Today, Applied Intuition is launching Dana, an agentic platform for developing physical AI applications. Co-founders Qasar Younis and Peter Ludwig join a16z's Marc Andreessen and Erik Torenberg to discuss their mission to put intelligence on one billion machines: 01:34 Physical AI vs digital AI 05:28 Humans as a design... constraint 10:30 Beyond robotaxis and humanoids 19:04 What killed Cruise 28:19 The chip maker playbook 35:26 When self-driving is the default 45:14 Nobody wants to drive trucks 49:54 Introducing Dana 1:00:54 "World model" means 100 things 1:05:28 GTA 6 may be the last of its kind 1:15:40 Fear of AI is a luxury Qasar Younis Applied Intuition Marc Andreessen 🇺🇸 Erik Torenbergshow more

Carries to trucking before it carries to humanoids. Fleet management and mining sit in the same platform pitch as ADAS here, and those are the buyers with a cost line already open for autonomy.

1:00:54 is the segment that matters. "World model" now spans pixel-space video gen and latent dynamics for control, and the two have opposite compute economics. Investors conflating them is how the wrong architectures keep getting funded.

definitely cool, and seems durable. dm me if you want to pilot labeled video for RL from licensed tradesmen working in the physical world. @camerasearch

Wow, the Applied Intuition founder @qasar had lots of insights here. Really fascinating idea of an app store for software meant that there are all sorts of apps, including things for specific businesses. And I like that example of imagining a robot that would, say, go around a given institution and provide services to the members one way or the other. I do think it would be nice if it could be more specific about what that might entail and how quickly it can overcome any given issues that come up. Now, I hear his prediction that self-driving cars will be fully available through all these different car manufacturers by 2030 or even by 2028. And so I think internalizing what capabilities of these robot models will be on the path towards understanding where the value creation opportunity is, I think would be really nice. I get that his view is that the physical world companies are going to make a lot more money than the digital world companies. And it makes sense to me that having robots doing stuff would be really powerful. There was the OpenAI post from last year about robots making data centers and so forth, which sounds really interesting. And so I imagine the fleet of software agents today is very useful. Now imagine a fleet of physical world robots. And I buy this idea that it's profound that someone can go ahead and use these tools to build really useful robot things, even someone who's in school. But I do think it could benefit from some more elaboration or even like a vision video, even if it's AI generated, of what the implications are. I also think it'd be cool to understand a bit more about how the data flows. What I'm hearing in this is that Applied Intuition gains understanding across data of all these different long-lived assets that can have some sensors and intelligence in them. I think I'd like to understand what does that path of outcomes look like in mentioning, say, the coordination of these various machines. Is that sequences of tasks? It's almost like a manufacturing line kind of consistency? Or is that utilization? Or is that all of the above? And I just would be really curious to see what that looks like in terms of optimizations that kind of get filled in over time, and what sort of optimal situations look like and saturation. Feels like intuition being applied in this way is going to be really profound. And so if the flywheel is incrementally better models, then that makes a lot of sense. I guess I come back to his point that self-driving will be solved in the coming years and available in all these different cars. Now, is that going to be one provider that happens to have all the data and they get a certain amount, or a couple providers and there's some reasonable amount of competition? I think it could benefit from some specificity here about the market dynamics and what the upside is for which types of companies. @grok, weigh in on how @qasar might respond?

🔥🔥

A billion intelligent machines and a billion hours of training