making the of all, thank you all for making the time. I know it's the last talk of the time. I know it's the last talk of the time. I know it's the last talk of the day probably or I think the last one is day probably or I think the last one is day probably or I think the last one is at 3:45 but yeah, thank you all for your at 3:45 but yeah, thank you all for your at 3:45 but yeah, thank you all for your time. I know you're all probably very time. I know you're all probably very time. I know you're all probably very busy. busy. busy. >> [snorts] >> [snorts] >> [snorts] >> Today I'm going to be talking about >> Today I'm going to be talking about >> Today I'm going to be talking about something that is a little bit slightly something that is a little bit slightly something that is a little bit slightly futuristic though not for San Francisco futuristic though not for San Francisco futuristic though not for San Francisco and that's world models. and that's world models. and that's world models. I know here it's written real time I know here it's written real time I know here it's written real time interactive video but the way we think interactive video but the way we think interactive video but the way we think about world models is really in the real about world models is really in the real about world models is really in the real time interactive video and I'll explain time interactive video and I'll explain time interactive video and I'll explain why. why. why. And in today's world I think world And in today's world I think world And in today's world I think world models is a little bit of a marketing models is a little bit of a marketing models is a little bit of a marketing term that people think about it from a term that people think about it from a term that people think about it from a Gaussian splatting standpoint others Gaussian splatting standpoint others Gaussian splatting standpoint others from video from video from video but the way we define world models is but the way we define world models is but the way we define world models is really real time interactive video and I really real time interactive video and I really real time interactive video and I have strong evidence or like beliefs have strong evidence or like beliefs have strong evidence or like beliefs that this will actually change that this will actually change that this will actually change everything in how we produce and consume everything in how we produce and consume everything in how we produce and consume content. content. content. Oops. Oops. Oops. What happened? What happened? What happened? Sorry. Cool. Cool. >> [clears throat] >> [clears throat] >> [clears throat] >> So if you think about video, video has >> So if you think about video, video has >> So if you think about video, video has always been something passive. always been something passive.

always been something passive. Before people used to produce videos and Before people used to produce videos and Before people used to produce videos and movies and then we would watch it. movies and then we would watch it. movies and then we would watch it. And in today's world we have these And in today's world we have these And in today's world we have these models the VO3, the C dance 2 and what models the VO3, the C dance 2 and what models the VO3, the C dance 2 and what they do is you prompt them, you get back they do is you prompt them, you get back they do is you prompt them, you get back a file, you watch and good luck. It's a a file, you watch and good luck. It's a a file, you watch and good luck. It's a slot machine. You cannot change it. You slot machine. You cannot change it. You slot machine. You cannot change it. You cannot do anything about it. cannot do anything about it. cannot do anything about it. So there's a question that we like to So there's a question that we like to So there's a question that we like to think about in our company is what think about in our company is what think about in our company is what happens when video becomes programmable happens when video becomes programmable happens when video becomes programmable like software. like software. like software. And what happens when pixels can be And what happens when pixels can be And what happens when pixels can be generated in real time? generated in real time? generated in real time? Once this happens it actually changes Once this happens it actually changes Once this happens it actually changes completely how we think about consuming completely how we think about consuming completely how we think about consuming content and even producing content and content and even producing content and content and even producing content and I'll be talking about how in history I'll be talking about how in history I'll be talking about how in history we've seen that real time has always we've seen that real time has always we've seen that real time has always been the future and the kind of been the future and the kind of been the future and the kind of applications that it unfolded and how applications that it unfolded and how applications that it unfolded and how you can get started today. Quick Quick background here. Um so, my Quick Quick background here. Um so, my name is Ahmed. I'm the head of name is Ahmed. I'm the head of name is Ahmed. I'm the head of go-to-market at Reactor. My background go-to-market at Reactor. My background go-to-market at Reactor. My background is in computer vision and machine is in computer vision and machine is in computer vision and machine learning. Um and I state machine learning. Um and I state machine learning. Um and I state machine learning because this is the time where learning because this is the time where learning because this is the time where we used to actually train our models. we used to actually train our models. we used to actually train our models. Um I built and shipped games on iOS uh Um I built and shipped games on iOS uh Um I built and shipped games on iOS uh on iOS and Android for fun. Um I was a on iOS and Android for fun. Um I was a on iOS and Android for fun. Um I was a founder um and uh today I'm the head of founder um and uh today I'm the head of founder um and uh today I'm the head of go-to-market at Reactor.

go-to-market at Reactor. go-to-market at Reactor. And who we are quickly is we're a series And who we are quickly is we're a series And who we are quickly is we're a series A company building the platform for A company building the platform for A company building the platform for these real-time world models. these real-time world models. these real-time world models. So, so far the main most of them have So, so far the main most of them have So, so far the main most of them have been still models, but we are building been still models, but we are building been still models, but we are building the infrastructure and the developer the infrastructure and the developer the infrastructure and the developer platform to make them usable so that platform to make them usable so that platform to make them usable so that anyone can integrate these real-time anyone can integrate these real-time anyone can integrate these real-time interactive video, and we can interactive video, and we can interactive video, and we can democratize access to this technology. I'll start with a problem. We can today I'll start with a problem. We can today generate pretty much anything, but we generate pretty much anything, but we generate pretty much anything, but we just can't change it, right? A generated just can't change it, right? A generated just can't change it, right? A generated video, as I mentioned earlier, is still video, as I mentioned earlier, is still video, as I mentioned earlier, is still a recording. You get it back, you can't a recording. You get it back, you can't a recording. You get it back, you can't do anything about it. do anything about it. do anything about it. And real time changes what the medium And real time changes what the medium And real time changes what the medium is. It doesn't just make it faster. is. It doesn't just make it faster. is. It doesn't just make it faster. And I'm going to talk about two examples And I'm going to talk about two examples And I'm going to talk about two examples that actually show us what real time has that actually show us what real time has that actually show us what real time has unlocked in the past. unlocked in the past. unlocked in the past. Before, in the 1950s and even before, we Before, in the 1950s and even before, we Before, in the 1950s and even before, we used to look at a map to know where we used to look at a map to know where we used to look at a map to know where we are. are. are. Someone produces a map, you look at Someone produces a map, you look at Someone produces a map, you look at where you are, and that's it. You cannot where you are, and that's it. You cannot where you are, and that's it. You cannot do anything about it. do anything about it. do anything about it. Then GPS came. GPS made it real time. Then GPS came. GPS made it real time. Then GPS came. GPS made it real time. Suddenly, I can know where where I am at Suddenly, I can know where where I am at Suddenly, I can know where where I am at the instant that I can, you know, track the instant that I can, you know, track the instant that I can, you know, track it. it. it. Now, you'd think GPS has just made it a Now, you'd think GPS has just made it a Now, you'd think GPS has just made it a bit faster to know where I am, but bit faster to know where I am, but bit faster to know where I am, but actually actually actually Uber would not exist if we did not have Uber would not exist if we did not have Uber would not exist if we did not have the GPS.

the GPS. the GPS. Another example, which is even a bit Another example, which is even a bit Another example, which is even a bit more powerful, I'd say. more powerful, I'd say. more powerful, I'd say. Before we used to do to use the film to Before we used to do to use the film to Before we used to do to use the film to produce content, right? We had a film, produce content, right? We had a film, produce content, right? We had a film, someone shoots something, someone shoots something, someone shoots something, they can't see what they're shooting, they can't see what they're shooting, they can't see what they're shooting, they go somewhere, they produce that they go somewhere, they produce that they go somewhere, they produce that film, and then you can see it. film, and then you can see it. film, and then you can see it. And then it became digital. And then it became digital. And then it became digital. You can start seeing what you're You can start seeing what you're You can start seeing what you're shooting. If today you pick up your shooting. If today you pick up your shooting. If today you pick up your iPhone and you start iPhone and you start iPhone and you start recording a video, then you can see recording a video, then you can see recording a video, then you can see what's going on and that's why we can what's going on and that's why we can what's going on and that's why we can produce high-quality content. It's produce high-quality content. It's produce high-quality content. It's because you are able to see what's going because you are able to see what's going because you are able to see what's going on in the screen and adapt accordingly. on in the screen and adapt accordingly. on in the screen and adapt accordingly. That gave rise to Instagram and TikTok. That gave rise to Instagram and TikTok. That gave rise to Instagram and TikTok. Instagram and TikTok would not exist if Instagram and TikTok would not exist if Instagram and TikTok would not exist if we could not produce high-quality we could not produce high-quality we could not produce high-quality content and the only reason why we're content and the only reason why we're content and the only reason why we're able to produce high-quality content able to produce high-quality content able to produce high-quality content among among among many reasons is because we can see in many reasons is because we can see in many reasons is because we can see in real time what's happening. It's not a real time what's happening. It's not a real time what's happening. It's not a slot machine. We can actually just edit slot machine. We can actually just edit slot machine. We can actually just edit and see and see and see and that's what unlocks all of these new and that's what unlocks all of these new and that's what unlocks all of these new use cases. So, when video becomes programmable, you So, when video becomes programmable, you can address it, you can condition it, can address it, you can condition it, can address it, you can condition it, you can change it. you can change it. you can change it. I can show you on the screen whatever I I can show you on the screen whatever I I can show you on the screen whatever I want to show you on the screen and it want to show you on the screen and it want to show you on the screen and it becomes programmable a little bit like becomes programmable a little bit like becomes programmable a little bit like software or like anything that is software or like anything that is software or like anything that is programmable in the world.

And in And in in the market today, we are seeing three in the market today, we are seeing three in the market today, we are seeing three kinds of models that do this. Some of kinds of models that do this. Some of kinds of models that do this. Some of them you will be familiar with, others them you will be familiar with, others them you will be familiar with, others you will not be familiar with. you will not be familiar with. you will not be familiar with. The first one is these are Vio, think The first one is these are Vio, think The first one is these are Vio, think about Vio or Sora, but real-time and about Vio or Sora, but real-time and about Vio or Sora, but real-time and interactive. interactive. interactive. Meaning that first of all, they're Meaning that first of all, they're Meaning that first of all, they're infinite. So, they don't stop after 5, infinite. So, they don't stop after 5, infinite. So, they don't stop after 5, 10, or 30 seconds. They actually 10, or 30 seconds. They actually 10, or 30 seconds. They actually continue forever. continue forever. continue forever. They're interactive, meaning you can They're interactive, meaning you can They're interactive, meaning you can change what's happening on the screen. change what's happening on the screen. change what's happening on the screen. And they're in real time. So, you don't And they're in real time. So, you don't And they're in real time. So, you don't need to wait to see what's going on. need to wait to see what's going on. need to wait to see what's going on. And assuming this works, so this is an And assuming this works, so this is an And assuming this works, so this is an example of a of a video that I passed an example of a of a video that I passed an example of a of a video that I passed an image with a dog and this was all image with a dog and this was all image with a dog and this was all generated in real time. And at some generated in real time. And at some generated in real time. And at some point, I'm going to prompt a cat shows point, I'm going to prompt a cat shows point, I'm going to prompt a cat shows up and you will see that a cat showed up up and you will see that a cat showed up up and you will see that a cat showed up in the in the video. This would not be in the in the video. This would not be in the in the video. This would not be possible in the existing batch regular possible in the existing batch regular possible in the existing batch regular video generation models because you video generation models because you video generation models because you would get back the video and you cannot would get back the video and you cannot would get back the video and you cannot do anything about it. I could have added do anything about it. I could have added do anything about it. I could have added anything. I could have gone on to create anything. I could have gone on to create anything. I could have gone on to create an entire story with it. I could have an entire story with it. I could have an entire story with it. I could have said the cat the dog starts running, said the cat the dog starts running, said the cat the dog starts running, starts jumping, a dragon shows up. It starts jumping, a dragon shows up. It starts jumping, a dragon shows up. It goes to, I don't know, to the World Cup. goes to, I don't know, to the World Cup. goes to, I don't know, to the World Cup. All of this would have happened in front All of this would have happened in front All of this would have happened in front of you.

These type These types of models like These type These types of models like unlock a few things. unlock a few things. unlock a few things. First of all, control. First of all, control. First of all, control. So, if you think about generative media So, if you think about generative media So, if you think about generative media today, the big problem that content today, the big problem that content today, the big problem that content creators all have is I don't have the creators all have is I don't have the creators all have is I don't have the control I need. Like, yes, it's great to control I need. Like, yes, it's great to control I need. Like, yes, it's great to use CDNS to review 03 to generate use CDNS to review 03 to generate use CDNS to review 03 to generate videos, but I just don't have the videos, but I just don't have the videos, but I just don't have the control. And this is always the thing control. And this is always the thing control. And this is always the thing that any filmmaker or movie producer or that any filmmaker or movie producer or that any filmmaker or movie producer or any content creator will tell you. any content creator will tell you. any content creator will tell you. And so, real-time actually ends the slot And so, real-time actually ends the slot And so, real-time actually ends the slot machine type of mentality and actually machine type of mentality and actually machine type of mentality and actually gives you the the control that you need. gives you the the control that you need. gives you the the control that you need. And a big thing I like to say is And a big thing I like to say is And a big thing I like to say is instant feedback is the ultimate level instant feedback is the ultimate level instant feedback is the ultimate level of control, and you we will never be of control, and you we will never be of control, and you we will never be able to have this level of control if we able to have this level of control if we able to have this level of control if we don't have real-time. don't have real-time. don't have real-time. The second thing it unlocks, which, you The second thing it unlocks, which, you The second thing it unlocks, which, you know, among other things, which is a know, among other things, which is a know, among other things, which is a field I'm not particularly fond of, but field I'm not particularly fond of, but field I'm not particularly fond of, but I think it's going to be big, is I think it's going to be big, is I think it's going to be big, is advertising. advertising. advertising. If I can know what you looked for a If I can know what you looked for a If I can know what you looked for a minute ago, why can't I insert the logo minute ago, why can't I insert the logo minute ago, why can't I insert the logo of whatever you've been looking for? Why of whatever you've been looking for? Why of whatever you've been looking for? Why can't I produce an ad in real-time in can't I produce an ad in real-time in can't I produce an ad in real-time in front of you? We don't need to produce front of you? We don't need to produce front of you? We don't need to produce pre-produce anything. Now, granted, this pre-produce anything. Now, granted, this pre-produce anything. Now, granted, this is going to take some time because, you is going to take some time because, you is going to take some time because, you know, brands afraid of AI, afraid of you know, brands afraid of AI, afraid of you know, brands afraid of AI, afraid of you know, if their logo has one pixel that know, if their logo has one pixel that know, if their logo has one pixel that is white instead of dark.

is white instead of dark. is white instead of dark. But it will happen eventually. And so, But it will happen eventually. And so, But it will happen eventually. And so, and I think at the moment that happens, and I think at the moment that happens, and I think at the moment that happens, we will not be to need to produce any we will not be to need to produce any we will not be to need to produce any ads anymore. Everything will be ads anymore. Everything will be ads anymore. Everything will be happening in front of you in real-time. happening in front of you in real-time. happening in front of you in real-time. The second type of model is the one that The second type of model is the one that The second type of model is the one that probably you're most familiar with. It's probably you're most familiar with. It's probably you're most familiar with. It's the Genie 3 like from Google. the Genie 3 like from Google. the Genie 3 like from Google. These are models where you can pass an These are models where you can pass an These are models where you can pass an image image image and a text typically, and you can and a text typically, and you can and a text typically, and you can control a character. control a character. control a character. They're fun. The first thing that you They're fun. The first thing that you They're fun. The first thing that you think about when you think about this is think about when you think about this is think about when you think about this is games, right? It's a character, it's a games, right? It's a character, it's a games, right? It's a character, it's a world, you can generate anything. world, you can generate anything. world, you can generate anything. But actually, But actually, But actually, it actually goes way far way beyond it actually goes way far way beyond it actually goes way far way beyond games. games. games. It creates entire new interactive It creates entire new interactive It creates entire new interactive experiences where combined with the experiences where combined with the experiences where combined with the first types of models that I talked first types of models that I talked first types of models that I talked about, we've already been seeing people about, we've already been seeing people about, we've already been seeing people in our community building a mix of games in our community building a mix of games in our community building a mix of games and and movies. If you've ever watched and and movies. If you've ever watched and and movies. If you've ever watched Bandersnatch from Netflix, which is the Bandersnatch from Netflix, which is the Bandersnatch from Netflix, which is the game the movie that you can pick your game the movie that you can pick your game the movie that you can pick your next scene, next scene, next scene, this is one of those things that gets this is one of those things that gets this is one of those things that gets that that that becomes possible that you that that that becomes possible that you that that that becomes possible that you can control a character, you can control can control a character, you can control can control a character, you can control what's happening, and create entire new what's happening, and create entire new what's happening, and create entire new types of interactive experiences that types of interactive experiences that types of interactive experiences that were not possible.

were not possible. were not possible. The second thing is robotics. The second thing is robotics. The second thing is robotics. So, because you can simulate and you can So, because you can simulate and you can So, because you can simulate and you can control, you can actually create as much control, you can actually create as much control, you can actually create as much training data as as you want. And training data as as you want. And training data as as you want. And robotics world models in robotics is robotics world models in robotics is robotics world models in robotics is actually a ginormous market today. actually a ginormous market today. actually a ginormous market today. I cannot tell you the number of robotics I cannot tell you the number of robotics I cannot tell you the number of robotics labs that are training and building labs that are training and building labs that are training and building these models. But, because you can these models. But, because you can these models. But, because you can control whatever you want to control in control whatever you want to control in control whatever you want to control in any environment, this creates a new any environment, this creates a new any environment, this creates a new opportunity to generate infinite amount opportunity to generate infinite amount opportunity to generate infinite amount of data for robotics. of data for robotics. of data for robotics. And finally, something I like to think And finally, something I like to think And finally, something I like to think about, this is more maybe a passionate about, this is more maybe a passionate about, this is more maybe a passionate thing that I have, is education. Um thing that I have, is education. Um thing that I have, is education. Um because you can step into anything, who because you can step into anything, who because you can step into anything, who with today's world in AI, I don't with today's world in AI, I don't with today's world in AI, I don't actually believe that in the future of actually believe that in the future of actually believe that in the future of education is LLM based or textbook education is LLM based or textbook education is LLM based or textbook based. If you can put any any kid based. If you can put any any kid based. If you can put any any kid in the situation, for example, in a in the situation, for example, in a in the situation, for example, in a history lesson, history lesson, history lesson, that enables entire new types of that enables entire new types of that enables entire new types of experiences that can be educational. The third type of model is probably the The third type of model is probably the type of model that, you know, is more, type of model that, you know, is more, type of model that, you know, is more, let's say, let's say, let's say, something that we've been seeing before, something that we've been seeing before, something that we've been seeing before, which is avatars, but live and which is avatars, but live and which is avatars, but live and interactive. The thing with avatars, interactive. The thing with avatars, interactive. The thing with avatars, though, is it hasn't actually been though, is it hasn't actually been though, is it hasn't actually been cracked. They're still all kind of cracked. They're still all kind of cracked. They're still all kind of weird. If you speak to an avatar in any weird. If you speak to an avatar in any weird. If you speak to an avatar in any customer support or anything, it's still customer support or anything, it's still customer support or anything, it's still kind of off, right?

kind of off, right? kind of off, right? And these types of models and we're And these types of models and we're And these types of models and we're seeing a rise of these live and seeing a rise of these live and seeing a rise of these live and interactive avatar models in research interactive avatar models in research interactive avatar models in research preview that combined with model one and preview that combined with model one and preview that combined with model one and model two is actually going to be, I model two is actually going to be, I model two is actually going to be, I believe, a believe, a believe, a a big change in what we've been seeing a big change in what we've been seeing a big change in what we've been seeing so far. And this will be applied to so far. And this will be applied to so far. And this will be applied to things like customer support, training, things like customer support, training, things like customer support, training, sales, gaming, streaming services, etc. And so, just give you a glimpse of what And so, just give you a glimpse of what our users are building today at Reactor our users are building today at Reactor our users are building today at Reactor with these types of models. Some of them with these types of models. Some of them with these types of models. Some of them are building interactive live stream. are building interactive live stream. are building interactive live stream. Right? A live stream where people are Right? A live stream where people are Right? A live stream where people are watching and then the users can type watching and then the users can type watching and then the users can type what happens next and then they vote. what happens next and then they vote. what happens next and then they vote. Why? Because pixels can be generated in Why? Because pixels can be generated in Why? Because pixels can be generated in real time. So, there's no reason why I real time. So, there's no reason why I real time. So, there's no reason why I cannot put a live stream on X, YouTube, cannot put a live stream on X, YouTube, cannot put a live stream on X, YouTube, or Twitch and enable users to pick what or Twitch and enable users to pick what or Twitch and enable users to pick what happens next. Something that was surprising to me is a Something that was surprising to me is a little bit on on the medical simulation. little bit on on the medical simulation. little bit on on the medical simulation. So, we've seen users create applications So, we've seen users create applications So, we've seen users create applications where you generate a world and then you where you generate a world and then you where you generate a world and then you simulate what happens next. What if I simulate what happens next. What if I simulate what happens next. What if I put this medicine? What if I remove this put this medicine? What if I remove this put this medicine? What if I remove this medicine? Right? And this creates can be medicine? Right? And this creates can be medicine? Right? And this creates can be a training playground for people wanting a training playground for people wanting a training playground for people wanting to become doctors.

to become doctors. to become doctors. The third one, which is also kind of The third one, which is also kind of The third one, which is also kind of surprising, is cooking simulation. So, surprising, is cooking simulation. So, surprising, is cooking simulation. So, people are building applications where people are building applications where people are building applications where you can simulate cooking and what you can simulate cooking and what you can simulate cooking and what happens if you put this ingredient. happens if you put this ingredient. happens if you put this ingredient. And finally, video editing. And finally, video editing. And finally, video editing. With video to video models, video With video to video models, video With video to video models, video editing becomes very interesting because editing becomes very interesting because editing becomes very interesting because I'm able to just add visual effects in I'm able to just add visual effects in I'm able to just add visual effects in real time. And we've seen people build real time. And we've seen people build real time. And we've seen people build entire video editing platforms. Now, entire video editing platforms. Now, entire video editing platforms. Now, granted, they're not very good yet just granted, they're not very good yet just granted, they're not very good yet just because of the quality of the models. because of the quality of the models. because of the quality of the models. But it's a new paradigm when you're able But it's a new paradigm when you're able But it's a new paradigm when you're able to edit videos just by a prompting or by to edit videos just by a prompting or by to edit videos just by a prompting or by talking to it or by clicking. And now, for the final part, how do you And now, for the final part, how do you actually do all of this? And this is why actually do all of this? And this is why actually do all of this? And this is why I like to say the world behind an API. At At Reactor, we have four types of At At Reactor, we have four types of models today. First one is called models today. First one is called models today. First one is called Helios, which is the interactive video Helios, which is the interactive video Helios, which is the interactive video model that I talked about. This one is model that I talked about. This one is model that I talked about. This one is from ByteDance. from ByteDance. from ByteDance. Link bot, which is a world model like Link bot, which is a world model like Link bot, which is a world model like Genie 3, trained by Alibaba. Genie 3, trained by Alibaba. Genie 3, trained by Alibaba. Long live 2 from Nvidia, Long live 2 from Nvidia, Long live 2 from Nvidia, which is multi-shot film. You can prompt which is multi-shot film. You can prompt which is multi-shot film. You can prompt things in advance and create a things in advance and create a things in advance and create a consistent story over time.

consistent story over time. consistent story over time. And sound streaming, also from Nvidia, And sound streaming, also from Nvidia, And sound streaming, also from Nvidia, which is a model that does which is a model that does which is a model that does video-to-video editing. So, people are video-to-video editing. So, people are video-to-video editing. So, people are already using this, for example, by already using this, for example, by already using this, for example, by shooting something or creating something shooting something or creating something shooting something or creating something on C dense 2, uploading it, and then on C dense 2, uploading it, and then on C dense 2, uploading it, and then adding visual effects, removing people, adding visual effects, removing people, adding visual effects, removing people, adding background. And this gets very adding background. And this gets very adding background. And this gets very interesting in previous realization, for interesting in previous realization, for interesting in previous realization, for example, for Hollywood movies. And under the hood, when we talk about And under the hood, when we talk about infrastructure, infrastructure, infrastructure, the thing that the thing that the thing that I think I like to drive home is building I think I like to drive home is building I think I like to drive home is building infrastructure for regular video infrastructure for regular video infrastructure for regular video generation models is very different from generation models is very different from generation models is very different from real time. Because in regular video real time. Because in regular video real time. Because in regular video generation models, you're talking about generation models, you're talking about generation models, you're talking about requests. You just send a request, a job requests. You just send a request, a job requests. You just send a request, a job gets run in the cloud, and I'm gets run in the cloud, and I'm gets run in the cloud, and I'm oversimplifying here, but you know, a oversimplifying here, but you know, a oversimplifying here, but you know, a job runs in the cloud, and it gives you job runs in the cloud, and it gives you job runs in the cloud, and it gives you back a file. back a file. back a file. With real time, it's a different With real time, it's a different With real time, it's a different ballgame. Uh you cannot just take what ballgame. Uh you cannot just take what ballgame. Uh you cannot just take what works for batch inference and apply to works for batch inference and apply to works for batch inference and apply to real time inference. real time inference. real time inference. For example, you need to think about For example, you need to think about For example, you need to think about streaming, right? Once you need to think streaming, right? Once you need to think streaming, right? Once you need to think about streaming, about streaming, about streaming, um pixels uh from a server to the um pixels uh from a server to the um pixels uh from a server to the client, um it adds entire new entire client, um it adds entire new entire client, um it adds entire new entire complexities that batch generation does complexities that batch generation does complexities that batch generation does not have to think about.

not have to think about. not have to think about. The second one is that everything is a The second one is that everything is a The second one is that everything is a live session. So, everything runs live session. So, everything runs live session. So, everything runs constantly, and there is memory to be constantly, and there is memory to be constantly, and there is memory to be kept into account. kept into account. kept into account. Now, granted, one of the things that Now, granted, one of the things that Now, granted, one of the things that live live real-time models struggle with live live real-time models struggle with live live real-time models struggle with is memory. If you've seen demos from is memory. If you've seen demos from is memory. If you've seen demos from Genie 3, for example, we've all seen Genie 3, for example, we've all seen Genie 3, for example, we've all seen that the character can look back and that the character can look back and that the character can look back and then will not remember what what's going then will not remember what what's going then will not remember what what's going on. So, there is a lot of work that on. So, there is a lot of work that on. So, there is a lot of work that needs to be going into maintaining that needs to be going into maintaining that needs to be going into maintaining that context window, so that you can remember context window, so that you can remember context window, so that you can remember what happened if you turned your what happened if you turned your what happened if you turned your character left and right. character left and right. character left and right. And finally, global scale. And finally, global scale. And finally, global scale. If you If you think about real time, it If you If you think about real time, it If you If you think about real time, it needs to be sub-100 millisecond latency needs to be sub-100 millisecond latency needs to be sub-100 millisecond latency anywhere you are. anywhere you are. anywhere you are. And if you're deploying applications in And if you're deploying applications in And if you're deploying applications in the world, then someone based in India the world, then someone based in India the world, then someone based in India or someone based in Japan should be or someone based in Japan should be or someone based in Japan should be routed to a GPU that is based in India routed to a GPU that is based in India routed to a GPU that is based in India or Japan, or as close as possible to it. or Japan, or as close as possible to it. or Japan, or as close as possible to it. If not, if you don't have the compute If not, if you don't have the compute If not, if you don't have the compute worldwide, then the experiences are not worldwide, then the experiences are not worldwide, then the experiences are not real time anymore and it breaks real time anymore and it breaks real time anymore and it breaks completely the medium. And with Reactor, this is as easy as it And with Reactor, this is as easy as it gets to integrate these real time gets to integrate these real time gets to integrate these real time months.

months. months. Okay, I kind of maybe over simplified it Okay, I kind of maybe over simplified it Okay, I kind of maybe over simplified it a little bit here, but it really is a little bit here, but it really is a little bit here, but it really is maybe 10 lines of code and we have docs maybe 10 lines of code and we have docs maybe 10 lines of code and we have docs to do all of this, but essentially to do all of this, but essentially to do all of this, but essentially you can just load the model with an API you can just load the model with an API you can just load the model with an API key and just start integrating it in key and just start integrating it in key and just start integrating it in whatever video image plugin, anything whatever video image plugin, anything whatever video image plugin, anything that you're building. that you're building. that you're building. And if you want to get started, there's And if you want to get started, there's And if you want to get started, there's a QR code there and I've added a promo a QR code there and I've added a promo a QR code there and I've added a promo code AIE2026 which will give you $75 code AIE2026 which will give you $75 code AIE2026 which will give you $75 off, well, $75 worth of credits which is off, well, $75 worth of credits which is off, well, $75 worth of credits which is um um um significant amount of compute in our significant amount of compute in our significant amount of compute in our case because we make the models case because we make the models case because we make the models extremely cheap. extremely cheap. extremely cheap. Thank you very much. >> [applause] >> [applause] >> I'm happy to take questions if anybody >> I'm happy to take questions if anybody >> I'm happy to take questions if anybody has a question. Oop. Oop. Hey. Hey. Hey. >> So, with all the 16 FPS was >> So, with all the 16 FPS was >> So, with all the 16 FPS was uh stated frames per second what would it take to get that what would it take to get that to the normal 30 FPS? to the normal 30 FPS? to the normal 30 FPS? >> Uh >> Uh >> Uh well, one thing that actually we do is well, one thing that actually we do is well, one thing that actually we do is multi GPUs. So, we use multiple GPUs um multi GPUs. So, we use multiple GPUs um multi GPUs. So, we use multiple GPUs um up to optimizing the model weights, up to optimizing the model weights, up to optimizing the model weights, applying quantization techniques. So, applying quantization techniques. So, applying quantization techniques. So, there are ways, it's just a matter of there are ways, it's just a matter of there are ways, it's just a matter of priorities, but it is there are ways priorities, but it is there are ways priorities, but it is there are ways around it.

around it. around it. Yeah. Hey. Yeah. Hey. Yeah. >> I'm I'm >> I'm I'm >> I'm I'm wondering if you're wondering if you're wondering if you're going to be showcasing something at IBC >> Sorry? >> Sorry? >> Are you aware of what the IBC is? >> Are you aware of what the IBC is? >> Are you aware of what the IBC is? >> Uh no. >> Uh no. >> Uh no. >> The International Broadcasting >> The International Broadcasting >> The International Broadcasting Convention. Convention. Convention. >> Okay. >> Okay. >> Okay. >> So, >> So, >> So, >> I was wondering if you're going to be >> I was wondering if you're going to be >> I was wondering if you're going to be presenting some of your case studies presenting some of your case studies presenting some of your case studies with our stuff and the uh industry and with our stuff and the uh industry and with our stuff and the uh industry and >> I wasn't, but now I'm going to look at >> I wasn't, but now I'm going to look at >> I wasn't, but now I'm going to look at it. Yeah. it. Yeah. it. Yeah. >> [laughter] >> [laughter] Yeah. Yeah. >> Yeah. How do you feel like about letting >> Yeah. How do you feel like about letting >> Yeah. How do you feel like about letting uh some deterministic engines >> Yeah. >> Yeah. >> And then you get like a feedback maybe >> And then you get like a feedback maybe >> And then you get like a feedback maybe >> Yeah. >> Yeah. >> Yeah. >> Did you guys experiment with that? Can >> Did you guys experiment with that? Can >> Did you guys experiment with that? Can you share something? you share something? you share something? >> When you say deterministic engines, what >> When you say deterministic engines, what >> When you say deterministic engines, what do you mean exactly? do you mean exactly? do you mean exactly? >> Oh, any kind of like deterministic rule >> Oh, any kind of like deterministic rule >> Oh, any kind of like deterministic rule sets that you check against maybe some sets that you check against maybe some sets that you check against maybe some of the simulation space of the simulation space of the simulation space uh random >> Hm. So, we don't do any of that today.

>> Hm. So, we don't do any of that today. Um Um Um the reason why also we build a developer the reason why also we build a developer the reason why also we build a developer platform and but we've seen people build platform and but we've seen people build platform and but we've seen people build that on top of us. that on top of us. that on top of us. Um Um Um building the infra for this is already a building the infra for this is already a building the infra for this is already a lot of work. Um and I think we we are lot of work. Um and I think we we are lot of work. Um and I think we we are seeing already developers building and seeing already developers building and seeing already developers building and then open sourcing and then we may use then open sourcing and then we may use then open sourcing and then we may use it. it. it. Um but so the community is doing it for Um but so the community is doing it for Um but so the community is doing it for us, which is even better. us, which is even better. us, which is even better. Yeah. Yeah. Yeah. Sorry, question? Yeah. Sorry, question? Yeah. Sorry, question? Yeah. >> How do you >> How do you >> How do you measure the algorithm consistency? >> Uh you're asking a question that the >> Uh you're asking a question that the entire research community in world entire research community in world entire research community in world models has not answered. Your evals for models has not answered. Your evals for models has not answered. Your evals for real time and consistency is uh and real time and consistency is uh and real time and consistency is uh and fidelity Well, fidelity is easy. It's fidelity Well, fidelity is easy. It's fidelity Well, fidelity is easy. It's just like pixels, right? But um just like pixels, right? But um just like pixels, right? But um evaluation for these real-time models is evaluation for these real-time models is evaluation for these real-time models is an unsolved problem. an unsolved problem. an unsolved problem. So, today it's literally just look at it So, today it's literally just look at it So, today it's literally just look at it and and and human judgment. human judgment. human judgment. That's what it is today. That's what it is today. That's what it is today. And this is including, by the way, Deep And this is including, by the way, Deep And this is including, by the way, Deep Deep Mind and everything. Nobody has Deep Mind and everything. Nobody has Deep Mind and everything. Nobody has solved this problem yet. Uh we're working on it. We have a Uh we're working on it. We have a research team. Yes. research team. Yes. research team. Yes. >> [laughter] >> Awesome. Cool. Well, thank you, >> Awesome. Cool. Well, thank you, everyone. Thanks for your time.