
10.3.23-Showcase-Osaka-Common_Sense_Machines

-
Video details
Startup Lightening Talk
-
Interactive transcript
DHRUPAD KARWA: Hi, everyone. My name is Dhru. Let me just make a quick correction. I'm not the co-founder and CEO. He couldn't make it, so asked me a week ago can you make it to Japan, and I said, yes, just about. So I'm the product lead at Commonsense Machines, also known as CSM.
And our company was founded by Tejas Kulkarni and Max Kleinman-Weiner, who were both computer science PhDs at MIT. And I joined CSM just over a month ago now. And before that, I was the co-founder and CEO of a company called Haiku Jam. Haiku Jam was a social poetry app where strangers from all around the world would write haiku poems together in collaboration.
The app grew to over 2 million users from across 150 countries and we exited to a leading mental health player in India last year. So on a personal note, because of my connection to Japanese haiku, and now with CSM, it's really a pleasure to be here in Japan with you for the first time.
So the problem that CSM is solving relates to creating 3D worlds. The demand for 3D content today is skyrocketing across so many different industries, from gaming to animation to simulation to prototyping. But making this type of content is really difficult. It takes a lot of expertise.
To produce a single 3D asset it often costs thousands of dollars and takes days, if not weeks, of multiple teams collaborating together. Now because of this complexity and prohibitive cost and time, it makes it difficult for companies to develop large and diverse 3D data sets. Take Shutterstock, for example, one of the largest content marketplaces out there. Shutterstock has less than a million 3D assets listed on its website, which is just nothing compared to the number of stock photos and videos that it has. So it's really high time that someone solved this problem.
Meet Cube by CSM, which is a beautiful AI product that lets anyone anywhere, regardless of your skill level, generate 3D assets, based on simple prompts like a line of text or a single image or a short video. Cube is available on across different platforms like mobile, web, through APIs, and also on Discord. Now I think the real power of Cube is just its sheer simplicity.
It works in three simple clicks. Click one, you upload a prompt, like this cool-looking sheep in a denim jacket and sunglasses. Click two, you hit the Generate button and let our AI do its thing. And then click three, you can export the 3D model and go and build your dream world.
Here's another example this time showing a video prompt taken from our mobile app, which Cube then transforms into this lifelike humanoid asset, which just looks ready to enter reality, right? The possibilities are endless. This could become a character in an action game.
It could become, it could feature in Marvel's next superhero franchise, or it could even be 3D printed and turned into a lifelike robot imbued with intelligence and common sense. So in terms of quality, currently CSM Cube is really state of the art. As you can see, our outputs are higher quality than anything else on the market today, including OpenAI's latest Shape-E model, which was trained on millions of proprietary data points.
Our team is super-small, focused, and determined to maintain this competitive advantage through improving not just the technology, but each and every aspect of Cube's user experience. So returning to the theme of robots, I wanted to share this recent use case, which came out of Japan. So a 3D artist in Tokyo discovered Cubed, and used it to generate some characters from single images.
He then animated these characters to music, using programs like Mixamo and Unity, and then shared these videos on social media. They went really viral, getting over like 60,000 views. And I wanted to share and play this one of this cute little robot he generated on Cube.
[MUSIC PLAYING]
So the artist was actually inspired to write up the entire workflow on his personal blog. And I feel community energy like this is not only inspiring, but it's often a leading indicator of great things to come, especially for consumer products.
So honestly it's difficult to believe that this product was launched pretty recently, just like two months ago, and has already scaled to over 50,000 users without a penny of marketing spend. And our cohorts in Korea and Japan are particularly exciting. They're growing organically 20% week over week, performing thousands of daily interactions on Cube.
And the average user in Japan and Korea generates 2.6 3D assets every single day. We're seeing applications across industries like gaming, animation, simulation, 3D prototyping, and printing, and we are ready with enterprise and studio plans as well as APIs to support these use cases and take these industries into a gorgeous 3D generated future. So let's partner and build new worlds together. Thank you so much.
-
Video details
Startup Lightening Talk
-
Interactive transcript
DHRUPAD KARWA: Hi, everyone. My name is Dhru. Let me just make a quick correction. I'm not the co-founder and CEO. He couldn't make it, so asked me a week ago can you make it to Japan, and I said, yes, just about. So I'm the product lead at Commonsense Machines, also known as CSM.
And our company was founded by Tejas Kulkarni and Max Kleinman-Weiner, who were both computer science PhDs at MIT. And I joined CSM just over a month ago now. And before that, I was the co-founder and CEO of a company called Haiku Jam. Haiku Jam was a social poetry app where strangers from all around the world would write haiku poems together in collaboration.
The app grew to over 2 million users from across 150 countries and we exited to a leading mental health player in India last year. So on a personal note, because of my connection to Japanese haiku, and now with CSM, it's really a pleasure to be here in Japan with you for the first time.
So the problem that CSM is solving relates to creating 3D worlds. The demand for 3D content today is skyrocketing across so many different industries, from gaming to animation to simulation to prototyping. But making this type of content is really difficult. It takes a lot of expertise.
To produce a single 3D asset it often costs thousands of dollars and takes days, if not weeks, of multiple teams collaborating together. Now because of this complexity and prohibitive cost and time, it makes it difficult for companies to develop large and diverse 3D data sets. Take Shutterstock, for example, one of the largest content marketplaces out there. Shutterstock has less than a million 3D assets listed on its website, which is just nothing compared to the number of stock photos and videos that it has. So it's really high time that someone solved this problem.
Meet Cube by CSM, which is a beautiful AI product that lets anyone anywhere, regardless of your skill level, generate 3D assets, based on simple prompts like a line of text or a single image or a short video. Cube is available on across different platforms like mobile, web, through APIs, and also on Discord. Now I think the real power of Cube is just its sheer simplicity.
It works in three simple clicks. Click one, you upload a prompt, like this cool-looking sheep in a denim jacket and sunglasses. Click two, you hit the Generate button and let our AI do its thing. And then click three, you can export the 3D model and go and build your dream world.
Here's another example this time showing a video prompt taken from our mobile app, which Cube then transforms into this lifelike humanoid asset, which just looks ready to enter reality, right? The possibilities are endless. This could become a character in an action game.
It could become, it could feature in Marvel's next superhero franchise, or it could even be 3D printed and turned into a lifelike robot imbued with intelligence and common sense. So in terms of quality, currently CSM Cube is really state of the art. As you can see, our outputs are higher quality than anything else on the market today, including OpenAI's latest Shape-E model, which was trained on millions of proprietary data points.
Our team is super-small, focused, and determined to maintain this competitive advantage through improving not just the technology, but each and every aspect of Cube's user experience. So returning to the theme of robots, I wanted to share this recent use case, which came out of Japan. So a 3D artist in Tokyo discovered Cubed, and used it to generate some characters from single images.
He then animated these characters to music, using programs like Mixamo and Unity, and then shared these videos on social media. They went really viral, getting over like 60,000 views. And I wanted to share and play this one of this cute little robot he generated on Cube.
[MUSIC PLAYING]
So the artist was actually inspired to write up the entire workflow on his personal blog. And I feel community energy like this is not only inspiring, but it's often a leading indicator of great things to come, especially for consumer products.
So honestly it's difficult to believe that this product was launched pretty recently, just like two months ago, and has already scaled to over 50,000 users without a penny of marketing spend. And our cohorts in Korea and Japan are particularly exciting. They're growing organically 20% week over week, performing thousands of daily interactions on Cube.
And the average user in Japan and Korea generates 2.6 3D assets every single day. We're seeing applications across industries like gaming, animation, simulation, 3D prototyping, and printing, and we are ready with enterprise and studio plans as well as APIs to support these use cases and take these industries into a gorgeous 3D generated future. So let's partner and build new worlds together. Thank you so much.