--:-- Anjney Midha: That's the story of our kind of creative progress as humanity: people build these tools, which enables the next group of people to focus on a higher level set of problems.
--:-- Anjney Midha: Just in the last three weeks since launch, developers have built over 20,000 new activities on the platform, and that's generating 4 billion minutes of user interaction per day. The scale is sort of mind-boggling. What's going on?
--:-- Anjney Midha: I think something like 93% of Gen Z plays games, whereas back when we were kids, it was super weird to be playing games. We're now seeing some really significant examples of companies being built entirely on Discord.
--:-- Anjney Midha: Bill Gates once said that a platform is when the economic value of everybody that uses it exceeds the value of the company that creates it. That definition sets a pretty high bar for the few companies that surpass it. But one company does come to mind, and that is Discord.
--:-- Anjney Midha: What officially started in 2015 can really be traced back to 2009, when Discord co-founder and CEO Jason Citron started building tools and infrastructure for games. Fast forward to today, and Discord now has over 200 million monthly active users. Some might even argue that the metaverse is actually here; it just doesn't quite look like The Sims.
--:-- Anjney Midha: Now, in today's episode, you'll get to hear from Jason, alongside a16z general partner Anjney Midha, who actually sold his company, Ubiquity 6, to Discord in 2021. There, Anj set up and ran Discord's first dedicated developer platform, including launching its partnership with Midjourney, all before joining a16z last year.
--:-- Anjney Midha: You can probably very quickly tell that Jason and Anj have this shared history, especially because they got to sit down together in our San Francisco studio. Discord became such a thriving platform. But what did Discord really do differently here? Together, they discussed community-driven product development, how Jason himself went from player to developer, and their focus on extensibility since the very beginning.
--:-- Anjney Midha: So with Discord's recent release of embeddable apps, what can we expect now that it's easier than ever for a developer to build? If I was to go back when I was in college, which was almost 20 years ago now, none of that stuff existed. I mean, I built games back then and it was like firing up C++ and reading the DirectX APIs and spending a week trying to get a window to open with a triangle on it.
--:-- Anjney Midha: Prior to this release, there was, of course, already a flurry of new applications built on the back of Discord, like Midjourney or Leonardo. So let's find out what's next.
--:-- Anjney Midha: As a reminder, the content here is for informational purposes only. It should not be taken as legal, business, tax, or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any a16z fund. Please note that a16z and its affiliates may also maintain investments in the companies discussed in this podcast. For more details, including a link to our investments, please see a16z.com slash disclosures.
--:-- Anjney Midha: I am so excited for this episode. Thank you for joining us, Jason Citron, CEO of Discord, and dear friend, former colleague, and probably the person I know who's tried to start game studios the most number of times. And you ended up building several successful platforms along the way.
--:-- Anjney Midha: So today we're going to talk about all kinds of things focused on developers, infrastructure, the future of the Discord platform. But before we get to that, for folks who might not be as familiar with the crazy story that led to here, why don't we go back in history? What was the vision for Discord when you first started out?
--:-- Jason Citron: So way back in 2012, I was sitting around trying to think about what could be an exciting business to build. Having spent most of my career—in fact, all of my career and my childhood—steeped in video games and multiplayer games, I had this hunch that multiplayer gaming and gaming in general was going to become much bigger than it already was at that time.
--:-- Jason Citron: Back in 2012, gaming was pretty big. But it was at the early innings of mobile and still trying to figure out where gaming was going to go. I thought there'd be an opportunity to build a communications app for people who play games that would span all the platforms and all the devices, as gaming would become bigger and more cross-platform. And that has largely played out that way.
--:-- Jason Citron: Today, in 2024, gaming is the largest form of entertainment, bigger than music and movies combined, growing fast. People love to play games. It's gone mainstream. I think something like 93% of Gen Z plays games, whereas back when we were kids, it was super weird to be playing games. So that was kind of where it started.
--:-- Anjney Midha: What were the moments where you were playing games and you went from being a player and a consumer of games as a product to going, "You know what, the tools I'm using here could be better"? What was the moment where you shifted from being player to developer?
--:-- Jason Citron: I fell in love with games when I was a little kid because they were a way for me to connect and spend quality time with people in my life. I remember sitting with my dad—I must have been four or five years old, this was like late 80s—and he introduced me to this game called *Where in the World is Carmen Sandiego?* on his old Packard Bell computer. I just remember being so excited about coming home at the end of the day and being able to sit with my dad and spend some quality time with him exploring this world and trying to find this crazy lady.
--:-- Jason Citron: Over the years, growing up, I played multiplayer games on consoles and then on the internet as that became a thing in the late 90s. Along the way, I met someone who basically said, "Yo, I know how to make video games." A friend of mine—I was like 13 at the time—and I said, "No, you don't. You can't just make video games." And he said, "No, no, check it out." We went to my computer and he fired up this thing called QBasic and showed me how to draw a circle on the screen. I was like, "Oh my God, that's amazing. I could make video games." That was when I became an engineer and a programmer and I learned how to code.
--:-- Jason Citron: Fast forward, I went to school, grew up, and got into an opportunity where I was able to start a company. Through the process of building a game on the iPhone, I actually launched a game the day the App Store opened in 2008—one of the first 50 titles on the App Store. That took off like crazy. As we now know, mobile has been the biggest computing platform in the world.
--:-- Jason Citron: Through that journey, I realized that I had made a fun game called *Aurora Feint*. The technology we had built—kind of like leaderboards, chat rooms, login—was something that other developers really wanted. At the time, many game developers didn't know how to build infrastructure. I had learned how to do infrastructure and also how to make games. So we spun out the backend tech and built this social network for mobile gaming. This was probably 2009. It was called *OpenFeint*. We opened up *Aurora Feint*. Lesson about branding from that: no one knows how to spell that thing.
--:-- Jason Citron: That was the first moment when I started building tools and infrastructure for games. That company did pretty well. Then in 2012, after I had moved on from that, I started again building another game. We began as a game studio in 2012 called Hammer & Chisel. I thought the path to building the communications app would be to start with a multiplayer game. I had this hunch that core long-form gaming was going to come to mobile in a big way. So we started building a team-based competitive multiplayer game on iPad in 2012. One thing led to another. The game didn't really work out. But through the process, in late 2014, we started talking about what if we just went to market directly with a chat app for gaming. My co-founder, Stan, kind of had the insight for what that concept could look like. We started building it in January 2015 and brought it to market in May 2015. That was how it all started.
--:-- Anjney Midha: There's a theme emerging here. At least twice now, you've approached building a product as an application developer. You can think of a game as an app, and you discovered along the way that there's a bunch of really hard infrastructure that needs to be built first, especially when it comes to real-time multiplayer gaming. The history of computing is such that usually real-time gaming is one of the most demanding infrastructure environments. You discovered that there were a ton of infrastructure problems along the way, and you ended up actually building those tools for other people to use. You've since built one of the fastest growing, biggest real-time communication platforms in the world, which is sort of insane to think about the scale of Discord.
--:-- Anjney Midha: When you play forward from that moment in 2015 when Discord came out to today, how has the community on Discord helped shape the roadmap? The people actually using the infrastructure, whether those are users or developers?
--:-- Jason Citron: Our community, our user base, has been part of the conversation of what we're making from the first day. When we started talking about building Discord, of course, we played a lot of multiplayer games ourselves. So we had a good sense for what the product should be and how it should work. But as anyone building a startup knows, we're building products in service of other people. So immediately from the beginning, we started talking to our friends and their friends. When we showed them even just mock-ups, we asked what parts got them excited and which parts seemed confusing.
--:-- Jason Citron: Very quickly, once we got a prototype off the ground, we started giving it to our friends and having them try it, seeing what they liked and what they didn't like. Oh crap, we had to rebuild the voice tech three times. We missed an important set of features that we thought maybe was not important, but it turned out it was. That was part of the ethos for how we built from the beginning.
--:-- Jason Citron: Over the years, we've continued to build products that way. We always try to come back to what we're hearing from our customers, from our users, and the different types of people who use our products. Then we mux that with what we're excited to build for ourselves, and of course, what we think will be great for us as a business because we are a company.
--:-- Jason Citron: Along the way, there have been many, many moments when large shifts have happened in our roadmap because of customers. I'll give you an example. Initially, when we built Discord, it was very focused on being a voice and text chat app for guilds—people who play games in groups of like 15 people. The max group size on Discord, I think, was like 30 people, maybe 50, so it was pretty low. We realized pretty quickly from talking to people that they wanted to use Discord as almost an IRC-like public chat room replacement. In that context, we saw people filling their servers up with 50 people, and then they were like, "I can't add more people. What's going on?" We thought, "Oh crap, we've got to make this work for folks." So we invested in raising the cap and adding more infrastructure to support that.
--:-- Jason Citron: Then developers started building moderation bots and extending these Discord servers with other capabilities that we never even imagined, made possible because we had an open API powering the platform. Along the way, many of these things happened: these communities got big, generative AI became a big thing on Discord, the crypto community was pretty big for a season on Discord. But throughout it all, gaming and playing games with your friends and hanging out with your friends was always the main thing that people were doing.
--:-- Anjney Midha: Even if they would go spend time in a public community or futzing around with generative AI?
--:-- Jason Citron: Yeah.
--:-- Anjney Midha: I think one of the most underappreciated things about Discord is that the product has found a way to do two things at once that almost no other companies are able to do at scale: have a singular focus on a particular type of user and their need—in this case, allowing friends to spend time together while playing games—while also making the platform and the product so extensible for other people to bring their own use cases to the platform.
--:-- Anjney Midha: I remember a couple of years ago, I was talking to Stan and he brought up a screenshot of the first version of the homepage you guys had put together. On the front page, day one, you had a call out for integrations and SDKs. You had an open API on day one as part of the hero marketing. Clearly, you were thinking about making the platform extensible for all other kinds of use cases 10 years ago. Where did that come from? Can you talk a little bit about the challenges of both building a delightful product for users first, while also maintaining this extensibility for other kinds of use cases that you may not have designed for explicitly?
--:-- Jason Citron: The extensibility was built in from the beginning. It's cool that you went back and looked at that. I think if you go to the Wayback Machine, you can find it still from 2015. The idea was that we knew people were going to want to build custom integrations with different games as part of thinking about what a group chat for gaming looks like. We imagined if you have an EVE Online corporation—a group of people playing this outer space massively multiplayer game—or you were playing Final Fantasy Online, a fantasy adventure game. These different games have data and things that you might want to pull into your group chat experience. But we knew we were not going to build all of these things for the hundreds or thousands of games that people might care about. So we thought, we've got to make it so other folks can integrate their custom stuff from their games into Discord.
--:-- Jason Citron: Related to that was this insight I had from being an observer in the gaming business for so long. In games, there's this concept of modding where people can mod games. A developer will create a game and then oftentimes ship with it the tools they used to make the game. I think id Software popularized this in the early days with *Doom*. It's kind of the first one I really remember getting big. I think they called them WAD files. There was this whole scene online where you could go download WAD files. Then they started packaging them up and selling them, and it added a ton of life to the game.
--:-- Jason Citron: Today, when we look at the top titles that people play, a lot of them actually began as community-driven mods: *Counter-Strike*, *Team Fortress*, *League of Legends*, GTA role-playing. The list just goes on. *Fortnite* began as version four of a mod that I think came from a game called *Arma 3* many years ago. This idea—give your community tools to create and customize and extend your game or your software, and they're going to surprise you and take it in places you never would have expected—seemed to me like how you make good software.
--:-- Jason Citron: When we built Discord, it seemed obvious to us that we wanted to create an API that allowed developers to extend Discord, to be more creative with it, to do things with it we never would have expected. Some of the things we expected were connecting to the EVE Online backend so you could have your own forum or pulling in world boss spawns as notifications. But the stuff we never expected was generative AI. Who could have guessed that?
--:-- Anjney Midha: Right. Now we're seeing, many years after you made those investments in the craft of designing a delightful API, fantastic tutorials, a great developer experience for people to mod Discord itself, we're now seeing some really significant examples of companies being built entirely on Discord. The Discord platform has 200 million monthly active users, and that's led to entirely new companies being built on top with the open architecture you described. People may be familiar with Midjourney as one example.
--:-- Jason Citron: Yeah, and there are a few other ones too. Midjourney is the most famous one. It's a canonical example of how this stuff happens. We have this open platform where we're allowing developers to customize it and extend it. The Midjourney folks had obviously been working on their model for a while. I think they tried bringing it to market in a few other ways, and then they just tried a Discord server and a Discord bot. This was an example of exactly what I was talking about. I never would have predicted that you would have, A, generative AI in the first place—what a crazy thing that we've created computers that can do these things—and then, B, that someone would figure out how to take advantage of the magic of a Discord server and our platform and build such a cool experience. There are a handful of these now. It's been a pretty cool thing to see.
--:-- Anjney Midha: Midjourney is one example of a generative model—a text-to-image model. We've also seen an explosion of text-to-music tools as an example of how extensible Discord is. What other kinds of use cases are you seeing emerge on the platform that you're excited about?
--:-- Jason Citron: The generative AI category is exciting for us. But we really come back to this idea that people on Discord mostly spend their time hanging out with their friends in these smaller, invite-only spaces with less than 15 people. In that context, when I think about generative AI tools and using our platform, I really think about what developers are creating that give groups of friends more fun things to do.
--:-- Jason Citron: One really cool thing we see with the generative AI stuff is people take the bots and bring them into their invite-only servers. Then they can use them to create and explore and work on projects with their friends in a more private setting, as opposed to being in the public chaos of some of these large servers.
--:-- Jason Citron: Other really cool experiences we see are things like ways to listen to music together. SoundCloud has a really cool bot that you can use to play music when you're in voice chats, for example. We're working with some other partners to try to bring more music to the platform. There are a bunch of games that people have made that are pretty cool, and many more are coming.
--:-- Jason Citron: We just launched a couple of weeks ago a new set of capabilities for the platform that will allow developers to go beyond the text box and build these rich, visual, interactive experiences powered by HTML5. You can build a web app and essentially deploy it into the context of Discord. We're seeing lots of really exciting stuff starting to get built that I think will give people a lot of really fun things to do with their friends.
--:-- Anjney Midha: Let's spend a couple minutes on that. This is something you and I spent years working on together.
--:-- Jason Citron: Yes, we did.
--:-- Anjney Midha: Let's take people a little bit behind the curtain of what it took to actually go from the moment where we realized that developers wanted to express their creativity beyond just a command line interface—which was what bots were initially designed around as a form factor—and going from there to expanding the entire canvas for them to the whole screen, really, with web apps.
--:-- Jason Citron: Years ago, when the bots platform started to get popular, there was actually a Hack Week project at our company. We do this thing every year called Hack Week where we basically stop our normal work, everyone gets together, and we have a little creativity festival—that's the best way to describe it. I think it was 2018. We actually had tents in our office. It was a whole cool thing with food and stuff. One of the groups that year had the idea: wouldn't it be cool if we added an HTML5 canvas to our apps platform, and then people could make games and do other interesting stuff? There was a team that built this. I was always like, "That's a really cool idea. We've got to explore that." But as company building goes, you have a long list of ideas and you have to prioritize when you get to them.
--:-- Jason Citron: A couple of years later, the time was right to look at the idea. That's when we met. You were actually working on your own startup at the time, building this as a standalone project. After some conversations, I was like, "Man, we should just do this together." So we acquired your company. Then we created a team and really formalized this idea of how do we take our platform from the text-based era into the visual, rich, interactive experience era.
--:-- Jason Citron: To start, rather than just opening it up, we actually built a few games ourselves to really test the platform, make sure it was designed well, that it worked for players, that the interaction loops were good. That took a couple of years to sort that all out. Now we're at the point where we're ready to open it up. It's in developer previews. It's been a journey. It's been a fun journey.
--:-- Anjney Midha: Yeah. Okay, we're going to go a little bit deeper there. I remember one of the most exciting parts of working at Discord was that we had this incredible respect for infrastructure. The company had built its own WebRTC streaming service, had built its own voice and video infra. We built a bunch of serverless networking for developers for the apps SDK. I remember there are these moments in the product engineering cycle when you're building infrastructure that you inevitably have to ask yourself, "What is this going to be used for?" I remember having debates with you about how to answer that question and prioritize the most important features. We came up with this ritual of jam sessions, if you remember, where the team that was working on tools and infra and SDKs for the developers would come in and jam with you, where you would often role-play the developer.
--:-- Anjney Midha: So today, fast forward, I'm going to ask you to pretend we're in a jam session.
--:-- Jason Citron: Okay.
--:-- Anjney Midha: We just launched the Activities SDK, the embedded apps SDK, and now you're a developer. What would you like to go build with this entirely new set of capabilities that you've been handed by Discord?
--:-- Jason Citron: The way I think about this is first, what am I going to build? I try to think about what I as a consumer want. It's almost two steps past the infrastructure. Developers are thinking about serving their customers, so they have to take their hat off and then put on their consumer hat. For me, I started to think about what kind of games I might want to play and how they could fit into my Discord servers in fun ways.
--:-- Jason Citron: I would really love someone to build some multiplayer titles that don't require a lot of time investment but have really cool moments of storytelling that you can interact with your friends asynchronously around. That sort of weaves into the Discord experience through text and through the visuals as well.
--:-- Jason Citron: I have a small side project I'm working on right now. It's a game kind of like this—an arcade shooter that has a leaderboard mechanic but with a modern roguelike vibe. It's just random stuff I'm making as a side project to both dog food and test our platform. Part of this is how I get myself in the mind space of what our developers want, what their customers might want, so I can give good feedback to our team who is also doing this so we can serve people effectively.
--:-- Anjney Midha: One of the things that I think you've always been pretty good at is asking the question of what is possible on Discord that isn't possible elsewhere. You're often able to laser in on these specific capabilities that are ultimately technical primitives but exposed to users in a way that allows them to do something with their friends they couldn't do before, in a way that's easier or faster or better or more convenient. That's how things like persistent stage channels happened. That's how forum channels happened. That's how the embedded app SDK happened. As you're working and crafting your side project right now, what in your mind are the top two or three things that the Discord embedded platform that you just launched really shines at that's hard to do elsewhere?
--:-- Jason Citron: The magic, I think, is the fact that you have the social context of the space you're in. It's really, really easy for someone to pick up a title, and then that game can depend on the fact that there's a group of people that are connected to the person playing it. You have access to that data in a privacy-safe way. Unlike other games where maybe you log in or you get a user to come into your game and then you have to get them to build a friends list or invite their friends to play, in the context of Discord, because of the way the games work, they're instant and they have the social context.
--:-- Jason Citron: You could build a game loop where one person plays and then immediately the result of that play session is other people get exposed to the game. They can play it, and they don't have to go do anything to set that up because the friend brought it into the server. I think about cool mechanics like leaderboards and challenges that really depend on having this group of people that have access to this shared space, both synchronously and also on the go in bite-sized ways.
--:-- Jason Citron: It's a different kind of interaction model. I could open my phone, play a game, and then when I get back to my computer, I could pick it up. You could play from your phone when you're somewhere else if you want to come and compete with me when you have a few minutes in your spare time. This is the kind of model I'm playing around with with this title.
--:-- Jason Citron: Part of what I'm so excited about with these tools is to see the creativity that other people bring. I know what would be interesting to me, but much like when we designed the platform for Discord in the beginning, so many of the cool things that happened with it I would never have predicted. I'm most excited to see what people do that I don't even think about.
--:-- Anjney Midha: I think it's worth taking a beat there to just recap what you just said. Up until now, for a developer to build a cross-platform experience like that, it takes years. The end result is that a product or an app—whether it's a game like the one you're building or a non-game like Midjourney—can literally go from zero to... I think Midjourney is now at more than 20 million users in the server or about there.
--:-- Jason Citron: Yeah, I think their public server is about that size. It's pretty large.
--:-- Anjney Midha: In essentially less than a year.
--:-- Jason Citron: You should write the marketing brief for that.
--:-- Anjney Midha: Well, I mean, you did. I guess you actually did.
--:-- Jason Citron: You're spot on. I think that's what's so magical about it. The trends that entertainment and gaming in particular have been on for the last decade are cross-platform. People are more and more expecting games to not be tethered to devices. It's just a screen. You should be able to switch screens and play the same experience. You want your content and your progress to go across those screens. You want your friends and your social graph to go across those screens. You want the things you bought to go across those screens. These trends are happening in the world broadly in gaming too.
--:-- Jason Citron: What we're trying to do is package that all together in a really easy to pick up way. Not only do the games have all those things, but we also bring the social graph to the game. We manage auth for you because it's built into Discord. It gets deployed on every platform that we're on, which is most of them. We have payments built in on the platforms where the platforms don't, and discovery and all this stuff. I'm just excited to see what people can make.
--:-- Jason Citron: That's the story of our kind of creative progress as humanity: people build these tools, which enables the next group of people to focus on a higher level set of problems. We've abstracted out some of this infrastructure for folks so people can spend more time on their gameplay and the creativity of that than futzing around with auth and building social graphs and all this kind of stuff.
--:-- Anjney Midha: One of the most amazing things about building in an openly extensible way is that you get the combinatorial creativity, the explosion of your primitives, your platform, with other tools that may be coming online or maturing at that moment in time. Sitting here in 2024, if you just contrast it to the version of Jason in college who was building his own games...
--:-- Anjney Midha: And you hand him all the open infra of Discord's developer platform and the recent explosion in these creative generative models that can allow you to turn text into images and text into audio and allow code generation and so on. When you look at how the production pipeline of building an entire experience like a game has changed with generative models and you combine that with the Lego blocks of Discord, what are the kinds of new game formats or new types of interactive entertainment that you think are possible today that just weren't possible maybe when you were in college?
--:-- Jason Citron: There's so much that has changed. It's kind of amazing. Between the generative AI tools, social and distribution infrastructure like app stores and something like Discord, plus the modern game engines—we can't leave those out. The kind of game engine you can get off the shelf today, whether it's Unity, Unreal, Godot, Phaser, all these things, they're just incredible. If I was to go back when I was in college, which was almost 20 years ago now, none of that stuff existed. I built games back then, and it was like firing up C++ and reading the DirectX APIs and spending a week trying to get a window to open with a triangle on it.
--:-- Jason Citron: I think what ends up happening is because people can be so much more productive and the markets are so much bigger, we're going to start to see more and more games that are focused on more and more niche topics and niche mechanics. Big games get big budgets, so they tend to be less risky. Smaller games can be more risky because the budget dynamics are different. So you could imagine a world where one developer could build an entire game like Stardew Valley, which was one developer, but I think it took him like five or six years. The next Stardew Valley might be built by one guy in a year. What that means is you may get 10 Stardew Valleys and they may all have different themes and topics. So we may all just get more entertainment that's customized to our particular proclivities because so many more people are making games. The cost is down and the markets are bigger, so there are more people than ever who are looking for this stuff.
--:-- Anjney Midha: This is also, I think, one of the most underappreciated parts of Discord. It has unlocked paradoxically niche at scale. Through the server context, there are now thousands and thousands of niche communities on Discord who have then found people who love each of those niches globally. What always struck me when we were looking at the activity in these servers and what kinds of apps and bots they were using is the extensibility of the platform allows those niches to do things with the platform and build a bot or an app that's custom designed for that niche community's use case. I think there's one you told me about a while ago, which was the Harry Potter fan fiction server. It had an app that that community had built for the friends who were huge Harry Potter fans to role play being at Hogwarts.
--:-- Jason Citron: Yeah, yeah, yeah.
--:-- Anjney Midha: So if I'm asking you to channel your inner Will Wright, for example, one of the most successful game genres of all time has been simulation. The Sims, all the tycoon games that allow people to express this world building desire where they're able to almost use games as a tool for creativity, where the game is itself building and creating with other people. Roblox has done a phenomenal job at doing that in a pre-generative AI world. Minecraft. There's Minecraft, Roblox, there's The Sims. Given how massive those genres were in a pre-generative world, when you give the generative models to a developer, plus the insane distribution of Discord, 200 million people on day one, do you think we're going to see a new kind of genre there of something that blends simulation, like The Sims, with real life, with your real friends group?
--:-- Jason Citron: It's entirely possible. I think these things are really hard to predict. As someone who makes more on the tools side of things, what I think is definitely going to happen is that because it's going to be so much easier to create, we're going to see more random stuff. So the chances of something interesting happening are going up. But the things that cause new kinds of game genres to emerge are oftentimes changes in distribution, business model, or production capabilities. In this moment, we're seeing some of these things change. Discord is offering a different kind of distribution mechanism with different context. Generative AI is definitely changing the landscape for how people produce games today. So I think we're probably going to see something interesting happen. But yeah, I don't know. It's hard to predict.
--:-- Anjney Midha: This is the most fun part about working on dev platforms. You get surprised.
--:-- Jason Citron: Yeah.
--:-- Anjney Midha: Okay. Just to take a step back for context, the Discord team finally launched in full general availability after years of crafting and honing the developer experience, the embedded app SDK, on March 18th. If I have my numbers right, just in the last three weeks since launch, developers have built over 20,000 new activities on the platform, generating 4 billion minutes of user interaction per day. The scale is mind-blowing. Why is this resonating so strongly right now? And what are the top emergent behaviors you were seeing in the first few weeks?
--:-- Jason Citron: I think it's resonating because developers intuitively understand a lot of the stuff we've been talking about. They're looking for distribution channels with captive audiences on the other side, with low production costs, so they can explore their own creativity and build products for themselves and their friends and bring those to market. The games industry, I think, is in an interesting situation right now in particular. So this new channel, we thought people were going to be excited about it. Then at Game Developers Conference a couple weeks ago when we announced it, I was actually surprised at how much it resonated with developers. We had a couple talks there and the lines were out the door.
--:-- Jason Citron: As far as what people are building, a lot of the things that I know folks are working on have not actually released yet since we opened that up. So I don't want to say anything that hasn't come to market yet. But I think that 20,000 number sounds big and it's exciting. In reality, what it signals to me is a lot of excitement and curiosity. We'll see how many of those come to market. I suspect there are probably 20,000 people poking around. I think there are thousands of them that are actually really making something. So I expect we'll probably see hundreds of those things come to market over the next six or eight months. Without getting into specifics, some of them are interesting new ways to stream games, some are new games, some are interactive experiences, like ways to enjoy different types of entertainment together. Some of them are silly things like comic book related projects. And then there's a lot of stuff in there that I actually haven't even seen because there's so much.
--:-- Anjney Midha: I was talking to a developer at GDC who is working on a Discord embedded activity that is tinkering with some of these new generative models that you were describing. One of the things he crystallized for me, which I think maps the experience we had in the early days of MidJourney launching on the platform, was this idea that game development and AI app development have this very strong similarity in the development process. You pre-train a model and then you put it out with your community. Then what the community does with the early days of the model and what outputs of the model they prefer helps you run a reinforcement learning loop to improve the model and give users what they like. That's very similar to the game process, where you put out a soft launch title, usually with LiveOps, and then you start seeing which parts of the multiplayer experience your community likes. Then you pipe that into future LiveOps releases. If you had to describe why the Discord platform has found so much success with generative AI developers, is there fundamentally something similar about the game development production process and AI app development that has resulted in Discord hosting one of the world's most successful consumer AI businesses right now, which is MidJourney?
--:-- Jason Citron: That's an interesting point. I think both of those things make sense, but I actually think if you zoom out a little bit, there's another interesting trend that this is part of. I think it's this idea of consumers wanting to be closer to the people creating the things that they use in their lives. It's a broader dynamic of co-creation with consumers as you're building products. In the case of AI, there's quite literally a direct feedback loop where a lot of these models are using thumbs up and thumbs down type reactions on the outputs to directly feedback and improve the model. In the case of games, it's a little one step removed where perhaps the players are talking with the developers, using the products, and giving them feedback. I know a lot of devs use Discord to do early play tests. They hop on voice chat, show off builds, and spend time with their users. But I also hear startups doing it and other companies doing that too, not just games.
--:-- Jason Citron: I think gaming and AI is kind of at the forefront of this, but we see lots of other tools. There's a command line app that I use as an engineer called Warp that has a Discord server. They hang out with their community in there and talk about feature improvements and how to make their product better. It goes on and on. I think the trend is that consumers want to have say and influence over the products they build. As a product creator, having that direct line of feedback with a tight feedback loop with your early adopters really helps you shape what you're building and make it better. It ends up being this really powerful back and forth where you can improve your product. People get excited about it. You build evangelists. Then when you do go to market and launch, you have this built-in community energy that can help spread the word. Sometimes those things will lead into Kickstarters and Patreons and other stuff. There's this whole community-driven product development thing that I think Discord is part of, or maybe helping drive in some way. It's another one of those interesting emergent things. You build these platforms and these tools with something in mind and then other interesting things can happen with it. We really started focused on being a place for people to come together and play games with their friends. That is still the bulk of what people do today. But all these other interesting things happen, like companies setting up servers to do co-development with their consumers.
--:-- Anjney Midha: Wow, that's super cool. There's this company that I have the chance to work with as an investor called Luma. It's a generative AI model. Before they even had a website or a mobile app, they launched as a Discord app. In three or four days, they had 30,000 to 40,000 people in the community show up. Half were from the games industry and half weren't. They started using the model in ways that they didn't expect and allowed them to realize there was a much broader set of uses for their tool. That began a dialectic that allowed them to change the focus or tweak the list of priorities in their product development roadmap, then ship that to that Discord user base. When I saw that happening, I realized there's this art in software development of finding product market fit early on. We used to talk about this concept of the highest expectation user, the HXC. It is remarkably hard to find HXCs to take time out of the day to get attention of users today early on. But Discord is such a phenomenal tool at aggregating these people in one place who care about what you're building, that the speed at which you can iterate with them is unbelievable.
--:-- Jason Citron: The app development process has three steps. You ideate with your community, then you launch, and then you find a way to actually monetize and turn them into a sustainable business.
--:-- Anjney Midha: You've solved the first two parts of that journey. What do you think the last mile looks like on Discord?
--:-- Jason Citron: Our focus right now, starting with this embedded apps launch a few weeks ago, is to bring to market the full loop of how you, as a developer, build a sustainable growing business on Discord. Right now we have all the parts and much of it is in developer preview. It's going to be rolling out over the next few months. That will begin with how do you get your game listed in our app directory and in our app launcher, which sees millions and millions of people every day coming there to find fun things to do with their friends and ways to customize their server. Then once they have added your app, how do you make money? We have in our own titles, we're running payments and we have payments available to some apps today. So you'll be able to directly monetize. If Discord's on a phone, that will run through the phone payment systems. On desktop, we have our own stuff that we've built that you'll be able to plug into. You'll get the expected monetization hooks that you'd want out of the box, and then re-engagement through there, and all the backend dashboards and reporting that you'd expect. You'll be able to build, launch, monetize, and do that whole loop as a developer on Discord, and your apps will work.
--:-- Jason Citron: This is another one of those things where when we went and talked to a lot of the people who make apps and bots on Discord today, many of them have been over the years using off-platform payment mechanisms to try to cobble this together. Whether it's setting up their website and implementing something like Stripe, or trying to get people to go to Patreon or whatever. All of these things are super high friction for customers. They have to go off platform, log in, do a whole bunch of nonsense, and then you have to build all this stuff and manage it. So what we've done is we've created an easy one-click solution inside of Discord that works just like you'd expect any app store to work. A consumer can either purchase a one-time transaction in your app or make a subscription, so you can decide how you want to monetize your service for whatever makes sense for you.
--:-- Anjney Midha: Looking back over the 10 years that Discord has been around, the story of Discord has been consistently observing what the biggest pain points are of people trying to communicate and do things they love with their friends and making it just 10 times easier. This craft of giving people a way to do what they're already trying to do by duct taping or combining different tools all in one place, while making sure Discord doesn't become bloated, doesn't become slower, doesn't become more expensive, has been this remarkable journey of ruthlessly making what people are trying to do already easier and easier in one place.
--:-- Jason Citron: That's the journey of most great products and services: how do you make it so that whatever someone's trying to do is better, faster, and cheaper. We often talk about removing objections or reducing friction in the process for a person who's trying to accomplish something. Whether that's an end user who wants to open their app and be able to quickly message their friends, or a developer who's looking to build and deploy their app or some creative project to people. We love that we get the opportunity to wake up every day and help people spend time with their friends and play games and enjoy life. That's what it's about. I can't wait to see what people build. Maybe we'll check in a year from today and instead of 20,000 apps, it's going to be 200,000 apps. We'll see. But you've got some pretty amazing stuff already on the platform. I'm so excited, personally as a developer, to get started on my side project this weekend.
--:-- Anjney Midha: Cool. Well, thanks for having me, Anj.
--:-- Jason Citron: Thanks for coming.
--:-- Anjney Midha: If you liked this episode, if you made it this far, help us grow the show. Share with a friend, or if you're feeling really ambitious, you can leave us a review at ratethispodcast.com slash a16z. Producing a podcast can sometimes feel like you're just talking into a void. So if you did like this episode, if you liked any of our episodes, please let us know. We'll see you next time.
