44: A new way for creatives to use AI

I'm so happy that we have the founder of Flora AI on, because this is, I think, a real move from being just a chatbot where you're texting with an AI to really getting a new kind of user experience, a new kind of user interface that lets you put it all together. So let's dive into it. Who are you?

Hey, I'm Weber. I'm the founder of Flora. Have a background in poetry, venture capital, new media, art. And we started Flora to solve our own problem, have more creative control over AI, build creative systems, put everything in one place. So it came out of a place of a labor of love. And we've been working on it for a year and now excited to kind of do a quick demo to show you guys what it looks like.

Yeah, let's get right into the demo because I think it's mind blowing what you guys are building. And it really shows that a new kind of user experience, new kind of user interface as possible with AI and we don't have to just use our voice or type to the AI and then we can use it to create new things in a whole new way. So can you share your screen with me? And let's dive in.

Cool. Let's get started.

Yeah.

So the basics of Flora is that we have every creative AI model in just three blocks. Text, image and video. If you connect these together, you can get access to all the best creative AI models to walk through a single node. I'll walk you through the anatomy of a node. Here you can select as many generations as you want. And this is basically text to image if not connected to anything else. Here we have all the open source models. Here you have the style models. We're adding the ability to train your own soon as well as all the aspect ratios.

First of all, you said all the models, you have Clang and mid Journey, all these different models in here.

Yeah, we have everything that has API access. We put it all in one place. Image to video, video to video, text to video, image to image. Everything you would need.

This is really a tool for somebody who's trying to build visual things with all these new video and picture models that are doing some amazing things lately. Like Luma and whatever.

Yep, exactly. We're trying to build the best possible interface where you can interface with all these models at once across text, image and video. And really quickly concept from your initial idea to end video state or end image state. So we're currently used by folks like, you know, visual design agencies like Pentagram, people making films, music videos, Large agencies as well.

And how much does this cost per month for somebody to use?

So you can get started for free, and we give you some free credit so you can go ahead and get started there. And then our lowest tier is 16, and then we have another tier of $48. Both of those come with a certain number of credits.

Okay. And since you're bringing a new kind of user experience, new kind of user interface where you have to do, you know, do things, how long does it really take you to learn? Like, okay, I'm a Photoshop user, an Illustrator user, and I'm getting into this new AI world. How long does it take to learn to wire it all up?

Probably like five minutes to get started. Mastering it is different. But all you have to know is that you just connect one block into another to transform media. So I'll give you an example of that here.

Okay.

If we take an image node and then we pull this out, you'll see we have text, image and video. If you connect it to any one of these, we'll find the right model for you to turn video into video, image into image, depending on what you have connected here. So for instance, if I connect this to video, this is now image to video, and you have every single model here. So you can go ahead and generate many versions of that. If you do something like text, let's connect this out and choose text Here you can actually describe the image using an image prompt, or it'll look at the image and give you the actual prompt that you can use to go and generate more images. Then if you connect this to image, this is all of a sudden just image to image. You can change how much you keep the structure of this initial image while changing the style by prompting it.

Could I have it like chatgpt, write me a script for a movie or something like that, and then turn that into a board and then bring that into here so that I can start working on the visuals for my little commercial or my AI film or my AI music video or something like that.

Oh, definitely. Yeah. So we'll start with text. And this is where a lot of people that are using floor will start with concepting. We have GPT, Claude and also Google's model here. So we can say write me the high level storyboard for a short video script about a flower turning into a tree. And then we can go ahead and generate that so you don't have to do this in GPT or cloud anymore. And then you can connect this out into text and what this will do is essentially look at what's coming out of here and then from there you can prompt it even further to get more and more detail.

So, so this really speeds up your workflows in building stories and videos and images and things for a book or for a movie script or an AI music video, something like that. I know AI music videos are popular with a certain community here that I.

Follow on X. Yeah, yeah, definitely. So here, for instance, you have a few examples. You can ask for a detailed image prompt for option one. You can also do another one for the second scene of that music video or idea you have. And if you connect it to image now, you can go directly into image and then you can also animate it directly into video too. So you can really quickly go from your idea to the visual output that you're looking for.

It's really mind blowing. So thank you for being on with me. We need tools like this to tell stories in this new modern way. It's. It's hard to keep track of all the pieces and you're talking to chat GPT a lot and then you're having to go to Mid Journey or, or Luma. You know, it's confusing for people. They don't know how to get started, you know, and they don't know what's available. And here you can try a lot of things real quick.

Yeah, no, definitely. That's kind of the idea. We want to build a very intuitive creative tool that's also really creatively powerful and also lets you build repeatable workflows. So I'll show you an example of some examples of how this is being used today. So some examples here. We have some example workflows here that you can really easily get started with, broken out into different types of things. I'll show you an example of how folks at Pentagram are using this. They're a really great design agency. In the past, they would go into Adobe illustrator, make 100 variations of a logo by hand, choose one, and then go from there. Now you can just generate hundreds of variations at once. Pick the one you want and then go and refine that. So instead of the metaphor they told us of how it makes them feel is they were like in the past, it felt like we were trying to carve a marble sculpture with just a chisel. You've given us a pickaxe. You can move away the big chunks and then go and refine from there. And of course, we're a very professionally focused creative tool. We want you to be able to do creative work with Your team. So you can say, you can leave comments here like Figma and say something like, oh, let's go with this one. And if you click Share Copy Link, you can send it to anyone and they'll be in this project with you in real time multiplayer. So you can really, really quickly iterate, live, generate concepts together and so on and so forth.

Got it. My friends at Nexart just came out with a new model for digital photography. I don't think you've hooked that up yet. Can I hook that up and use my digital photography model that they created, which they call Marco or Icon?

Yeah, we'll look into that actually. Yeah, we, we carefully do.

Do you have to hook them up or can I hook them up? Yeah, that's the.

Yeah, we, we typically look at all the best tools coming out and fit the best ones into our framework of text, image and video. So yeah, thanks for flagging that. We'll look into that and see if that's something that we would add here.

Yeah, I put up a video with the founders there. Where I'm going is this world of AI is continually changing. There's new models, there's new things. I have listed 6,100 AI companies in just my holodeck list, which is a list of companies for creative people. Like what you're trying to get to. There's like 700 and something companies. Right. There's a lot of change. So I'm wondering how flexible your tool is and bringing new things or new models into this.

Yeah. Now we also have a similar list. When we decide, okay, what are we adding in? What do folks want? We find that there's two approaches here. One, you can let people add in whatever they want here. Two is we curate, we lean more towards the curation side just because for a lot of creative professionals, they don't want to keep up with all of it either. So we kind of do that hard work for them, figure out what's important to put in and curate that a little bit. But in general, of course, we have a lot of models in here already that we kind of let people use.

No, that's cool. Yeah.

And I'll give one other example here to show how you can go from your idea all the way to end output really, really quick. This is an example of someone making a short film completely in flora. What they're doing here is they've drawn this really rough sketch here, set it into image to text to analyze what's going on here and then use that to go and generate the Specific storyboard scenes in a very, very high fidelity. And then from there, this is another variation of that. They can feed it into image to video. And then now all of a sudden, you've gone from text to video really, really quickly. So you can do that in like, you know, 15, 20 minutes and stuff like that. And if you have a workflow you really like, what you can actually do is, let's say we don't like the kind of first iteration of it. You can select all of this, hit, run all, and it'll run everything at once. So you can really quickly get an iteration. Although I would recommend doing that sparingly, given that would would charge a lot of credits.

That's really cool. So. So you can bring your own photos, your own text, your own videos into this and then talk to it about it.

Yep. You could even like Control C, control V, stuff directly on the canvas, like Figma.

Oh, that's awesome. Can I talk to it and say, you know, here's my creative vision, Maybe even you here and write a creative vision for something I want to do. You know, I want to make a little movie of a neon world with this kind of character and this kind of setting. Could I bring that in here and then talk to what you're doing?

Yeah, that's actually something on the roadmap that we think would be really cool, where if you can just talk directly to it and then you see it just spawning stuff out for you, that would be a crazy experience. So, yeah, that's something that we're definitely thinking about for sure.

That'd be cool. Certainly. If I ask it, how could I stylize a photo? I have a photo here, bring that in and then say, I want to stylize this into a video with this kind of attribute. Make it look like it was hand drawn with neon wires or something like that. Could it tell me what I should be doing or does it just do it?

Yeah, I mean, you can do that today by adding stuff here, using image to text to describe it, and then from there you can go off and branch and say, hey, I like the aesthetic style, but could you make it of something else? And it'll give you the prompt for that and then you can go generate with that. We are also building a feature that we're not ready to announce yet, but it'll give you control from a single interface, everything that's happening on the canvas, and we're adding a voice mode to that. So you'll be able to talk to Flora and then have Flora kind Of.

So you're going to be able to vibe design, right?

Yeah, yeah, yeah, yeah, yeah. It's funny because like in theory designing is already vibes, but we can make it more vibes for sure. There's definitely ways to do that.

Andre Karpathy made that popular term in the coding world. You can just talk to your cursor, right? And it creates code for you. Do you hook in the cursor and start maybe making some user elements, some icons, some buttons and stuff that you might bring into cursor to help you with your programming and your user interfaces?

Yeah, no, that'd be really interesting. At the moment we're not focused as much on functional pixels. I would say there's two types of things that you can generate. One is the whole idea to design the code market with Figma is probably working on something internally to do that. There's bolts and lovable stuff like that. And then there's the non functional pixel world, the pretty pixel world where the pixels themselves are just kind of different colors. We're more focused on that world for now. But yeah, down the line that might be something we look into given generating layouts and stuff is also really useful.

Can I tell it what my end thing is going to be? You know, I'm creating an invite for a wedding, you know. Well, that's going to be very different than creating an AI music video or creating a picture for putting about my product on Amazon or something like that, right?

Yeah, no, definitely. I think that's the reason why we focused a lot on making that text note experience great. Because a lot of what you make really great with the visuals with image and video comes from the idea being good first. So as a result, that's kind of what we have, GPT quad in there. So you can really quickly concept and kind of generate the right ideas for those.

Very cool. How did you get into this? Because this is crazy.

Yeah, it was an interesting journey. So my background was initially in poetry and then I realized I have to make money for a living. I actually went into invest.

There's no money in poetry, is there? Until you turn it into a music video.

Yeah, yeah. So I went the opposite route and went into investment banking and the theory there was like I'm going to work at a top investment bank, get a job at a top venture firm, invest in the best startups and then do my own. So that's what I did. Worked at top investment bank, worked at this venture firm called Menlo Ventures. That's a really great firm. Love the People there. I did realize after some time there though, that, you know, I wouldn't back myself because I didn't have any specifically unique insight about the world and I had no experience or whatnot. So I decided that I needed to go pursue my passion. So I left my job at Menlo and then moved to New York. The idea was to work at a coffee shop, have a quarter life crisis, get into the art world, which is what I was passionate about, and figure out what I wanted to do from there. Right before I left though, the firm looked at investing in RunwayML and the text video company. And that was super interesting because they were innovating with technology in the art world. And honestly, I didn't know you could use technology to make art. I thought you could only use technology to make money. So I learned more about the founders and they came from this program called nyuitp where they use technology to make art. So when I moved to New York, I got into the whole art and tech scene and I met the head of I.T.P. the program. And I think you saw a lot of potential in me. And the folks there did as well. And they gave me a scholarship and let me join right away. And I thought, hey, like, you know, this is what's interesting to me. I don't have any other plans right now. Let's go to art school. So I went to art school, learned how to code and started building installation art. And I can quickly show you a little bit of what that looks like as well. But I was interested in this idea of, of two things. So one is I went to the Met and I love Picasso, but I realized that his paintings haven't moved in like 100 years. On the other hand, if I go to MoMA, the most popular artwork in the Museum of Modern Art is just a mirror because no one understands the rest of it, but they love looking at themselves. So I thought, why don't I just combine the two? And I ended up building using AI to actually build this. So this is a real time AI mirror. Basically it just takes the real world as input via a webcam, runs it through an image to image model, 20 frames a second, and then outputs the result. And people started finding this pretty interesting. I started building a tool along with my future co founder at the time, Alex. We started working on a tool to make it easier to do this for ourselves and others. And it was very simple. The node based software back then was just webcam real time AI and then output just hosted on the web and people started using it and my API bill started racking up. So we thought about just starting an agency using that tool to automate our stack so we could speed up that process a lot and trying to scale from there. And then we realized, like, hey, if it's something that we want, I bet a lot of other people want it too. So instead we did a startup where we're focused on building the software itself and then using that to power all the other creative teams. And so we've been working on it for the past year. And yeah, that's kind of how it started. Overall, along the way, we've done some other art projects, like to launch the startup. We actually did a performance artwork where I put a GoPro on my head and livestream my life for two weeks in real time AI. So we had a live. I was inspired by that. And actually along the way, oh, yeah, this is me with the GoPro. I actually met Justin Khan randomly, like, right before I started. And I asked him, like, yo, how was it, like, doing the live stream? Because I'm about to do it, like for 12 hours a day and for two weeks, and I'm kind of like worried it's going to disrupt my life. And he was like, dude, I did it for six months, 24 7. Like, two weeks is gonna be fine. And I was like, oh, okay. And yeah, it was pretty fun. We got a lot of traction from it and that's how we kind of launched a startup and it was really fun.

So I was actually on his show when, in the early days when he was doing that, he came over to my office while he was live streaming. That's crazy.

Yeah. I don't know how he did it for six months. It's crazy.

It's a little insane, you know, but insanity leads to companies, right?

Yeah, it appears so. It appears so.

You know, you're wearing an A16Z gaming hat, and gaming is about to go through, I think, a metamorphosis, a real paradigm shift. What's your vision, you know, for your company? You know, are you thinking about a 3D world that you're going to be helping people create for? And how do you fit in with the video game world? Because they're a little separate, right? A video game is different than creating an AI music video.

Yeah. A lot to talk about here. And the reason why I got back by a 16z game, speedrun, is I was actually talking to the GP of the games fund, John Lai, just casually, on a creative basis of like, what is the future of interactive media?

Right.

And My art was more towards that lens and in some ways floor down the line may get towards that. I would say, like my framework is like our program. ITP means Interactive Telecommunications Program. What we do there is interactive art. That means that it responds to you as a user. If you think about it, video games are the ultimate form of that. Right. I have this x, y axis framework like this. Where on the one hand it's like how interactive is an artwork, on the other hand is how generative is an artwork. I think the future of media is here. And the reason why is because generative techniques go very well with interactive techniques. Because if you're changing something every second, how do you make sure that new thing is good? I think it's usually AI generated right away. And then you have this really, really interesting dynamic experience. You're starting to see really interesting things here. I had a primitive version of it with the live stream of my art. We were seeing crazy stuff like deck art. These AI generated video games. A lot of interesting stuff that's in stealth right now that I'm aware of. That's really, really cool. And I think it'll end up being a mix of how much of it is structurally generated using LLMs to write C code in the same way Unity is built on top of it, and how much of it is like diffusion based of animating stuff on top. I think that the vision of the future that's probably most likely, at least in the creative world, is a combination of the two, where you're using LLMs to auto generate the structure and the skeleton and then your AI generating with diffusion on top of that. That seems to me like the most, I don't want to say computationally efficient, but the highest control way with the least amount of resources. That takes the benefits of both, like a very bottoms up approach and also that tops down approach of diffusion. But yeah, that's kind of like a starting point on kind of your question there. But there's a lot to explore there for sure.

Yeah. Diffusion as video games. Are you thinking about bringing in virtual beings or virtual characters into this so that I can start working on an avatar that I would actually create with these techniques? You know, make an AI girlfriend or something. We gotta find a better term than that for these things. But they're coming, right? Photorealistic things that you talk. That a user talks with and can create environments around them, manipulate the real world and the virtual world around them.

Yeah, no, that world is super interesting. I would say there's like two types of how I Think about Flora one is like not even as a business, just as a tool. Right. For creatives. As a tool for creatives, we want to add as much as possible. I want to have that like kind of hey, gen type experience directly in Flora and with everything else. If we add like text to speech, 11 labs, all this stuff, you could program your own fully programmable hologram that responds to you user text notes to program in their personality and direct how they should respond and then be able to deploy it somewhere. And then there's Flora as the business and what we need to prioritize and building off features with what we have right now of text, image and video. There's a lot of market to go capture right now and UX to build out around that. So we need to prioritize what we have right now. But yeah, audio 3D holograms or talking heads, however you want to call it all in the roadmap. For us, it's just a matter of when do we add it and how do we add it in a way that's really, really intuitive with everything else where it feels really, really simple to make the version of that talking head that you want.

Most people in my life are not nearly as educated about what's going on in the AI world as you are. What excites you about what's going on? What you know, give us a state of diffusion and LLM world that excites you right now?

Yeah. I'll answer this question from like a cultural perspective. You know, the thing about like the early adopter to early majority to like late adopter and there's a big gap normally between early adopter and early majority. I think we're about to like cross that gap because we're working with Pentagram and they're asking us for generative deliverables and things like that. That audience of early majority folks that want to adopt AI because they actually see productivity gains and benefits from doing that is happening now. But interestingly, the tools are behind there. There are so many really interesting things that you have to go on hugging face to go find or go dig into comfy UI to go find that would really, really benefit the creative world. And they're asking for it already. They just don't know how to do it. So our job right now is to as quickly as possible ship a lot of these things in an intuitive manner so that all these really great creators in the world that are interested in it can start making really, really cool stuff. Right. And the other framework we have here is I Would say there's like three categories of people in relation to AI. There are AI natives. This is most of the people you see on Twitter kind of talking about stuff. They kind of have the knowledge that we do. They get excited when a model comes out, but they actually haven't used other creative tools before. Their creative tool skill set is like Runway, Luma, all that stuff.

Probably not. Haven't used Illustrator or Comfy or Blender or Visual Studio. Even programmers have that kind of history, Right. There's a lot of programmers on X. Yeah.

And it feels like the general psychology there is like, I'm really interested in AI as a technology itself. The stuff I'm making, I'm not actually using that for like, you know, making money or creative output. It's just for fun, right?

Yeah.

Then you have the AI curious folks. These are folks in the professional industry that know creative tools, they know Illustrator, they know all this stuff, and they do creative work for a living. This is actually a very big segment of the population that I don't think a lot of people making creative AI tools are aware of. And the reason why those folks have not adopted AI so far is because no one's built a tool for them. They're kind of looking at all this stuff happening in AI native world, and they actually feel like they're getting left behind, especially given the messaging from the AI native world is, AI just killed Hollywood, AI just killed designers. And the designers in Hollywood, people are like, hey, what's going on here? Can we use AI? How do we do it? So they're kind of scrambling around, trying 20 different tools, not really sure what's going on. What we're trying to do for them is give them everything in one place, add Flora in their roster alongside Figma, Illustrator, stuff like that. So we deliver Flora in the context of those creative tools to them. Oftentimes Flora is more intuitive than something like midjourney, because they're not used to using prompts. They're visual people. They want to see it in that interface. And the last segment is AI haters. We don't think too much about that. A lot of that will change over time. But they're kind of like this calling siren to society of like, hey, like, let's still, like, slow it down a little bit. There's a lot of disruption going on. There are things we haven't thought about in relation to IP giving credit. And a lot of those things are valid too. But that kind of also is a voice for AI curious people. We saw this with Crypto as well, where, like, some people were really for it, some people were against it. People started shipping over to being for it in the creative industry. It's also this big question for them in particular, am I for AI or not? And that's a question that, you know, once they are for AI, they can't really go back.

Well, I've been watching this community since the beginning. You know, 24 months ago, Luma could barely make a blurry Gaussian flat or a neural radiance field. Right. And now they're making really fairly good videos. I mean, they're very good. They're getting close to Hollywood. Right. You know, quality. There still is a gap between, you know, something you film on a. On a camera camera and, and something you generate in AI, but it. It. That gap is closing every week, right?

Yeah.

And so I think that's also pulling people in. It's like, okay, the, the. The technology is now good enough to create creative things.

Yeah, being in the AI industry is just really fun because. Yeah, like, we started in January 2024, and since then, what, like the, the SORA demo they showed in like, February of 2024 that blew everyone away is just a standard now. Yeah, it's crazy. Yeah.

Yeah, there's better.

Yeah, it's better now. What I'm really excited for is, like, more and more control mechanisms over it, over the generations. Because, like, I think, like, we've been. I'm starting to think about, like, a framework for evaluating AI outputs or like, AI stuff in general. Like, you know, the thing about, like, precision versus accuracy. Like, sometimes, like, you know, a model can simulate natural phenomena like waves crashing and stuff really well, but it's not sticking to what you ask for. So there's like a difference there too. And, you know, there's different criteria in which you can evaluate. I love what Loom is doing with like, first frame, last frame, for instance. That's really helpful seeing some stuff coming out with, like, video loras. So, yeah, a lot of these control mechanisms I think will be important for broader adoption. And I've been talking to many, many creative teams in the last two weeks since we launched and went viral. Everyone's interested in learning how to incorporate AI in a professional sense. And when I get on these calls, the main questions I hear are one, like, okay, this is cool. You have everything in one place. And then we quickly go to them asking me how to do very specific things. How do I take. How do I change the lighting in this scene using AI? What model do I use? How do I just make it? So this product photography just moves slightly to the left a little bit for, like, five seconds. And there's this almost, like, really interesting curiosity of, like, people starting to poke at these models and figure out what they can do with them. I think the more interfaces we can build to enable that control, the easier it is for enabling those metaphors for the creative class, the easier it will be for them to adopt and also just make really great stuff.

Back to video games. There's games like Tetris, which are 2D games, and then there's games like Call of Duty, which are 3D games. You're walking around a 3D world, manipulating the world, interacting with the world and stuff like that. I mean, Edward Saatchi, who created Fable Studios and created this thing called Wolves in the Walls, which won him a. An Oscar or something, I don't know, Academy Award. He created a house that I was in in VR, and there was a little girl jumping around this house and drawing on the walls. And then she came over and started talking to me, and she handed me a virtual flashlight, and I grabbed it from her, and I. I spent 10 minutes just playing with that flashlight because it worked like a real flashlight. You shine it around the world, and it lit up the walls just like a real flashlight. And I shined it in her eyes and looking at me and talking to me, and it worked like a real flashlight. Right. That's my expectation of entertainment in the future, is we're heading into a highly interactive virtual world. How is Flora gonna get us into that? Creating that 3D world with triggers, with all sorts of interactive pieces, even geofences. You walk, you know, into a kitchen of a house, all of a sudden, things happen, Right?

Yeah.

Tell me about your dream about where things are going, and how does Flora get us ready for that world?

Yeah. So the origin of my artistic practice, which led into Flora's mission statement, I suppose, is how do you make it possible to speak your idea into existence with a high degree of creative control? That experience you just described, how do you just ask for it and then get all of that with all the interactivity right away? I think the answer is something kind of like this. Probably not exactly accurate, but at the top, a natural language interface layer where you ask for it, then an orchestration layer that gets all the different logical interactive pieces together, because you do need that logic of if she hands me a flashlight and I point that flashlight around, the beam hits the wall in a certain way, there's a lot of orchestration. Required. And then underneath that, I think it's some sort of base physics engine, maybe something unity like. Because I think using diffusion to simulate a ball dropping has not worked very well. I think that's probably a better way to do it. And over time, like as world models and all these different things get better also. So I think Flora is that middle layer, that orchestration layer. Today we are focused on like 2D and then moving 2D, because that's what the industry needs right now. But over time, you know, as Fei Fei Li continues doing stuff at Spatial Labs, my friends at Stanford Research Lab continue doing work on world models. We want to integrate that in. So that way you can have as. As robust of an experience as you can right now. When you run a video model, it takes like 3 minutes, 2 minutes. In the future that might be like that. Right. And that's what we were pushing the boundaries of with real time AI of these image models, you can call 20 times a second, maybe you can call video models in that, like really short time as well. Maybe you can do 3D simulations at that short time as well. Although that's probably three or four years out. Once we get there, though, the ideal experience is like you just ask for that world. You want the orchestration layer, figures out the right models to pull together. It builds the base structure and you're just in it. And then whatever you do inside it, you know, it's all accounted for in that.

A lot of this stuff sounds probably freaky deaky stuff to normal people, but, you know, I was at one of the first Twitch parties at south by Southwest, and there the DJ was wearing a VR headset and had a virtual world around him. The 400 people at the real party saw that on screens all around, right? Yeah. And then he had a virtual camera that was filming in the virtual environment and he could move that 2D camera around and that was live streaming out to Twitch and that got 75,000 people watching on Twitch. So I've always wondered when we're going to see this mixture of virtual worlds, creative tools like flora with a 2D camera to. Because I've. Here's my thesis, the creative class that I'm watching, you know, AI artists, AI filmmakers, AI music video makers. They're gonna soon be rock stars, they're gonna be performance artists because they're gonna be able to work in real time in an environment like this. Right. Speaking things into existence. And it'll be very entertaining to watch somebody who's very creative do that.

Yeah, yeah. No, that's that. First of all, that sounds like a really sick experience. I wish I came up with that idea. And, yeah, you can steal it.

I created the first documented reality app on the iPhone store by spitting out of my blog. And a guy at Yelp built it, right? Yeah, yeah, you can steal my ideas. They're not patented. Yeah.

What's the Picasso quote? Good artist copy. Great artist. Steel. Yeah, no, actually at itp, my grad program, there's an entire group. The good thing about that program is you got people from everywhere. You got, like, music people that are randomly interested in technology. You got DJs kind of coming in. You have people coming from Venture, I guess. But the most interesting, one of the most interesting kind of groups, there are VJs, visual jockeys, right? There's like Disc Jockey that do the music, and there's Visual Jockey that use stuff like Cost Designer interactive installation software to make, like, the background visuals behind it. Right? And yeah, like, if, like, DJs right now will like, you know, really quickly try to make a VJ set to, like, accompany their music, it's kind of like a live music video for. For the dj. But if you can make that experience, like, so easy to control for them, especially given their skills, is mostly in, like, Ableton and music, where they can control, like, every element of the experience, that would be crazy because then they can fully integrate their creative vision and control kind of the vibe extremely well. So that would be really, really cool. For sure.

Can you imagine doing this at the Sphere in Las Vegas? You know, you got 4,000 people watching these 16,000 pixel screens. I mean, they're insane. And the music quality there has 165,000 speakers in that building. I was at the final day of CES and they had a concert there with generative artists putting their art up on that screen. It was insane how good it was, right? And having somebody down in the middle doing this, using flora to generate stuff for the screens, it would be crazy.

Yeah. I feel like the. The Sphere is to create a technologist. What the Met is to, like, find artists. It's like the ultimate place you could perform, the ultimate, like, venue. So, yeah, I mean, eventually would love to collaborate with someone to do something for that. That'd be. That'd be so insane.

Yeah. Can you tell it? I need to generate a video for the Sphere or a bunch of models that'll run and an Nvidia card that'll drive 16,000 pixel screens. Because that's quite different than generating a little video for putting on YouTube in 4K. Right.

Yeah. I actually think the hard part will be like, plugging into something like Resolume or Touch Designer. Those are like the venue softwares to, like, get the. Get it mapped correctly. Yeah, I haven't looked too much. We actually have someone on our team who built like, an interactive dome for NASA to, like, look at different parts of the solar system. So he probably knows and I can.

Ask him, but yeah, it's crazy. Generate for the sphere. I could see it in a future version of Flora.

Yeah, you can say, like, simulate the solar system and then like, it just kind of like fully generates all around you. That'd be crazy.

I mean, the. The artists who go to coach, when I went to coachella, they had 110 LED screens around you, some of which were on robots. They would come down and spin around and, you know, having something like that, like, oh, I'm. I'm performing at Coachella in that month. I need to create a whole art installation for that, for, you know, Skrillex or somebody. Right. It's a crazy world that you artists are living in, isn't it?

Yeah. And actually, you talking about the memory, realize something. That music software is far ahead of visual software. Like, if you go to a concert, the sound goes to everyone, but the screens can only really be seen. First of all, the screens are just kind of one flat thing at the very, very front. It's not immersive, if that makes sense. Yeah. And I think that's because, like, with music software, like, we have Ableton now, like a single person. In the past, if Mozart, like, made a banger, he has to find an orchestra to, like, play it. Now you can have like a kid in New Jersey just on Ableton, make something, post it on the SoundCloud, and everyone can hear it. On the visual side, it still takes a lot of time to. To make everything and also, like, in the context of a crowd for everyone to see in a very immersive way. But yeah, that's interesting.

Yeah. And we're heading into a world with glasses. I mean, I've seen the wave guides for glasses coming from Loomis, from an Israeli company, and many others. Meta is working with, you know, making an Orion Glass that has ability to see virtual stuff in front of you in full color. You might say those are only. Those are two years away or three years away, but they're coming. And. And when they do, I mean, it takes you into a whole new immersive world as a user, and it'd be interesting to see where you go with that, you know, over the next two, three years.

Yeah. Like if you go to a concert and everyone's wearing those glasses, I guess you need to find some way to like have a better connection method than Bluetooth.

Yeah.

Automatically know like they're there because Bluetooth, like, you know, on these zoom calls, headphones are always disconnected and stuff like that. But yeah, that'd be really interesting. I do wonder though, and you probably have good insight on this, like. Yeah. How are all these things going to be coordinated in the future? Because like, let's say you walk into a coffee shop and their menu is an AR or whatever. Like, how is space almost defined in your view?

Web XR is the thing to watch there because then your user doesn't need to load up an app to interact with the coffee shop. Right. Just by walking into the coffee shop, it could impel a URL, which would impel a 3D environment to pop up or a little menu for you, you know, on the desk in front of you or something like that, so that the Web XR movement is one to watch. There's a lot of unanswered questions because, you know, we don't have glasses really, that we wear all day long that interact with the world like this. I mean, I have a Vision Pro and it does some of that at home, but it's really not well integrated yet with, with generative tools that could, you know, put a spongebob on my coffee table or something like that. Right. You know, yeah, it's a new world coming at us and that's why I wanted to talk to you because you're, you're pushing the user interface. You know, do you see this as a new boundary where, you know, because right now everybody's comparing competing over, you know, voice models, you know, talk to Grok or talk to chat GPT or text to it. You know, they're not really thinking about really bleeding edge, make the user experience very different. That's what got me, like excited about you because you're really trying to bring the AI world into a new thing with user interface.

Yeah. You know, if I was more technical, maybe I'd be trying to build my own model, but, you know, I'm not. So this is the only thing I could do. But yeah, I mean, what we're really obsessed with as a company is we view ourselves as an applied HCI company, taking all these technologies and figuring out the right way to deliver them to society. Right now it's primarily creatives, but yeah, down the line that may change. We Want to be able to let anyone be able to create really easily, but still have a lot of power. And that's always a trick. If you can make it really easy to make something really powerful, that's the ideal experience right now. Where we're starting with, we think is the right kind of underlying structure for the next 10 years as a lot of different kind of tools come out and change and we may not have gotten it right. We probably optimize for the near term, one to two years. If things change dramatically, world models, that world gets defined a little bit more. We may have to change the way we structure things. And we will. We will never kind of just stay with one interface just because it's making money. We need to kind of, like, move forward and, like, allow people to access these things better and better. So, yeah, that's kind of how I view what we're doing for the most part.

Okay, let's go back to talking to the beginners because, you know, while there is a nice community using Luma and Clang and Sora and, you know, all these new tools, it's a small community, really. There's going to be waves and waves of new people coming into this because they're going to get attracted to, you know, creating their own pictures or videos or worlds eventually. How do they get started? What's the first, you know, few things that they should learn about, you know?

Yeah, I think I heard a quote once, like, the best teacher is a recent student. So if people are listening, I'll ask you to take this advice with a grain of salt, because I've been deep in it for like a year, but I think it's probably getting to that first aha moment of almost like taking a step back from the creative process and, like, feeling more like a creative director. I think that's when it starts to kind of click of what AI can be like. I talked to this one designer who, like, went to art school, but he was never really great at drawing. I was never really great at drawing too. Now he's, like, deep into Comfy UI and all that stuff. But he started with no knowledge. But the reason why he got so deep was for the same reason people get really, really good at drawing. They want to have as much control as possible. He's just doing it from his perspective, controlling these AI systems to get him to exactly what he wants. And he's gone really down deep. But I imagine for him as well, it started with that first aha moment of, like, if I just figure out how to Prompt this correctly, I can get pretty close to what I want. Oh, let's go look at control nets. Let's go deeper and deeper. So I think that's probably. It's kind of a mental shift. Once they see kind of the benefit of it, that probably gets people started.

Yeah. There's now a new AI music video channel that I'm watching. I forget the URL of it. Send me a note. I'll give it to you. I'm learning a lot just by watching it, because now I see what's possible with the Crown tools. Right? And I see, oh, this guy created a whole 3D world and is walking through it and stylizing and changing it.

Oh, that's possible.

How did he prompt that? How did he do that? Or she did that, right? Or. Or how did, you know, Trisha create this pink brand? You know, and how did she get her face into the. The system? You know, I know it's possible that knowing it's possible, it gets a lot of people on board, right?

Yeah, no, definitely, yeah. When I was at itp, like, you would just see, like, older people in itp, like, people the year above and just making crazy stuff. And then you ask them how you do this, and they're like, touch Designer or some niche software. And then you're like, all right, time to watch YouTube tutorials, figure out how this really esoteric software works. So, yeah, I think it's very much, like, driven by outputs. And I think that's the great part about the AI Twitter community. Everyone's posting stuff. Some people are kind enough to list out how they did it, and that kind of benefits everyone. I also noticed that with AI in the AI art community, it seems to be accelerating really fast because there's not this, like, a gatekeeping element. Because everyone knows, like, you know, we're all kind of, you know, just trying different prompts. Like, what is the value of a prompt? Right. Like, I'm happy to share that. So that's been a really cool thing to see, and I think it progresses the aesthetic as well, really, really quickly. You see these trends kind of come and go because it's so easy to just sample someone else's style, iterate on it, build on it, and go from. From there. So, yeah, that's definitely a really good point.

Anything I should be asking you? Because I'm not an artist, really. I'm not a movie creator, director. I'm not, you know, doing this every day.

I don't know. It seems like you're really kind of plugged into a lot of this stuff.

I'm a. I'm a keen supporter and, you know, consumer. So I'm watching all this stuff. I built lists of 3000 AI artists. So I. I know enough to ask a few questions, but I just wondered if there was anything.

I'd like to ask you a question.

Okay.

So, you know, you've been at the forefront, like, reporting for many different tech waves, right? And there's generally friction with a lot of things. Right. But every technology comes with a different societal friction. With AI, it's kind of like this replacement thing. With crypto, it was like this. Are you for it or against it?

Yeah. Facebook, when they came out with the newsfeed, a million people protested the newsfeed. Now every service uses news feeds. Right. But it was controversial back in the day.

Yeah. I'm curious, like, what sort of technology wave in the past do you think this is most similar to?

I haven't seen anything like this. I mean, in my list, there's 6,100 companies, and there's a lot of resistance. I mean, we're about to get autonomous cars, robots in our world, augmented reality glasses, they're all coming. And. And the AI world is just. Every hour there's something new and that causes its own resistance. People hate change. Right. You hear, if you get out of Silicon Valley and go and meet normal, everyday people, they're just like, they have no clue what's coming and they're resisting it. And it's that resistance that's real interesting to me. I've studied it. Every time there's a paradigm shift, there's always a resistance. But this resistance is off the charts, right?

Yeah.

You can sort of say, well, when the news feed came out, that was resistance. When, when the mouse and windows came out that there was resistance to that. A lot of nerds in college told me, oh, who needs a gui? Real program, real computer users can use a text window, you know, to type to it. You know, we don't need mouse and windows. Many people told me that. Right. Autonomous cars, you see the same resistan. I'd never let. Let a little chip, you know, driving me around. I. I'm one guy in Kansas. I did a lot of consumer research for the auto industry. One guy in Kansas told me, I'm a narcissistic control freak and there's no effing way a little computer driving me around. Now we have people, people who are living in San Francisco and LA taking waymos every day, and they love them. They say, I'll Never drive a car again. Right. So they switch, they flip over into a new world and we'll see the same thing here that as. As this new world keeps getting better and easier and more powerful and more engaging and more, you know, amazing. It'll drive more and more people into it. Because this is where the future jobs is going to be too. Right. People who are using at Flora in a few years, they're going to be creating brands and experiences and games and businesses. Right.

That's really interesting. I think that anecdote you shared of that person in Kansas and that sentiment there. I was when I first got to New York in like summer of 2023. The GPT stuff had kind of just happened a couple months ago and I was walking down the street and I heard some drunk guy being like, you know, why does it matter? AI is going to take my job anyway. And like that sort of sentiment, I think the sentiment from AI and the reaction to that in particular across like, you know, robots, autonomous driving the creative class is this feeling of like I'm losing control. Which is different than like everything else in the past where it was just like, I don't want to do it this way. Right. With social media, I think it kind of felt like one is like, you know, I don't want everyone to know what I'm doing. But I think it was more about getting sucked into the whole social media realm and taking attention away and stuff like that. I do think that a lot of times there are the good companies and then the bad companies that seem to emerge as dynamics. I was talking to the number two at Snap recently and he kind of talked about how back then Snap was somewhat perceived as the good company because they were ephemeral, whereas Facebook was like, everything you share is there and they're focused on metrics to get you more engaged. But Snap was like, we just want you to communicate as quickly as possible and whatnot. Yeah. And I think a lot about in the creative world, like what that is and partially how to position ourselves as like good, but also just how to be good. Right?

Yeah.

I think it's like the main takeaway I've gotten so far is that like creatives don't want to be automated, but they do want to automate. They do go through a lot of bullshit in their day to day life that takes a lot of time and gets them away from like actually having their vision executed. So in terms of the messaging, but also most importantly the product itself, we have to bias towards kind of that desire, which is Also a desire that I have. So that's kind of why I asked that question. And it's helpful to hear how you're thinking about it.

That's pretty deep, you know. Ansel Adams son gave me a walk around Yosemite national park and talked to me about his dad. He said when his dad first got to Yosemite, he was holding a Kodak Fox camera. He was a. Actually a pianist back then. And his first pictures he got back just were shitty pictures compared to the experience of being there. There was a huge gap between being there and why you got back on your pictures. And he spent his whole life trying to close the gap, you know, trying to get closer to what the motion that he felt by going to Yosemite National Park. And it seems like the same thing is happening now. It's like, how do we create beautiful things?

How do.

How do we take people to beautiful places? You know, both real and, you know, made up, you know, because. Or a mixture of the two, right? You might someday be at a rave in Yosemite national park and you're augmented reality glasses and things are changing all around you and morphing and. And, you know, becoming interactive. You touch, you walk through some grass and music changes, right. Or something like that. So many things are about to happen because of people like you. So thank you.

Oh, I mean, I. We started building this for ourselves so, you know, we're just kind of doing it for fun, honestly, so we can make cool stuff. But we're really grateful that other people like it too. I have a question on Ansel Adams. I mean, when I see his work, I think I feel like quite a bit of like that emotion of being at Yosemite. They're really great photos. Did he ever think that. Did he ever think that he got there?

No, I think. I think if he was live today, he'd be like looking at you and looking at the tools that we now have to create new things, particularly the Vision Pro. You know, be like, oh man, let's dive in. His son actually, he invented several things to make like. Like he made the advertisements for Penn Station for Eastman Kodak Co. And he invented a tripod that what your panoramic does on iPhone, right, to make a wider picture. So he had notches in the tripod so you could take four images and stitch them together for the advertisement. He was always a. Well, he. He resisted too. Actually. I talked to my friend Thomas Hawk, who's a great photographer, and he says, ah, he resisted color, right? Because he didn't like, he didn't like color or didn't feel impelled to join in on. On color that much. He did do some photography, but he. He really resisted that and. And attacked other photographers who use color, you know, instead of black and white. So I. I don't know. You know, everybody has their opinions and their point of view and their thing they're trying to protect.

Right. That's super cool that, like, you know, he took it so seriously. He was working with the Kodak company to come up with things to help him get more creative control.

Yeah.

And, yeah, it makes me think about who we should be working with to, like, help them. You know, we're obsessed with getting creative control over what they want.

He really invented a lot of algorithms. I mean, when you visit his darkroom in Carmel, his enlargers on train tracks is big and larger and it has a grid of lights in the back so he could turn on and off lights to make parts of the picture darker or lighter. Right. And we take that stuff for granted. Now we can just click on an iPhone and go, make that part darker. Yeah.

That's so fascinating that he would be so, like, wanting to control every aspect of it and yet, like, almost as an aesthetic choice, wouldn't want to do color. And, you know, that sounds like just like a personal aesthetic choice. And I respect. That's pretty cool. Yeah.

Yeah, it's. Yeah. We're in a new world now. I. If he came back to life and came and saw this world, he would probably flip out. Because back then there was. Wasn't computers, there was an intern Internet. There wasn't, you know, tick tock or Instagram. There wasn't, you know, all these new AI tools that are hitting us, like, constantly. Right. Yeah.

I would say the closest metaphor to like, you know, paint to what we're going through right now of just like being able to generate media is kind of that painting to photography kind of pipeline. So, yeah, I read up on like, Kodak story, and they started with those really small brown. The brownies, I think.

Yeah.

Over time, they got more and more. Yeah, that. That's a really interesting metaphor to look at from a. From a cultural standpoint.

Well, cool. Well, I've spent enough of your time. I, you know, I love, I love what you're doing.

Keep.

Keep moving because you're pushing the industry forward and I appreciate that.

Dude, it's great chatting. We could chat about stuff for hours, I think.

Totally, totally. Where did people find it and find you?

You can find it@florafauna AI I guess you can find me on Twitter at Weber Wong Wong.

What's Twitter X? I still call it Twitter.

Elon.

I'll slap my hand. You know, you might turn off my power.

Yeah, yeah.

Anyways, thank you so much. Let me stop recording here. Thank you.

Thank you, Robert, always.