We talked to the CEO of ComfyUI

ComfyUI is reshaping how studios create AI-generated content, with 50,000 downloads per day and adoption across Netflix, intelligence agencies, and major VFX houses.

In this episode, Yoland Yan, CEO of ComfyUI, shares the origin story of the open-source AI workflow platform that has become an industry standard. We discuss how co-founder comfyanonymous created the tool with zero machine learning experience, why control and quality matter more than simplicity for professional creatives, and how the platform's 60,000+ nodes enable everything from marketing pipelines to classified government applications.

--

The views and opinions expressed in this podcast are the personal views of the hosts and do not necessarily reflect the views or positions of their respective employers or organizations. This show is independently produced by VP Land without the use of any outside company resources, confidential information, or affiliations.

Transcript

So node-based interface became this very natural existence in here to allow you to do that. And that is pretty much what ended up allowing people to adopt Comfy because it gives you as much control as possible into these AI models. And it's to this day why Comfy is really unique and different and being adopted at the kind of central AI tools in a lot of these studios.

All right. Jolen, comfyanonymous. Thanks for, thanks for coming on. It's great to meet you. Actually, yeah, this is the ComfyUI part 

right off of that. Maybe I didn't give us enough 

context. You're not comfyanonymous. 

I am not the comfyanonymous. My, my co-founder is 

co-founder. 

I, I'm the CE. 

Okay. 

Yeah, because the comfyanonymous.

That is a good lead into our first question. The though, like what is, can you give us kinda your background and the origin story of. ComfyUI. 

Yeah. Yeah. It, it's, it's actually funny that it confuses a lot of people because, uh, Yannick go by Yan and my last name is Yan. So it's like, you know, uh, things start to get very ComfyUI and Yannick usually isn't very, uh, you know, communicative with some of the open source, uh.

Players and I am, and we had this incident where folks on, on hugging face side was like, wow, this guy's bipolar, uh, half of the time. He's like, not responsive and, uh, you know, sometimes, uh, uh, somewhat aggressive and half of the time he's so nice and so clear in terms of his communication. What, what is going on.

Yeah. Like it does confuse a lot of people. Let me just go back in terms of, uh, the history of the, the product or project. Initially, this, uh, started as a pure open source project. Uh, this was beginning of, um, 2023. Um, actually the background story of how Yex started. Uh, all of that itself is also another like rabbit hole and fascinating, uh, stories.

But just to give you the high level, he was, uh. Uh, prior to this, never wrote a single line of ML code. Uh, worked for 10 years as his dad's dental implant factory as a c plus plus programmer on automation in Quebec City. So, like, you know, pretty much the furthest you can imagine, oh, uh, from, you know, a Silicon Valley ai, uh, tech company and wrote it in two weeks.

Put it out there. And at this point, you know, uh, or pretty much a year later, he became probably the best inference, uh, person in the visual space. Uh, and the project after he launched at the beginning of 2023, start to gradually pick up. If you look at the history of, uh, uh, confis journey has always been this very, very steady growth.

Never this huge jump. People hyped over it, you know, uh, as we probably are seeing from open clock, it, it takes a lot of. Effort for people to onboard to this product and it gradually, uh, grew and the ecosystem start to flourish. Over time about, uh, a year into this journey. Uh, that's when, uh, I came in a bit earlier than that.

But, uh, that's officially when, uh, we kind of got started. I, on my end of my journey, uh, built, um, my own product on top of Comfy and then started to see the value of, um, this entire ecosystem at the same time. We were also really curious in terms of where the project is going over the long run and ideally, if possible, kind of help guide the future towards, uh, more beneficial on our end to be able to build on top of Comfy and got kickstarted.

The conversation started contributing, uh, to the space Get connected, got got connected with Yek and other open source contributor. And came to the realization that it is a incredibly grassroots process. No one's, you know, uh, there's no Jesus taking the, uh, the wheel type of scenario. Everyone in here is just, you know, focused on a very niche area.

I think. comfyanonymous. Yannick at the time was just doing huge amount of model improvement, inference, uh, optimization. And other folks in the ecosystem might manage the extension, um, and someone else would, might do the, uh, front end and a very uncoordinated effort. And that's when we realized the problem and the opportunity to.

Kind of centralized the effort under a single, uh, organization and start a pitch towards them. And about, you know, at the beginning of, uh, uh, 2024, uh, that's when things officially started, uh, uh, working, us working together and brought in, uh, quite a few open source contributors in the ecosystem. So that's kind of the origin of starting point, I guess, of both in terms of the tool and the company.

Happy to kind of go into more. The past, you know, ever since we got started. What are some of the progress we've made in here? Um, but yeah, I, I don't want to blabber on for too long. 

Do you know what inspired him to wanna make this tool in the first place to kind of start building out, you know, early versions of Gen AI imagery?

Yeah. Actually the, there are three key characteristics of Comfy, uh, or what en, en enables people to do. One is control, the other is quality. And then the third one that's, uh, uh, scalable and scalability. What started this project at the first place is still what remains to be true today in terms of why, you know, Comfy.

Is this central of, uh, adoption in the studio VFX animation world overall? When he first started, he wanted to actually build a, you know, have a system that renders the initial part of the image using one diffusion model and using another, uh, diffusion model for the, uh, higher resolution, uh, generation. Uh, if you guys played around with a lot of open source model, you might know that some models are very good in terms of their aesthetics, um, but they might have limitations, for example, in terms of the.

Resolution that they generate. And sometimes if you ask those models to generate higher resolution, what they give you is a, you know, four quadrant of, uh, uh, images, because that's what those are the resolution they're trained on. So they, he, uh, at the time wanted a model that is really good with aesthetics on the initial.

50% of the steps in terms of generation and then using another model to finish up when the foundation of the image, uh, is already laid out and there's just no product out there. There's nothing that gives him the control like that unless he just rewrite a code himself. And even when you write the code, there isn't a good visual feedback on a, you know, screen that.

Tells you like, oh, okay, yeah, you changed these parameters. This is the result you get. So node-based interface became this very natural existence in here to allow you to do that, and that is pretty much what ended up allowing people to adopt Comfy because it gives you, uh, as much control as possible into these AI models.

And it's to this day, why Comfy is, uh. Uh, really unique and different, and being adopted as the, as the kind of central AI tools in a lot of these studios. 

Amazing story. Love it. Do you have the, does Yannick or you, do you guys have the original image? 'cause that could go up on the AI history MuseMe, you know, a hundred years from now.

We do. Oh man. Uh, this company has gone through a lot of, uh, you know, in terms of it's. Brand, uh, position. One thing I sometimes show people is, uh, I'm gonna pull up over here in terms of the Comfy org website, we really leaned in hard, maybe a little bit too hard at the beginning in terms of this, uh, uh, grassroots energy.

This is the initial website of the company. Uh, this is the very first image generated Nice. With Comfy ui, and that's literally like, you know, the homepage as you can see. It's, yeah, the homepage itself is also very just like make me cringe a little bit at this point, looking back. Um, but also back then that was, uh, you know, uh, what we're leaning into in terms of kind of.

Wanting to be this open source, uh, and, you know, very grassroots energy type of organization. And gradually over time start to become more mature, both as a product and as a brand. And now the website looks different. And, and in the future the website is gonna even be more professional and, you know, very much speaking the voice towards the professional creatives.

Yeah. I mean, can you tell us about kind of what the current landscape looks like with Comfy, sort of the kind of current products and where you're looking to go with where it develops? 

Right now, uh, there's some of, uh, you know, kind of high level, uh, examples I can give in terms of Comfy, Comfy popularity and, and adoption.

Mm-hmm. There's even more like. You know, in terms of, uh, high level roadmap, what we're looking at over the long run, given the, the current where we are in terms of the local product, um, it's actually fairly prevalent. We now get, you know, close to around, uh, 50,000 download per day in terms of like the local installation.

Wow. Um, there're approximately, total about, you know, 4 million users out there and given the. Professional, creative ecosystem. You know, it's actually a non-trivial amount of, uh, number in here. I think the total estimate on our end of, uh, the entirety of, uh, professional creative in the world is about, you know, 50 million, uh, overall and already are approaching about 10% of the, those group of users.

Uh, and then ideally over the long run, we'll, we'll keep on expanding in this category. Uh, overall. And largely these are, uh, kind of books that want exactly those three characteristics, control, quality and scalability. In the end, when you have a pipeline, you can kind of, uh, push that out into, you know, a pro production environment and keep on running it against different assets, different input over the long run.

At least from our end, we want to build Comfy into this. What we call operating system for visual ai. Essentially it, what in effect it means is that. In the future, you know, let's say five to 10 years timeframe, more than 50% of generative content goes through this Comfy workflow layer. Um, it doesn't have to be, you know, kind of through our interface or product.

Um, it can be, you know, running on local, it can be running on, uh, our cloud, it can be on mobile, SDK, it can be on, you know. Anywhere. Uh, but as long as, you know, Comfy is the engine or workflow orchestration system that empowers a lot of the creativity. I think essentially we've changed, you know, the entire landscape in a way that, uh, the tool is owned by the creatives as opposed to, you know.

Historically, um, I'm sure everyone knows that largely the, the creative tooling space is usually locked in, in a close ecosystem. And, you know, and one, I even argue that it's, um, fairly toxic in terms of the relationship between the tool builders and the users out there. And. That's the paradigm that we want to shift, uh, in the new AI era.

Yeah, I think it's really admirable that you have a grassroots origin story and uh, you know, to some extent Comfy still feels very much like, uh, an open source independent company. You know, if you compare it to like the big tech giants in visual space, like your, you know, your Google, your OpenAI or whatnot.

And I think that grassroots story, it really resonates with the creatives because. Nobody loves, you know, I think everybody loves a grassroots story about a creative origin story like George Lucas and Star Wars, right? Like that started out of his garage and things like that. So I think if that is in your DNA now, that is definitely working for you and it'll continue to work for you in the future.

Yeah, we actually, the. Uh, one thing that, you know, people probably don't realize enough is, uh, the amount of success Comfy has, uh, enjoyed. Um, a huge portion of that actually came from the open source ecosystem. Uh, these are, you know, custom node, um. Extensions, uh, within the confi ecosystem, the builders around that, uh, you'll be surprised every day they're just a new node.

People created, vibe coded these days, um, that put out there, that solves part of the problem. And, um, open that to other people that can adopt into it. There are more than 60,000 nodes in, in our ecosystem right now, which is. You know, if you compare in terms of extensions to other platforms outside of create, uh, creative World, it's um, probably one of the top type of platforms out there to, to have that many open source developers.

Same thing for workflow builders that people were, you know, combining these nodes together for the. Visual creativity and, um, pretty much everyone in this entire circle of ecosystem are contributing, um, model builders as well. So yeah, that's one of the major reasons why Comfy ended up being, you know, such a huge success in comparison to any, you know, close systems out there.

This was, uh, sort of an ongoing question at I always has brought up, but ComfyUI is free, it's open source, you can download it, install it. Only recently you've had Comfy Cloud, but. Before then, like how, how did it make money? How was it lean, how was it to, to be developed and you know, like flourishing and continuously, like continuously improving?

I would say the initial journey of, uh, Comfy as a personal project was, you know, completely zero funding. It's not even. There's not even sponsorship for Yannick in this process. Uh, obviously he was, you know, partially employed, uh, uh, during some of his time at stability to continuously maintain the open source project.

But, uh, ever since kind of we came together as a team, it's mostly initially through, uh, venture funding that, uh, we found folks that are very aligned in, in this vision and. My God. There are, uh, we talk to a lot of people, uh, in this ecosystem and we filter out a lot of, uh, you know, from outside view, very solid investors, but fundamentally, we just knew they were, you know, not aligned with the vision or, you know, come in with the perspective.

Hey com, confuse a good brand. Maybe now it's a time to. You know, shift towards a Comfy Pro version. So yeah, at, at least at the early part it was finding the right investors and fund the team to keep on moving forward. In terms of, uh, more recently our, you know, cloud launch and other commercial kind of product launch in here, and this is generally describing why.

It's actually an incredibly powerful model to empower the. The, the growth of, uh, uh, creativity and creativity tooling over the long run. Comfy at the core is built as a, you know, uh, the open source product is built as the one person, one machine type of experience, a huge amount of, uh, optimization and work with put into, and this is pretty much state of the art for.

You know, past two years is around, you know, how do you enable people on a, a single system that, uh, just, you know, run these models and get as much out of it using your own local computer. At the end of the day, as a individual creative, you're not gonna buy, you know, 20, uh, you know, RTX 6,000 and put that on the cloud or run that in your basement.

It's, you know, financially irrational to actually do that regardless, uh, even if you are able to run it at all. All time. So from those kind of perspective, a very natural, uh, pricing or business model around it would be providing a very powerful backend that we are able to optimize for on the cloud in a very, you know, uh, distributed fashion that no longer requires you to kind of.

Maintain this kind of complex logic, uh, of running things in the cloud. Hence why a lot of people adopted company cloud as, as a product in here, in, in that world. We actually didn't need to do any, you know, feature gating as typical open source product. You know, a lot of these developer product would have some of the features out in the open.

Some of the, you know, let's say, oh, I will provide. 50% of the nodes in the open source community version, and then the remaining nodes all happen on the cloud. Really not necessary because, um, the speed and performance of the product is. A feature is one of the major feature. Are you going to wait for two days in terms of your rendering pipeline or would you be willing to kind of, uh, pay for the value of the orchestration on the powerful cloud backend that can finish the entire run in 20 minutes?

Right? Those are the kind of key differentiation that enables people to naturally flow towards using Comfy Cloud. And we actually even see people who are. Both using local, when they have some, uh, a decent machine and concurrently they will run things on the cloud just because, you know, both systems have some amount of wait time and right now your creativity is going faster than the compute or rendering time any, um, any machines can provide you.

So in that world, the more you provide out there, the more people will use it. So it's a very healthy kind of usage based model for us. 

What are some of the applications that you've been seeing, you know, either in studios or larger companies? And any of those applications been surprising to you or unexpected?

Oh, man. Uh, surprising. There, there are a few, um, some of which I don't know of how much I can, you know, talk about, um, in terms of the surprising part. But in terms of general application, uh, a large portion of our users are in the, you know, studio, uh, marketing VFX animation world, and that's primarily where we focused a lot.

For those group of users, the pipeline is usually, you know, a high level pipeline is very similar, like individual workflows might differ a lot in inside this company. They might have a pipeline that swap out, you know, autonomously swapping out, you know, uh, logos that are not. Their own brand as part of their video footage and make the video look more natural.

Another company might have a voice replacement model that, uh, you know, allows, uh, the footage or like the, the voice generation to be more, you know, natural or translate to a different language. So, but in general, the, the three key criteria is. Still the same for all of these users. The, the, the application being, you know, giving as much control to the creatives, you know, um, having the highest quality and then also in the end, a pipeline that's scalable over the long run.

So yeah, like the, there's a huge long tail of. Of these applications across different, uh, studios, but the, the fundamental nature is, is very similar. 

I saw on a recent post on Comfy LinkedIn website that, um, Comfy UI is now a job description requirement on a lot of AI facing jobs. How do you feel about that and how does that help guide you in future product roadmap and things like that? The, the fact that you, you now, this grassroots thing has turned into a requirement. In a lot of cases. 

That's definitely super exciting on our end. Um, and this has actually historically been the case that, uh, we, there's a Netflix post like almost, uh, two years ago or like, you know, a year and a half ago when, when we just got started, you know, said, Hey, inside Netflix is looking for Comfy creatives.

And then at, at this point, there's actually a lot more throughout the entire ecosystem. I think Netflix is, uh, um, fairly ahead of its. Uh, of, of their time. One thing we want to do in here is, um, and historically I would say. Quite transparently, Comfy, didn't do that great of a job to empower kind of, uh, the on-ramp or learning process from, uh, individual creatives onto the platform and going into the future.

This is definitely an area we want to address. We want more people to actually come in. With the experience to say like, 

oh, 

Comfy, I, I learned that in, you know, this course. I learned that in college. I, I successfully ran through the process of onboarding within the Comfy product and already know the basics.

So these are areas where we, what we do want to help address. And from a product side, there's also a lot of planned features that will enable people to. Go to a smooth on ramp process. App Mode is one of them. Um, App Mode enables people to escape the sort of, uh, workflow layer and then just operating at a higher, uh, level that, you know, you will see in your day to day or normal type of, um, uh, creative AI tools.

Yeah, I think that's kind of, uh, how we view it. 

Yeah, well, kind of speaking about the onboarding and sort of building out the workflows, I kind of had a question about how you're thinking, uh, how you're thinking about well, a combo of like the rise of vibe coding and agentic ai and sort of if you see building features into the product.

From my personal use case, I've like tried Quad code to build out a Comfy workflow. 'cause at the end of the day it's like JSON files. And it didn't do so it, it didn't quite get it. So I didn't know if, um, yeah, I'm curious if you're kind of thinking of, you know, the workflow end future ways to make it more friendly, whether it's like an MCP server or something that can interact from just plain language.

Like, I want to build a workflow that does X, Y, Z, and then you get a starting point with a company workflow that exists and that you can modify. 

Yeah. Yeah. It, it's a great point. A lot of people have brought this up, uh, to us so far. I would say like it's definitely a direction that we walk, we are pursuing over the long run.

Just, you know, the general theme of making a product easier and enable LLM systems to you. Even as part of the on ramp, um, make it easier for people to learn and adopt the tool and also in the future, edit the tool. If you think Comfy as somewhat of a visual programming language, I would say we're still in the, you know, precursor or um, the VS code days.

Even from that perspective, there's a some product level of, uh, uplifting. We need to keep on pushing on that actually get to the VS code level of quality of, uh. Um, a product, but yeah. Uh, one nice thing about Comfy being open is that other people can do this. Just yesterday actually, uh, uh, client, this, uh, AI coding, uh, tool.

Organized a, you know, hackathon on their own version of Comfy MCP Alibaba created a tool called, uh, Comfy Copilot, uh, that does exactly this. And a lot of times you can already, you know, a clo can do some, some amount of these kind of with one shot try. But yeah, if you actually fully, you know. Build a agent system to generate these workflows is gonna be a lot more powerful and more accurate.

Right now, at least the nice thing of being open source is other players, uh, in the ecosystem that will come in to help. And that is just generally a repeating pattern you see from the Comfy ecosystem. Someone will come in to deliver these features and help the rest of the ecosystem, and eventually on our end.

Um, it becomes a critical point or path for us to deliver these features from the core confi experience. Then we will definitely, uh, execute on that as well. 

Just on that note, um, there are now quite a few node-based AI workflow orchestrators, right? Uh, some of them are, you know, like, um, what was the company that got acquired Joey by the Foundry.

Well, there's a, oh, um, 

Griptape. 

Grip, grip tape. 

Yeah. There's Griptape. There's invoke, there's vy. 

Vy free, 

free pass. No. Yeah. Nodes are very invo right now. Yeah. And, and how do you see Comfy sort of maneuvering this in the future? Do you see the product getting easier to use or the opposite, getting more and more defined by professional use cases and in some cases it's getting harder to use.

Um, or do you ride the line in the middle like you have been 

on the. Uh, one thing we keep on, you know, emphasizing within the team and also to the, um, outside user and rest of the ecosystem is Comfy. Never really want to sacrifice the power, the core power inside the system. Uh, 'cause that's actually why, uh, people.

At the highest end of the quality and control adopt Comfy is because it has these core power. If tomorrow I all of a sudden dumb down, Comfy in the form of, Hey, let's just do. Image to video, to text to 3D and to audio. That's it. You know, this kind of general process in here, you will, you know, probably bring on a lot of the new users, but you also ended up completely losing the, uh, the opportunity of for, you know, making Comfy this highest grade of, uh, uh, professional user experience.

That being said, I do think user interface is a accessibility problem. And that's something that we actually can both maintain the power of Comfy while, uh, addressing some of these easy to use, uh, uh, issues. I actually think it's interesting that a lot of times, you know, from our understanding, when you use other, you know, node based systems or, uh, uh, products that came out.

Over the past, uh, half a year or so, it's still a complex system. It's, uh, still something that requires onboarding and people need to understand, oh, this is how you connect things together. So even from that lens, there's complexity all around. From our perspective, we just want to cater towards, um, the professional users and also the, uh, group that connects towards this group.

Uh, professional group App Mode is actually the. Prime example in here where it didn't originate from us thinking, oh, let's, you know, kind of build Comfy into a consumer product. It actually originated from, uh, professional creatives that want to, uh, be able to have this type of feature so they can enable their workflows to other, you know, creatives in their team that doesn't use Comfy, but do have ideas that want to test out against these workflows.

Um, so those are, or even in some case scenario, they built a successful pipeline that just continues run, and they don't really want to go into the, you know, node graph layer to make more modifications. So they want a simpler version of this to, uh, engage with the workflow they built. Not, it doesn't mean they're no longer, you know, professional creatives.

It just simply means it's a accessibility layer that you enable for. Their needs. That's the lens we're looking at. 

Tell me about, uh, App Mode and sort of the overview of the new, the new product that you're launching. 

Yeah. App Mode, basically, if you imagine all of the workflows as some kind of, uh, templates, App Mode is basically this layer that, you know.

Allows these templates to be easy to use. Uh, what you can do is on an, uh, workflow graph, uh, you can select a subset of, um, um, parameters you want to expose on within App Mode and all. Uh, once you complete with that, it's actually a fairly simple process of just. You know, uh, click and, um, select a bunch of parameters in here and you can swap this, uh, workflow into the App Mode.

And within the App Mode only those parameters are surfaced and you have a more of a, you know. Pure, simpler interface to directly interact with the workflow. And if you just change the prompt, then the result will con run through the, uh, workflow. But in the end, only the prompt is getting changed. 

So is that mode a a, uh, a method to getting a local executable, or it's, I mean, what is, how does a user use it?

Who doesn't need to touch? Comfy 

would enable people to, you know, keep on. Spreading App Mode type of workflow in the same JSON format, in the same, you know, I see, uh, experience. You can send it to other people and they can install this, you know, app, uh, either on their local instances or on cloud as a shared, uh, link.

And then through that kind of mechanism, people can, you know, use it everywhere or share with other people that they can just install the notes they need or models they need and they need and execute that on their. Local instances. 

So it's a cloud-based workflow abstraction layer. Would that be accurate to say?

It not? It, it doesn't necessarily, you know, linked towards cloud. Anyone local can also establish that as an app and that will be registered as part of your workflow file to say, Hey, by the way, this is a workflow that's App Mode. Um. Preferred or, you know, kind of at least open up in the App Mode experience.

And then you can send the same workflow to other people, or you can put it on the cloud and then give people A-U-R-L-A link. That's the kind of general experience we would imagine people to do either directly through the typical process of, uh, share the, the workflow, uh, file itself or, you know, share a link.

So like you could build a complicated workflow under the hood and you want to give your client a simple link where it's like, here you give, upload an image. And then style transfer or something happens in the Comfy workflow that you built, and they get the output image, but they don't have to worry about the spaghetti of comp flow workflow under the hood. 

Yeah, exactly. 

With the local sharing option. You know, you mentioned that you could just share a JSON file. How does it manage if you're missing the models or the nodes? 'cause that's always a pain point when I try to download other workflows and then it's like the model's out there, and then you gotta try to like track it down and install it in the right folder.

Uh, how is there an improvement in being able to manage that? And 

it, it is a separate process in here. 'cause you know, as you mentioned, is the same pain point happens for workflow mode as well. It doesn't necessarily need to be, you know, um, the same or it, it doesn't, it's not a problem that just happened in App Mode.

That is an area we'll, uh, keep on improving. We actually already made some amount of progress in here to be able to embed the models, you know, location and as part of the workflow. So now it will get prompted to say like, Hey. Uh, go download these models to enable this node to, uh, en, en en enable this workflow to work.

And then the other is, uh, the custom node, uh, itself, uh, to be able to embed those informations, uh, its versions and all of that as part of the workflow. But there are. You know, even further improvements I think, um, in the future. Uh, one thing we did realize is that there are a lot of open, you know, workflows out there already that just doesn't have these information and they still occupied a huge amount of people's interest.

And those, how do we. You know, convert and gradually move the entire ecosystem workflows to have more metadata, um, to have better reproducibility over the long run is something that, you know, definitely will come as part of, you know, our continuous feature improvement for throughout the year. 

That's very exciting.

And, um, I have to ask, since you're on, what are the chances of Comfy being acquired this year? 

Um, zero. It, the, the truth is, um. You know, obviously without getting getting into much of a detail, there has been, you know, a lot of interest historically. Uh, and you can actually see, you know, the creative industries being a little bit.

Worried about a, a player like Comfy to, you know, continuously grow, get more adoption and over the long run potentially, you know, completely disrupt with the open source approach in here we are as a team, are very mission driven in that sense. Um, you know, so he, speaking from the founder's perspective at least, uh, we all can probably work in really interesting, you know.

Research, uh, or, uh, you know, fun environment. And the reason we picked this path for both Yannick, me and Robin is because we see the necessity for us to build Comfy into this. Foundation layer and enable the, the future that we want to see. Uh, some, something I used to tell kind of, um, you know, early VCs in terms of why we're building this is from my personal view, they're in the history of Silicon Valley.

Um, there are two type of companies. One is what I view as. You know, you did some marginal improvement on top of what generally the, the world of, uh, this trend is going towards to maybe like, you know, if weren't for you, there's gonna be another player in here that would have done similar things or maybe slightly worse.

So all you contributed in this journey is that marginal improvement of your product experience and your execution. And there is another type of company, from my perspective, actually changed the. History or change the path of where the future is going towards to. And it's the second category of company we want to build towards too.

So from that lens, it really makes no sense for us to actually join a bigger organization to build into that. 'cause I think, um, in those views, it actually very likely will temper with the, the, the success and the, the, the vision we set for this company. Unless there's someone that just fundamentally aligned with this vision as well, which in my opinion, they're on on.

And we're probably the first one to get there. 

Nice. Since this, uh, podcast, Yolan is for AI filmmakers and people that are more towards, um, visual, medium and creative spaces, what's your, what's your message to them about Comfy usage? A lot of them are already using Comfy, and they're finding great. You know, applications within it.

Mm-hmm. Um, Joey and I share Comfy workflows back and forth here and there. Joey uses it for his productions. I use it on my day job. How do you see our industry specifically as we are continuing to evolve with the newer tools, but also retain that, you know, human first, creative first approach? 

I think a few messages in here.

One is, uh, at the core. I think Comfy and our view of where the creative world is going towards too is extremely human in the loop. Uh, and that's part of the core ethos in here. You're seeing a lot of noise in terms of, you know, the AI stops coming out, and then the, you know, SA fear-based message that drives marketing and adoption that.

Industrial design is done like it, I, I said like a very pet peeve on my end to just constantly see these things on Twitter. Just people like, oh, this is over. Like, you know, Hollywood's cooked. No. The only thing that will happen is they're gonna be more creatives that come into the ecosystem that is able to deliver.

Their creativity as a high at, at the highest level. And that absolutely just simply require, uh, human in the loop approach in here. 'cause at the end of the day, what sets, uh, the, the difference between the slop of the world, which will over. Flow the entire internet in terms of volume, not in terms of attraction versus the what stand out among all of the content is your creativity, the control, the artistic thoughts among it, and Comfy is essentially that.

A product that, um, you know, believe in this ethos, and it's also the, the trend we're seeing that set, you know, Comfy apart from anything else in that ecosystem. That's definitely one of the major kind of message I want to deliver for. For the creative world. If you see that, you know, these kind of bullshit narrative, just keep calm, carry on the, the, these Twitter trends eventually will die.

I guess Comfy will never be used for like, um, script to movie usage. Like that just doesn't make any sense. Right. 

Sorry, what do you mean by that? That Comfy never be used for script, for movie. 

So one of the, like the AI slot proponents, one of their point is like, well we, we will soon to get, get to an automated system where you input, uh, a script, like a story, and then out comes the movie and then there's like no human in the loop effort.

I think, uh, there it is very possible that will be kind of a feature and, uh, a possibility in, you know, whether Comfy or other systems, but the result is what matters. The what. Comes out of it and where do you want to take that direction and what further changes editing, uh, you know, process do you run in here is something that's very critical.

And I think the visual, uh, space has this high dimensionality that sometimes, you know, doesn't even register in people's mind of, to imagine if you look at text or you know, asking AI to write an email, 80% of the time it probably gets close enough. And then even that close enough, you'll go in and then do some fine tune adjustment, and they all keep on getting better and closer towards your voice, but you're constantly having this ability to go in to just make some adjustment in the end to complete the creation process.

Whereas, uh, on the visual side, if I ask for a very fine grained described prompt in terms of, uh, an image of a oil painting, of a water lily, um, even in that world, there are like, you know. Infinite amount of possibilities or control where the leaves are, like I can describe that, but like, you know, uh, does the AI really well, like exactly, precisely get the, the location.

Like there are many styles in here that you can further adjust the, all of a sudden image just is at a higher dimensionality, then text, hence why, you know, like a picture that's describe a thousand word. And then video is on an even higher level because incorporates motion, you know, uh. And world environment.

So from that view, uh, even text is going through this, you know, or landing in this final result where, you know, you can have ChatGPT write an entire book for you. Um, but that is just gonna come out as. Another AI slot. You would want to take advantage of these, empower yourself to be able to produce a high quality content, but at the center of it, you are the creative, you're the person in the loop.

Same for image, E, even more so for image, uh, and even more so for, you know, uh, video and 3D. So those are the world where we see, uh, you know, it's totally possible that you'll get a system that give you the end-to-end result. But what happens next? Uh, where do you, uh, kind of. Put yourself into that creation process afterwards.

And how does your content stand out? Compare to everyone else that just click the prompt and, you know, five times and dump that onto social media. 

Great answer. Thank you. 

Yeah, a lot of what you're saying kind of feels like it echoes what I was hearing, uh, Ben Affleck talk about on the InterPositive, uh, acquisition today from Netflix.

And, um, and then also when they showed some of the behind the scenes clips, I saw Comfy on, uh, a lot of the screenshots from. The stuff they were doing, so, yeah. Yeah. I think it, you know, ties into your balance of control and having. Human in the loop. 

Yeah, huge fan of, uh, Ben's view in here. And obviously, like I, I imagine he's, you know, kind of, uh, at least familiar or aware of the product.

Uh, Ben, if you're listening to this podcast, uh, we'll love to talk to you, 

Ben. If you're listening to this podcast, we'll have to talk to you too. 

We'd all love to talk to you. 

I said it first, Ben. My email is, um, 

Yolan get dips. We'll go after I was there. Anything else you didn't, we didn't cover Yulin that you wanted to?

Yeah, some fun, uh, usage, uh, of Comfy. We've heard, uh, kind of intelligence agencies, uh, using Comfy, um, uh. Is that not much to comment over there? Not much I, I know about it. It just a kind of a government comment. Intelligence from folks who work, but yeah. Government intelligence agencies. Yeah. Like without specifically talking about like, you know, uh, government, it's just, uh, I would like to put it out there.

It's not an adversary kind of, uh, group in here. Even if it is, we don't know about it. It is a group that kind of. You know, talk to us about how they inside the, uh, you know, agencies have, uh, 200 people, you know, using or kind of like running pipelines of Comfy, which, yeah. Uh, that was pretty shocking to us.

And then in some of the financial analytics environment as well, that's another really weird ones. But beyond that, like, you know, gaming all of that, these are just the normal, uh, the usual kind of, uh, group that falls into this. But yeah, I was, first time I heard about it, I was like. Oh wow, okay. I, I did not know Comfy can be so kind of, uh, generalized.

Now I'm gonna be trying to brainstorm what their use cases are all day. 

Yeah, yeah, yeah, me too. Like, we basically kind of think a lot about it. I was like, are you creating deep fake content? Or you like, what is going on? Is this a, a defensive move? Is this offense move? It's kind of fun to, like, I would love to one day learn about the actual usage in here, but I doubt that will come out.

Well, they're generating propaganda. What else? 

50 years from now when the records are like declassified and you find out. 

Yeah. Yeah. I wish I, I hope they eventually declassify and I can look back that, ah, that's what caused, you know, the entire political, you know, uh, uh, problem in this region of the world.

All right. Well, a good place to wrap it up. Uh, thanks a lot Yolan. Really appreciate it. 

Appreciate 

you guys having you, 

Yolan, nice meeting you as well.