Transcription: OpenAgents Episode 116 - 3D Multiplayer

Source: https://x.com/OpenAgentsInc/status/1819071289740644563 Wiki source: https://raw.githubusercontent.com/wiki/OpenAgentsInc/openagents/Video-Series.md Media title: OpenAgents - Episode 116: 3D Multiplayer We demo an alpha build of the world's fi... Upload date: 20240801 Transcription model: gpt-4o-mini-transcribe Generated at: 2026-06-01T17:16:00Z

Machine-generated transcript. Review speaker labels and wording before using this as quote-grade source material.

[00:00] Speaker: So I built a tool for myself. I spend about 80% of my development time using this interface, and this tool happens to be in a world that I built using the tool. Hey, look. Uh-oh. Okay, there we go. Because, you know, I just need to have my wife playing right next to me. Okay, so the question is, how are 10 billion humans, 10 billion AI robots, and tens or hundreds of billions of AI agents going to interact with each other in a way that is transparent, inspectable, lets a user make sense of all sorts of the different data that's available, have tools, have interfaces? I don't think it's gonna be your standard kind of 2D ChatGPT-style ChatUI component, basic bullshit. I think that the UI patterns that we need are found in games. I'm a gamer. If you couldn't tell, the name of my company before renaming it to Open Agents Inc. was Arcade Labs. Played MMOs for many years. So yeah, guilds, reputation systems, multiple players, agents as NPCs. Because one of the things that we need, here's a tweet, and I don't quite agree fully with the premise here, but Rune, the OpenAI shitposter guy, says, I don't care what line the labs are pushing, but the models are alive, intelligent, entire alien creatures and ecosystems, and calling them tools is insufficient. They are tools in the sense a civilization is a tool. You know, maybe directionally accurate, but this guy says, yeah, the missing piece is persistent state and long-term memory. If you're interacting with an agent via some 2D interface that forgets your interactions when you disappear, or maybe it stores some things in a vector database that remembers, it's like, that's kind of cool, but if it's only learning from you, if you have agents that are only learning from you, that's not good enough. That's going to pale in comparison to agents that can learn from humans collectively and from other agents collectively. Having that occur in a transparent process that people can kind of wrap their heads around and not in these sort of like closed AI labs where you got to trust that they're not, you know, gonna end humanity. I think that there needs to be a game for this, or at least a gamified experience. Okay, so that's the introduction to the Open Agents MMO, you could call it. We are basically taking the auto dev code that we've had built and just folding that into here. That's mostly complete. Going to start, well, anybody can come now to openagents.com and create an account, and you'll be able to see people walking around even at openagents.com. Okay, so you log in. GitHub X, email address, username. We're using Clark for this. Passkey, blah, blah, blah. For GitHub interactions, you're gonna need to connect your GitHub. We'll have some more explanations about all that stuff soon. And then for the auto dev stuff, you'll need an invite code from me. We're gonna keep things kind of gated for some of the anonymous. Who's this motherfucker? Okay, yeah, give it a try. We're going to be adding things like in-game currency is going to be our sort of like Bitcoin-based currency called Joules, J-O-U-L-E-S. So like to power things that require heavy compute, you're going to pay with Bitcoin Lightning or we'll let you do credit card for an upcharge. And we're just going to start adding, been able to build this. I built most of this, like, in the last 24 hours. So we're building very fast. I'm excited to get more people like you in here and kind of helping to build out the world. But until next time, see you soon.