# Everything will be Simulated

Cristóbal Valenzuela · AI Engineer NYC 2026 · October 14

[Watch recording](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30551s)

[08:29:11](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30551s) Hey, guys.

[08:29:13](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30553s) Very excited to be here.

[08:29:14](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30554s) Thank you for staying for the last talk.

[08:29:17](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30557s) Welcome to my talk about the matrix and how we're living through it.

[08:29:24](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30564s) I have been working on Runway for almost a decade now.

[08:29:27](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30567s) Started here in New York, maybe around 10 years ago.

[08:29:31](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30571s) I want to work through some of the learnings we have seen through building the company,

[08:29:35](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30575s) building this research lab here in New York, and where do we see a lot of the work that

[08:29:40](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30580s) we do in video, in pixel generation, really moving forward.

[08:29:43](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30583s) So I don't exaggerate and I joke that this might be a simulation or not.

[08:29:49](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30589s) Maybe for perspective, I will walk you through a piece of history.

[08:29:53](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30593s) This is like watching prehistoric ancient ruins.

[08:29:58](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30598s) is our first AI film we

[08:30:00](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30600s) ever made. It was done in 2019. It's called The Orange because, as you can tell very clearly, it's an orange that falls from the tree and walks through different parts of the world. This is done, this is a video model, an image model that we had back then that was creating pixels in 128 pixels by 128 pixels that we then forcefully up-sample to this kind of resolution.

[08:30:24](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30624s) I think of this as art now.

[08:30:26](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30626s) This is like looking at a film from the 1900s.

[08:30:29](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30629s) You know, it's beautiful.

[08:30:30](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30630s) There's a lot of history behind it.

[08:30:32](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30632s) But this is 2019.

[08:30:34](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30634s) We started the company in 2018, 2017.

[08:30:37](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30637s) And so we're trying to figure out if this had any lags.

[08:30:40](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30640s) We're trying to convince ourselves that if you scale this,

[08:30:43](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30643s) it will naturally improve to the point where we can get somewhere close to this.

[08:30:48](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30648s) This is a state-of-the-art video today.

[08:30:50](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30650s) This is all generated.

[08:30:52](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30652s) None of these characters are real.

[08:30:53](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30653s) None of those locations are real.

[08:30:55](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30655s) I'm sure if you're probably here, you know this.

[08:30:58](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30658s) By now, you probably have seen this online, on Twitter.

[08:31:01](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30661s) The models are just incredibly good.

[08:31:03](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30663s) And the most interesting thing is that when I look at stuff like this, I still believe

[08:31:08](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30668s) this will look like the orange in just a couple more years.

[08:31:11](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30671s) So I want to walk you through how we're thinking not only through what we're doing at Runway

[08:31:16](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30676s) today but also in the next couple of years and the research that's shaping that.

[08:31:22](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30682s) We are a company that's at the intersection of both art and science.

[08:31:26](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30686s) We have deep roots in artistic and creative communities, and we have some of the best researchers in the world working on diffusion models and large video models.

[08:31:35](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30695s) And I like that combination because it allows us to bring the best of both worlds.

[08:31:41](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30701s) We're a team of scientists and artists thinking about the problem of scaling video and world models.

[08:31:47](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30707s) but in a way the core idea of the company was already discovered by cinema and by moving pictures

[08:31:52](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30712s) 120 130 years ago which is if you think about video and maybe you remember this from uh from

[08:31:59](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30719s) high school video is an optical illusion video is the illusion of movement that if you do if you

[08:32:04](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30724s) stitch frames fast enough you will create movement you will create the illusion of movement you

[08:32:08](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30728s) create motion pictures and so at run we were doing kind of something very similar we're taking that

[08:32:13](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30733s) idea to the next level, how do we do next frame prediction fast enough, scale enough,

[08:32:20](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30740s) in such a way that we can hopefully do more than just pretty and nice creative videos.

[08:32:26](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30746s) And so, Aloto is powered by, of course, scaling laws.

[08:32:31](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30751s) I think we believed in scaling laws not only applied for language models but also for video.

[08:32:37](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30757s) And this has been true.

[08:32:38](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30758s) We're now really ramping up the amount of compute that we put behind the models that

[08:32:43](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30763s) you already have seen.

[08:32:44](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30764s) And I think the interesting thing about seeing the models scale over time is the emerging

[08:32:49](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30769s) properties that appear, the intelligence that appears when you start taking this very simple

[08:32:54](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30774s) idea that cinema found itself figuring out 120 years ago, which is next frame prediction,

[08:33:01](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30781s) and taking next frame prediction really to the next level.

[08:33:03](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30783s) So if I take an image like this, which is not real, those two people are not real, and

[08:33:07](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30787s) And I just keep on predicting what the next frame will look like based on the previous

[08:33:11](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30791s) frames.

[08:33:12](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30792s) And if I do it fast enough, and if I do it with a network that's big enough, I'll get

[08:33:17](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30797s) movement.

[08:33:19](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30799s) I'll get delusion of movement.

[08:33:20](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30800s) I'll get something that looks so real to us that we humans can distinguish.

[08:33:24](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30804s) I look at that video and I've been doing this for 10 years.

[08:33:27](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30807s) I can't tell you if that's real or not.

[08:33:30](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30810s) And we measure that.

[08:33:31](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30811s) We've actually measured and tried to quantify if people can distinguish between real and

[08:33:36](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30816s) I think that Turing test was already passed.

[08:33:39](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30819s) People are not able to distinguish.

[08:33:40](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30820s) But the most interesting thing is here is that really what's happening as you predict

[08:33:44](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30824s) next frames is you're learning a lot more than just pretty frames.

[08:33:48](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30828s) You're learning about physics.

[08:33:50](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30830s) You're learning about object permanence.

[08:33:51](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30831s) You're learning about gravity.

[08:33:52](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30832s) You're learning about occlusion.

[08:33:54](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30834s) You're learning about movements.

[08:33:55](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30835s) You're learning about how the objects work in the real world.

[08:33:59](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30839s) And you're doing that all through just video, through images, through next frame prediction.

[08:34:04](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30844s) So if you look at a complex thing like this, a good video model for it to understand how

[08:34:11](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30851s) to make this a consistent video will need to naturally understand what all these subjects

[08:34:17](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30857s) are, how each of them relate to one another, how a camel might probably stand on a ball

[08:34:22](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30862s) like that, what water in a container looks like.

[08:34:25](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30865s) And if I do all of that together and I do it fast enough, I'll get, hopefully, a consistent

[08:34:29](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30869s) video that looks real to us.

[08:34:31](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30871s) So, a lot of the benchmarks that we use to evaluate our video models are not only creative,

[08:34:39](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30879s) it basically means do you like it or do you don't, but are actually based on physics,

[08:34:45](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30885s) on geometries, on how objects behave in the real world and how those objects behave, and

[08:34:50](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30890s) also simulator or the real world.

[08:34:53](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30893s) So, you take that idea of next-frame prediction and you do it fast enough and you get, of

[08:34:57](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30897s) And of course, now something that looks more like what we should expect from a video application

[08:35:01](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30901s) or a game application.

[08:35:02](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30902s) Again, this is just video, right?

[08:35:04](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30904s) And you take that idea to a next level.

[08:35:07](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30907s) What happens if we start this idea of next-round prediction, but do it in the context of a 3D

[08:35:12](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30912s) environment or something that looks similar to a game?

[08:35:14](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30914s) You can start actually getting closer to a physics engine, a simulation system.

[08:35:18](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30918s) But the simulation system has no physics laws.

[08:35:21](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30921s) We didn't describe what the behavior of the object should be.

[08:35:25](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30925s) The model just knows how to do that by poor force of scale.

[08:35:32](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30932s) And this gets kind of interesting because you can only think about playing worlds.

[08:35:36](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30936s) This is not GTA.

[08:35:38](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30938s) This is someone playing someone else in an environment and a software that doesn't access.

[08:35:45](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30945s) It's all simulation of the simulation, which gets really meta, right?

[08:35:49](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30949s) And so I'm not kidding when I said welcome to my metric simulation.

[08:35:53](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30953s) But the most interesting thing for me is that everything you saw before was not real-time,

[08:35:59](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30959s) that it took a couple of minutes or seconds for those videos to be made.

[08:36:03](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30963s) A lot of what we're seeing right now, when we think about the next three, four years

[08:36:07](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30967s) of video models and world models, is to deeply think about what happens when you start not

[08:36:14](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30974s) only scaling the models and making them much, much, much bigger, with much more emerging

[08:36:20](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30980s) properties but also what happens when you start making them interactive.

[08:36:23](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30983s) So take the same idea of next frame prediction, you take a frame, you predict the next frame,

[08:36:27](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30987s) then X, then X, then X, but then it can act at any point in time in that frame prediction,

[08:36:32](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30992s) which means I can simulate any possible divergence or action in that world.

[08:36:37](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=30997s) So in this world, I could play the video, but I can also come and touch every part,

[08:36:43](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31003s) any aspect of that image, and make it interactive to affect the consequences of that world.

[08:36:50](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31010s) And it works.

[08:36:53](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31013s) And of course, this is incredibly useful for a variety of different use cases.

[08:36:58](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31018s) As you can imagine, we're now spinning an entire robotics lab in Runway.

[08:37:03](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31023s) So if you come to our office in New York, half of it is what looks like a Hollywood

[08:37:07](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31027s) studio, half of it is a bunch of robots walking through and moving through the office.

[08:37:11](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31031s) And so we're able to make these models or start making these models the brains of how

[08:37:17](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31037s) autonomous and embodiment systems walk and reason about the world.

[08:37:21](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31041s) But there's a couple of more interesting use cases that I think go beyond the too obvious

[08:37:26](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31046s) one, which is video and robotics, which is what happens when you think about anything,

[08:37:31](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31051s) anything, any pixel on a screen as a video model.

[08:37:35](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31055s) This is a great example.

[08:37:37](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31057s) This is a video model.

[08:37:38](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31058s) What you see right now, well, this is an image, first of all.

[08:37:40](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31060s) That image could be real or not, but I could run one of our video models in real time and

[08:37:45](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31065s) make this an interactive application.

[08:37:46](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31066s) So let's say I want to turn the lights of the car.

[08:37:50](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31070s) Maybe I want to turn the car around.

[08:37:52](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31072s) I can just click through it.

[08:37:54](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31074s) I can turn it around again, maybe open the hood and see what's inside the car.

[08:37:58](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31078s) Now, everything you just saw here has no physical 3D representation.

[08:38:03](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31083s) There's no Unity.

[08:38:05](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31085s) There's no Unreal.

[08:38:06](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31086s) There's no Blender.

[08:38:07](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31087s) It's just a video model.

[08:38:09](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31089s) A video model has been trained to understand actions and clicks in an interface.

[08:38:14](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31094s) We call this a world interface model.

[08:38:19](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31099s) And the world interface model, I feel you guys understand now, it's an interface and

[08:38:24](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31104s) it's a world model, world interface model, allows you to do very cool stuff like this.

[08:38:28](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31108s) And so here's another application, another example of how you could think about using

[08:38:33](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31113s) a model like this in real time.

[08:38:35](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31115s) I can take an image, a representation, move objects around and expect them to move as

[08:38:41](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31121s) I was actually in that room.

[08:38:44](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31124s) Again, remember, there's no 3D representation of this.

[08:38:48](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31128s) This environment might not actually have been modeled before.

[08:38:51](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31131s) It's just an image and a model that understands how objects move in a space because of this

[08:38:55](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31135s) very simple idea that next frame prediction scale well enough can become a world model

[08:39:00](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31140s) or a simulation engine.

[08:39:03](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31143s) So with the eight minutes I have left, I want to show you this idea taken to another extreme

[08:39:11](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31151s) we're thinking not only about what happens if you can control the school applications and the school

[08:39:15](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31155s) interfaces but what happens if you keep extrapolating and i think by the way extrapolating

[08:39:22](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31162s) may be one of the most important skills these days in ai in company building what happens over the

[08:39:27](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31167s) next six months 12 months 18 months may be one of the most important things to to try to predict

[08:39:32](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31172s) understand um and so we ask ourselves that question a lot at runway what happens what will

[08:39:38](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31178s) happen if we keep scaling the models, if we keep adding more data, if we keep making them

[08:39:42](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31182s) better simulation systems, if we make them universal simulation systems that can predict

[08:39:46](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31186s) anything on a screen, literally anything on a screen.

[08:39:49](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31189s) So we've been working on this idea that we call Continuum.

[08:39:52](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31192s) And Continuum is the idea of real-time video models or role models taken really to

[08:40:00](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31200s) the next degree. This is an operating system. The idea here is that you might open your computer at some point in the near future and what you see streaming to your screen might be coming from a video model.

[08:40:10](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31210s) So I will show you next specific how this idea could shape. This is still very early research, but I want to show you this idea of combining video models as part of your operating system, as part of your daily work.

[08:40:24](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31224s) So here's an example.

[08:40:25](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31225s) I was researching San Francisco and the fog in San Francisco.

[08:40:30](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31230s) I wanted to learn more about when and how it happens.

[08:40:33](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31233s) This is structural data.

[08:40:34](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31234s) This is actually a website.

[08:40:35](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31235s) This is something you can take from the Internet.

[08:40:39](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31239s) And now Continuum has this idea that I can select any part of content in that website

[08:40:43](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31243s) and make it a portal.

[08:40:44](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31244s) We call this a portal.

[08:40:46](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31246s) And the portal is a real-time video.

[08:40:47](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31247s) It's fully interactive and fully based on the content you're experiencing

[08:40:51](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31251s) experiencing reading through, I probably won't read through all of that text, it's too much

[08:40:55](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31255s) text for me, so I just want to make a video and I can make a video that's represented

[08:40:58](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31258s) by the text you see there.

[08:41:00](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31260s) These portals can be 3D realizations, it can be any content that you're watching or interacting

[08:41:04](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31264s) with.

[08:41:05](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31265s) On the right, you can have all the visualizations and portals you've experienced before, anything

[08:41:09](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31269s) you've been trying to learn.

[08:41:11](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31271s) Basically, it's a memory, it's the disk of the imaginations or the visualizations you've

[08:41:16](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31276s) seen before.

[08:41:16](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31276s) So here I was researching about batteries.

[08:41:18](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31278s) And so I can go back to the portal I had around batteries before.

[08:41:24](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31284s) And since there are real-time interactive applications, I can click them anywhere.

[08:41:27](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31287s) I can move them around.

[08:41:29](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31289s) I can start asking questions, and the videos and the content itself will change based on

[08:41:34](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31294s) my interactions and the way I'm managing through it.

[08:41:41](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31301s) Here's another very cool example.

[08:41:42](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31302s) that the first time I saw it, when the team showed it to me, it blew my mind.

[08:41:46](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31306s) Since it's a real-time video, you can make it full interactive with anything.

[08:41:50](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31310s) Not only with clicks, but resizing the application should also change the behavior.

[08:41:55](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31315s) So I can resize.

[08:41:56](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31316s) I'm learning about photosynthesis here.

[08:41:58](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31318s) If I resize the application, I get a different experience.

[08:42:02](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31322s) And this is, again, full interactive application, so I can click through it, navigate through

[08:42:07](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31327s) parts of the cell or how photosynthesis works to get a better understanding of what I'm trying

[08:42:11](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31331s) to learn and do.

[08:42:14](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31334s) The one thing that still feels to me kind of unsigned about this is I don't know how

[08:42:18](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31338s) to call this.

[08:42:19](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31339s) I don't know if this is ‑‑ it doesn't feel like a visualization, a video, a game,

[08:42:24](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31344s) something in between.

[08:42:25](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31345s) And I think we need to do the collective work of figuring out how we call these kind of

[08:42:28](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31348s) interactions in the first place.

[08:42:30](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31350s) And I'm sure you hear the other thing is that you can change the structure of the prompt

[08:42:34](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31354s) or what's actually creating that interaction in the first place.

[08:42:40](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31360s) Here is another example. I'm going to type something. And this is one of the ones that

[08:42:50](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31370s) I feel has been the most under explored yet. We call this visual thinking. So we've connected

[08:42:57](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31377s) Continuum to a coding agent. And what the coding agent can do is it can use Continuum to visualize

[08:43:03](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31383s) through a problem. So imagine you ask your coding agent to create a 3D model for your bike,

[08:43:09](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31389s) and instead of the model trying to just think through that via code, the model can spin a

[08:43:13](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31393s) version of continuum to think through the problem visually. So if you ever think about someone,

[08:43:18](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31398s) I'm sure you have friends or maybe a lot of people here may be visual thinkers, what a visual

[08:43:23](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31403s) thinker is, is it thinks through problems and through ideas visually first. And that helps you

[08:43:29](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31409s) ground and project some of the best things you want to do.

[08:43:33](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31413s) This is somehow very similar.

[08:43:35](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31415s) Agents have the ability to do visual thinking.

[08:43:38](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31418s) Agents can spin up a sort of imagination that you can tap and interact with.

[08:43:44](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31424s) So not only the agent can interact with it, I can also see it myself.

[08:43:48](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31428s) And I can guide the agent in much better ways than I might have been able to guide it just

[08:43:53](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31433s) via text alone.

[08:43:54](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31434s) Which is for us a very powerful idea.

[08:43:56](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31436s) It's a completely new paradigm.

[08:44:00](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31440s) visual thinking. The last example I have is how you think about taking Continuum to become

[08:44:12](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31452s) much more than just using or navigating interactive structural content in the way I've showed

[08:44:19](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31459s) you before, but actually making them fully live experiences that are reactive. So in

[08:44:26](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31466s) And as Continuum understands the intention of what I'm trying to search from, it also

[08:44:32](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31472s) suggests interfaces that are specifically designed in real time just for that.

[08:44:37](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31477s) So here it created for me this Mars outpost that actually should be accurate with the

[08:44:42](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31482s) data about the actual robots that are in Mars right now or the ones that are projected maybe

[08:44:47](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31487s) in 2030.

[08:44:48](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31488s) And since this is fully interactive, I can interact with it.

[08:44:50](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31490s) I can go to places.

[08:44:51](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31491s) I can move around in the same way that I might move around in a 3D system or a 3D continuum.

[08:44:59](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31499s) And it works.

[08:45:01](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31501s) Just fascinating.

[08:45:06](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31506s) So when I say everything will be simulated, I'm really trying to understand or help not

[08:45:13](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31513s) really our team but the world get to the point where a lot of what we'll see today will found

[08:45:18](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31518s) obvious hopefully in a few more years or months.

[08:45:20](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31520s) We'll look back at Continuum

[08:45:22](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31522s) and it might feel very close

[08:45:23](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31523s) to the video of the Orance that we just showed you.

[08:45:26](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31526s) Continuum and a lot of the things we're doing

[08:45:27](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31527s) in video and in runway when it comes

[08:45:30](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31530s) to simulation are still in very early days,

[08:45:32](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31532s) but we're getting to a point where it actually

[08:45:34](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31534s) works and it really scales.

[08:45:35](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31535s) So what we're really excited about is that for the next

[08:45:37](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31537s) 8 to 12 months, we're releasing most of what

[08:45:39](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31539s) I showed here for everyone to use.

[08:45:41](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31541s) Thank you, guys.

[08:45:43](https://www.youtube.com/watch?v=jiUC6IUMdvc&t=31543s) Thank