LIVE: Peter Steinberger & Dave Morin | OpenClaw Fireside @ GitHub HQ

Heat. Heat. Heat. Heat. Heat. Heat. Welcome. Welcome. We're about to get started in just a few minutes. If everyone can grab some seats, make your way to the front. We'll get started in just a minute. Thank you. Good. I know. I'm going to be like 2 minutes. What's that? Amazing. Folks can come grab some seats, please. Can we turn up volume a little bit? All right, thank you all for joining us tonight. Welcome to OpenClaw after hours at GitHub during Microsoft Build. If you guys could quiet down just for a second. Thank you. Thank you for joining us all. We're super excited to have the community here to join us during this event to celebrate some of the cool work that's being done with the OpenClaw community. Uh I'm going to kick it off and hand it over to PJ Mets who's going to be our MC for the night. Welcome PJ, please. Holy smokes. Hello everybody. How are we doing? Okay. Well, look, look, I know this is going to feel like a a high school motivational speaker, but I know we can do better than that. So, one more time, how are we doing? Holy smokes. This is a fantastic room to be looking at. Look at all these people. Uh I can't believe we're doing this event here tonight. Uh I wore my bestest red hat for it. We turned on our red lights for you. Obviously, there's a theme going on, right? So, uh, I'm here just to talk about GitHub and for a brief second. My name is PJ Mets. I'm a developer advocate on the education team. I work with students and teachers all over the world making sure they know how to use GitHub. Uh, this is an event that like 8 months ago we had no idea we'd be at, which is kind of cool. Uh, and and to be honest, like I didn't think I was going to be here till about an hour ago. So, I'm super stoked to be here with you all tonight. So, what I need to do is I need to ask you all uh some questions as my watch alarm goes off. So, I've got some uh some special swag here. It's playing cards. So, who here likes to gamble? It's 2026. Everyone's gambling. So, what I'm going to do is I need to figure out like what some of the best claw names are out there. So, who's got a really good name for their claw and wants to share with every Oh, your hand was up immediately. Look at this. All right. What's the name of your claw? Alfred. Why? like uh like Batman's assistant and also like the app that like is like an assistant. So yeah, we we get one Batman reference. That's solid. Does anyone else have a superhero themed one here? What's your name? What's your claw's name? It's called Clawax.ai. Open claw to the max. That's the whole name. Clawax. Oh, I was like, you have to say the whole thing. Like it's a long name. All right, you're getting some cards for sure. Longest name in the world. All right, hit it. My claw's name is crabme. Crabme. Like crab memory. That's pretty good. Crab. But it's supposed to be a lobster, right? Yeah. Yeah. Lobsters, crabs, it's all the same thing. I got to get up here. I'm sorry. I got to actually talk to some people up here. I hope this microphone doesn't feed back. We're going to learn today. All right. What's your claw's name? Clawaxing. Claw maxing. Of course it is. Naturally. Naturally. You got one shouting. Mine's Virgil as in Dante's guide through hell. Hold on. That was really good. Virgil, like Dante's guide through hell. Fantastic. If I ever go to hell, I hope I have a guide. It'll be difficult otherwise. Okay, I got one more, but hold on. Like, this one's got to be really good. So, think about your claw name. Is it really good? You're standing up. You've got a shirt on, too. So, all right, let's do this. Your claw name is Paulie because of the San Francisco parrots. So it's parrot like Paulie. So we call it I love that. Yeah. There's these parrots here in San Francisco. Enjoy your cards. There's these parrots here that fly around. They're so pretty. That's a great name. This solid. So I'm stoked to be in front of y'all. Um little personal information about me. My wife's nickname is lobster. And it's not because of Open Claw. But now that all of tech is obsessed with lobsters, I see this stuff everywhere. So I'm like, here's a pin. Here's a sticker. She's like, "Where are you getting this from?" I'm like, "Don't worry about it. There's cool stuff happening in the tech world." So, we're super excited to have you all here. Uh, this GitHub space, this is called the, by the way, get together space. You're welcome. Uh, we're super excited to have you at this building at HQ and to see all of these community people coming together on something that again did not exist eight months ago. And yet here we all are connecting with each other and doing amazing stuff and doing amazing work on brand new cutting edge technology. So what I want to do is I want to bring up our first person and our first person is a very important person to talk to you all. This is going to be Kyle Dagel coming up to talk to you all about GitHub. So Kyle, please come on up and start us off please. Thanks BJ. How's everybody doing? Yeah, you can feel I don't know for those of you that have been at build the the past two days, you might be in a metered level of energy after expending a lot of energy the past. So, I don't know how PJ does it, but I'll just say uh yeah, uh welcome to uh GitHub HQ. It's so good to have you all here. This whole first floor space is meant for this. We we do get togethers, but our teams are remote and so the reality is uh we're all actually just looking at a video call. Uh we're just bringing community in here four days a week roughly. So I'd love just everyone to give an enormous round of applause to Janice and the office team for putting this together. All the fun treats and snacks and just making any of this possible. Uh it's a huge uh labor of love for her. Uh my claw is named Baxter. Uh I don't have a good story for that but I thought it would be you know uh just an easy name to keep referencing and my team affectionatally now uh talks about Baxter in Slack uh because there's many times where I'm preparing for things or going through my personal notes I say I'm waiting on Baxter and the team kind of laughs and says okay well what did Baxter say and I paste things back. I'm personally very excited about like the execution container environment so we can get either the claw into my work computer uh cuz right now it is sitting on the Mac Mini that I bought like everyone else while I was on a work trip and it plugged into my hotel room and then ultimately I brought it home. I think that experience having been at GitHub for 13 years now myself is like something we've all had that moment of the moment where it finally all clicks and the magic happens you know for you personally and at GitHub we've always known that's done by people that's done by maintainers someone had to sit down and figure it out put it together and whether that be us you know just providing the platform or us responding to Peter the team every time a new API rate limit gets hit and we desperately try to figure out how to cobble together a bigger limit uh to support the many many claws running. Uh that's just what we're looking to do just to support the maintainers in the community to come together build something magical. And look, I mean you all are the lucky few that got in. I believe there's 1,700 people waiting to get the magic email. 2,000 3,000. I mean the number grew precipitously in the last 4 48 hours. So, it's so good to have you here. I won't waste any more time. I'd love to introduce Omar, Dave, and Peter to come on up and chat about OpenCloud. Hello everybody, welcome. Um, my name is Omar Shaheen. I'm a corporate vice president at Microsoft on Microsoft Scout uh which we announced yesterday at Build and is our open claw powered uh work assistant which I'm super excited uh to bring to market. Um I'm very pleased to introduce uh Peter and Dave. Uh thanks for being here. Um and I have 34% battery life to ask you all these questions. Um all right. Yeah. So, first things um I want to share a little story. So, um I crossed paths with these two in kind of interesting ways. Um, I was a customer of Peter back in 2012 when we licensed uh PSP PDF kit at Microsoft to power our PDF uh rendering for one drive mobile apps and I was like super impressed like he's just a builder uh amazing engineer and um so when he popped up and OpenCloud popped up I was like I'm going to go check that out. And then Dave built amazing things. Um, Path was a social network uh that was super awesome. My wife and I used it back in the day to sort of share pictures with each other. And then Facebook Connect um which many of you uh may know is like one of the most explosive developer platforms ever built. Um so it's like to have this opportunity to be here, spend time with these people in this new in these new roles is like it's a privilege for me and I'm just I'm excited to get started. So, um, going to get started with some questions. Peter, you've done a lot of podcasts and I've listened to every single one. Uh, so this is going to be a 400 level, uh, question. You listen to all three and a half hours of them. All of them. You have to really challenge him. Come on. All right. Let's um start with a question I'm super curious about. How did the two of you get connected? Oh, that's a great story. Um, you know, I did not expect that OpenClaw would create all of this. And that's honestly like the thing I'm mostly proud of. like having a little bit of this old builder energy back that was kind of like there 20 years ago where people were excited about the internet and then suddenly it almost feels like the '9s to a degree where people are excited again about like building something on themselves and like owning this thing and like buying their own hardware and sinkering with it. And that's really what I that's really what I wanted um to achieve as a project like you own this thing you hack on this thing remember reminder for the first two months the only way to install openclaw was like get clone pmppm build and then gateway run that that was the installation way and it got me a lot of PRs um and And basically overnight uh everything exploded. My phone number leaked online. Like people were calling me. Yeah. Some of these reporters figured out if you just call someone often enough, the iPhone thinks there is an emergency and will still ring even if you're in sleep mode. Uh I did not know that. um my Twitter DMs like you'd scroll down but at some point it just like it just like bugs out. I think I still haven't read all the DMs but but like luckily uh Dave was in in my DMs as well and like what the [ __ ] this guy? Um but he seemed like interesting and he was like trying to push me to do something I didn't want to do. Um but then we had a call for me it was in the middle of the night. I don't know for sure evening for you and I kind of we heard like a thing. Yeah. I mean he's married but I could I could feel like he he was claw killed. Um and we just kept in touch and and I guess the stars aligned. Um, I felt that he's he's a good one and he helped me a lot with oh my god all the legal stuff on the foundation. I like if you think like making a company is hard, try to make a nonprofit. That's like company in hard mode. It's insane. Like like everything is more complicated. Just getting a bank account, especially as an alien of extraordinary ability, uh is is is everything's very challenging. But it was important to like get the structure right. Uh so even if I even if I sell my time to someone else like the project stays independent. Um and we actually we got some nice donations. We could hire people. Things are going in very exciting ways. I actually get some sleep again. Yeah. That's a quick story. And Dave, why did you reach out to Peter? You know, when I was 7 years old, my grandfather put a Macintosh Plus on my desk and I fell in love with this idea of the computer and I started building things with HyperCard and Apple Script and you know just like it was this feeling that I could create anything. And then I went on and you know you've talked about it like I got to work at these cool places. I worked at Apple, I worked at Facebook, I built this developer platform, and then I built Path. And all along the way, there's always been this kind of uh chasing this feeling of being able to create something myself. And I don't know, like Peter was saying, for like the last 10 years, the valleys kind of felt like the builder energy was sort of lost in crypto and I don't know, SAS and like these things that like business became the conversation that was happening in all these buildings around town. And a friend of mine, an old hacker friend from the '9s, um, reached out to me when I was at home over the holidays. Um, I was in Montana. I grew up in Montana. And he was like, you should install this Cloudbot thing. And I was, Careful, the D is silent. I know. Uh, and I was like, I don't have time, you know, I'm I'm busy with my family, like trying to take some time off. But I I decided to install it the next day. And within 24 hours, I was just like, that feeling that I had when I was seven was back. And I was just like, I have to meet the creator of this. Like this, not only does this have to be protected, but we have to get this to everybody in the world. And I started just making more things with it than I've made in a really long time. Not because I didn't want to, but I think there was something about the speed at which you can do things now that it keeps up with your creativity. And for whatever reason, it fit my brain. And so I reached out to Peter and I just said, I love this and we have to protect it and make sure the community uh you know is protected over the long term. And so that's kind of been the conversation ever since. And we've ended up in a bunch of strange countries together now uh everywhere from Tokyo to, you know, all over the world. So it's been really amazing. And like you said the you know when I was building Facebook platform we had an enormous explosion of developer activity. Um it took us a a year to reach a million users uh million developers. I think that happened this time like in 2 weeks. So, we've been kind of trying to keep up with it and it's been an enormous amount of work and a lot of phone calls between Peter and I where he was asking to sleep more and I'm like, we're we're trying, you know, um we've been trying to just make things happen and like you said, like doing this nonprofit is been entrepreneurship on expert mode. So, we've been doing our best and we've made a lot of progress. So, excited to talk to you all about it. I love that story and uh Hyperard was awesome. Um all right, so uh Dave, this question is for you. when you look at OpenClaw like as an investor ecosystem builder um what you know what told you that this was more important than just a developer tool like sort of you spoke a little bit about that builder energy you felt and going back but like when you think about like the whole world like how do you get the whole world to pay attention to something that's as powerful and magical as open claw I mean I think one of the first times that night Peter called me and we were talking there was a bunch of really intense stuff going on. He had a lot of people coming at him and I remember saying to him, I think you found Linux of AI. And I really sort of felt that way in that um you know, we've been we'd all experienced these other AI systems, right? We've been using chatbt, been using claude or whatever, and none of them really felt like something that you could build on top of that you could take not just and customize, you know, a little sort of thin shim on top, but that you could actually take the entire thing and build something substantial on top of it. And not only could you do that with this, there were a lot of core components that you could tinker around with like that Peter I think so thoughtfully uh designed, but you could also think very broadly about what you could build build with this thing. And so I was just like there's nothing else like this. Like I I sit there and I receive a lot of pitches. I talk to a lot of entrepreneurs. I've invested in over a thousand companies um over the last uh 20 years. I've seen a lot of different things built a lot of different ways and there was just nothing in AI. There was a lot of like let's build a vertical version of this thing on top of the model. Um but then the models would like eat that entire category of startup. And so everybody in the investor community was like how do we even invest in applications on top of this? And so open cloud was the first time where I was like I think I can see 10 years of uh and beyond of entire companies, entire people's lives dedicated to building things on top of this and that maybe we are kind of in 1997 and the web was just invented and agents are kind of like websites and we've got the framework now. And that's where our conversation really accelerated cuz we were like, if we can make this like the platform that people can go build agents on top of for a million or billions of different reasons, then that could be very powerful. And so that's been the ongoing conversation. And I guess that's that's the long answer to your question. Awesome. Thank you for sharing that. All right, Peter. Um, I want to ask you about scale. So open clause had a huge amount of demand in a super short time. um like what have the last few weeks taught you about scale? Um specifically also about community and expectations. That's a good question. I mean you 400 level man. I mean I mean scale is I feel the biggest issue on scale is still people like and attention. You can you can automate so many things but you cannot really automate taste or thoughtful design. And then the other the other thing is you can add more people to the project but it doesn't automatically mean things go faster. I mean, it's like a lesson that we that we hear so many times, but still humbly I'm reminded about it every time I try. Um, and then again like the meta level of open claw I also find so incredibly interesting because we're hitting we're hitting so many issues like like Kyle mentioned it before like I I have I have a Slack channel now to like the GitHub Fox where I can ask them to like press the button and pressing the button means resetting my rate limits and I think at some point one of the team like automated with a claw and got in big trouble. Um so the whole way of like how we build open claw all the little things we we we we build to automate and to help test it. Um, that's really the unlock for for for more scale. Like now I can my my my my co my coding sessions went I guess from 10 to like 56 when the models got faster. Now they're kind of like at 30 again because I I I found ways to like close the loop even further and like to end to end test everything uh and like also have auto review and then everything takes forever. My trust in what I build is so much higher that my attention span usually I only need it one time when I discuss what I need and then even the end to end verification is done by the system so I I scale better but I still I still I'm still human that's a problem. Yeah. Awesome. Um so I'm I'm actually quite curious about the foundation. Um and I wanted to ask like how you see the role of the foundation. Um and what can the foundation do for OpenClaw? Yeah, I mean we set up the found I mean when we were first talking about it uh it was very clear that this is an open source project and so we need to do this nonprofit and the goal of a nonprofit at the end of the day at least in the United States is to serve the public good. And so what is the public good? The public good is the maintenance and continued engineering of the openclaw project, right? And Peter and I have also talked a lot about that there's this secondary thing that might even be a higher level mission than the actual project itself, which is that if you look at the research out in the market right now, um AI is currently less popular. Not it's more it's less popular than ICE and Jeffrey Epstein in the United States which this is a serious problem right um and what's the most amazing thing about openclaw I mean look at everybody that's here right now we've seen like we said people in every all kinds of countries we have different stuff going on all over the world and everyone's showing up because it's fun because it's a little bit weird because we've kind of brought this builder energy back to things. And we're bringing people closer to AI by showing them that they can not just interact with it and talk with it 10 times a day that can build anything that they want. They can build a business. They can build they can make beer. They can, you know, do all kinds of interesting things. And so the foundation can also play a role in trying to change the narrative um and to help people get closer uh to AI. And I think that's something that really I think bonds Peter and I quite a bit and is something we're trying to put into the uh into the foundation. Um and then you know of course the foundation can also provide a livelihood and so we're super excited. We're going to talk about a little bit more in the next couple of weeks but we've got several people on board full-time now. Um we've got an amazing set of people involved that um has been largely uh due to that we've brought in some very large donations or actually we have to call them sponsorships. Um and uh I've been able to bring you know my world I spend a lot of time in the world of capital and putting together resources to build projects. And so I've been able to bring together a lot of different types of uh people. We've had the University of Michigan show up. Uh I know there's a lot of go blue people out in in around the valley, but the University of Michigan gave us an enormous donation. They also started a a a institute for agentic engineering uh that'll start teaching classes this fall. Um we have uh Mount Cyani about to come on board as a sponsor and they want to see scientific discovery for neurological diseases and so we're we're receiving a lot of really interesting conversations that are beyond just the usual suspects. And so I think that that says something about what we're trying to accomplish. I don't know what else you would add, Peter. I would make the language stronger that it not it's not maybe, it's a it's a definitely. Um, also realizing that we only finalized this structure a few weeks ago. So I'm very excited in the future to like do more events. Um, not just like also full day hackathons where like we bring people together that that build stuff on top of open or any other agents like who cares in the end in the end it's about it's about getting people together that are excited about what we can build with the technology that's already out there. That's like my thesis almost. Um, we don't need better models. Like the main limiting factor right now is imagination. Because even even if the models would not get any better, we could change the world for the next 5 to 10 years easily just by cleverly applying and and building things with the tech and the intelligence already out there. That's awesome. Um, and you're right, like not seeing the usual suspects is kind of also an interesting angle to like who's paying attention. This is not the normal sort of folks that are investing in, you know, capabilities like this. Um so going back to the imagination point um you know one of the things one of the privileges I have I think is like waking up in the morning firing up Discord going to the main standard channel and seeing like what happened in the last 7 hours and you know you you are usually um fixing things and then inventing new things um all all at once. Um what are you most excited about? Maybe something that you like want to implement but haven't had time yet. like what's the thing that you are scratching up against? Ah, so many things like the I'm still like now finally we have really good people on board. I can I can give them more stuff. So I feel slowly I get my my time and attention back and actually trying out more things as I as I did in the very beginning and then the the flood of people came and like all the all the issues came and I drowned for a while. Um, there's multiple things that really excite me. I think the current way we do memory is not great. It It works. It's surprisingly effective for how it simple is, but I have some ideas. This could be a lot better. also right now the the agents does regular tool calls but the whole world is moving towards code mode and we already have experimental code mode including including like interesting ways how to like convert MCPS to just be a library and like maybe even make it so that the agent can like grab API calls. Um I already added some stuff. It's not on by default. The evolves I did so far were like showed that things get significantly faster. And again, it's like yes, we're getting better models, but we are still so early. There's so much we can still improve on the harness. Um like I see open to a little bit like a factorial game is infinite levels and like I go in Discord and there's memory and I'm at maybe at level three, but I I definitely want to go to level five very soon. And then there's like the core loop and then there's like community and it's like this is like infinite areas and infinite levels. Are you still playing Factorio? Uh again like my open claw is my factorial right now. That's what I thought. I thought you'd say that I started playing last year and then cloud code showed up. I was like no time for that. Um okay I have a funny story to tell. Um, so I one of the things about the open clock maintainer community was like I showed up on one day and I was like, "What am I supposed to do?" And somebody said, "It's high agency, man." I was like, "Okay, I know what that means." We keep saying that to people and it works sometimes. Yeah. I was like, "Okay, high agency." Like that means don't f up. Like be thoughtful, you know? Go read. Ask your agent to like crawl through the last 20 days of stuff and tell you like how things work. But um I started working on this iMessage thing and like Peter Peter's built like 75 other things uh that are like part of the ecosystem and I like got all this stuff working and then there was this bug uh in Peter's like GitHub action thing that he had to do. So I sent him sent him a text message. I was like Peter I found a bug and he sends me a screenshot that's like hey Clanker Omar found the bug fix it. And I was like oh this is so cool. Um, so was that is this like a silly one-off story or is that a glimpse of like how software development is changing? I think it's more a glimpse of the more data you connect to your agent and then you give them little skills to like help them along the way, the more you can automate in the pipeline, right? And my pipeline now is like over something. Look at look at this to here's the PR with the fix and I verified it end to end on a on a VM. Um so so it's no longer works on my machine. So you need to like make this longer like I shouldn't even be able to I shouldn't even do this. There should be there should be a hook where like my claw automatically gets a ping and then decides is this something I could prepare. I mean also token matching but But I feel we we're still not at the end. And like even here, why does it need to end at the PR? It could just it could just build me build me like a a branch release that I could already test and and if I like it, I just like to do the Caesar. Yes. And and and it's the new version. Like it's all about like your imagination. How much can you automate? And it's a step function. Every little thing I build improves my workflow a little bit. And and these things compound. Mhm. Even if they sometimes start a silly experiments or have silly names, um some of them may be silly, some of them become incredibly valuable. Yeah. And I think um one of the things that people don't quite understand is that you and the team are not just building OpenClaw. you're building this whole ecosystem of capabilities for safely moving at the speed I think we all want to move at because it's just like the velocity is just so profoundly like amazing. Um I mean to jump jump in there like there were definitely some bumps along the road. I don't know if you if you've seen the the Apo releases where we we at some point realized we reimplemented the shitty version of MPM. so it's like part of what's so attracting attractive for open code to me is about this. Okay, let's forget everything how we build software. Let's try a new way and then yeah, some things don't work and then like we learn from the mistakes, but damn, we're like we're getting very close to the dark factory now. Totally agree. Um, so I'm going to ask one one last question. Um I mean maybe similar to Dave I like I think we started our careers maybe around the same time but like I remember web 10 and I remember being here in the valley actually during the com bust and then web 2 and then like broadband like I remember going to work which is where the fast internet was and then home is where the slow internet like there were the all of these like very clear moments along this like 25 year arc of my my tech career. um mobile cloud. I remember when YouTube came out and people were like that's never going to work. Um so I I sort of can't help but think back to like all of this transformations that occurred. Um people are like what are these calls like and it's like it's hard to describe until you experience it yourself and you have that moment where you're like you talk about the WhatsApp moment. I had my moments. You probably had your moments. So um do you think this point in time is fundamentally different than some of these other their changes? I mean I would say my perspective is there's this builder role which is very different about this moment than maybe all the previous ones that have come come across uh the last 25 years. What are your thoughts? Yeah, I I think that it's exciting because I think in the 2010s we went into this like grind core culture where it was like I'm going to grind in a building in San Francisco and like get my money and it was just like not about building and tinkering and the the whole thing Peter just said like this sort of the fun of tinkering I think happens when the aperture widens when people are like so enchanted by a new technology that they can't sleep, you know, and I think we're back to that. Like I definitely felt that way in the in during the internet era, right? Like there was this time where everyone would go to these meetups cuz you were just all trying to make sense of what's going on. Like what? There's so much opportunity. I I I don't know what to do with it all. I need to bounce off some people in order to uh have better ideas. And so I think we're just back to a a cycle of that. And I do think the tools are really fun. Like these are amazing tools. Like they they're they're the most fun tools. I think that's why we're all so excited. It's like we've all been kind of talking to computers using this esoteric language or series of languages for a long time. And like only 50 million people in the whole world of the 6 billion people that use the internet have been able to speak that language. And now all of a sudden you can speak English and anyone can make things. And so part of it is I keep telling Peter this that every day I get into the main maintainers channel or I'm like working on something and I constantly tell Peter I'm like are we as out on the frontier as I think we are? And I hope we can teach people some of the stuff that we're learning. I think we we actually have a kind of a almost a responsibility to do that because we're we're out there like, you know, now it's not going to where the fast internet is. It's like I go to the maintainers channel cuz that's where the really awesome access to token is, you know, and we're we're doing a lot of really interesting things and it's so much fun and these tools are amazing and people talk about in a lot of different ways, but I just think we're back to a creative period and that's really exciting. I think maybe the thing to really say to people is that it's not over. Just because the first batch of AI companies is going public doesn't mean that this all is over now. Like I actually think the era of agents is going to be like as big as the era of apps on top of the iPhone or on top of the internet, right? Like it's going to be that big. And it was like way bigger than the late '9s, right? Um, so I think there's a there's a huge amount of opportunity that we can all go explore together. Yeah, I love that. Um, I think that your point about creativity is like the most important thing that like we get to spend our creative energy building and it's more accessible technology. Um, so I just want to thank Dave and Peter for being here. Um, thanks for having me. Um, and thank you again to the GitHub folks. Like I I before January, I think I had like two GitHub repos and I have 70 now. So, um, and there's a lot of people like me um out there. So, like I love that uh GitHub's a big part of this story as well. Um, and so with that, we're going to wrap up and transition. Yeah. What I'd love to do actually, I mean, as you said, Peter, the the it's the whole thing is is a community behind OpenClaw. And so what I thought would be cool would be to get some of the the maintainers up, the people that are here. We've also got a lot of the maintainers watching online on the channel right now. So, I thought we could get everybody up who's in the maintainer channel and get a photo together and then that way the people who all have claws can uh have a little uh say thank you to the maintainers. Everybody, so we're going to have a little photo. Come on through. That's all the maintainers. Scott, he's coming now. Any more? There is more. Do you want to ones that are? Yeah, exactly. The folks online here as well. We're going to say thank you to them as well. Yeah. So, everybody, if you'd like to give them a big round of applause. And then we've actually commissioned a very uh small token of our appreciation. So, yeah. Yeah. Yeah. It's the free token. Not tokens, I'm afraid, but uh this is it's a small token. So, uh if I get a sec. So uh it's officially the fastest growing open source project of all time to OPEN big round everybody. Thank you to the maintainers. Thank you so much. Yes. Thank you. All right. Hopefully you can hear me. That's cool. All right. Uh if you'll Take a step off. You're very welcome. For a long time, I wasn't sure like I never had public confirmation. Exactly. Sorry. Thanks all. All right. Hi everyone. I'm Ashley. Just take a quick moment to introduce our next speaker. Speaking of open-source AI and now security, I'd like to introduce the CEO of the Linux Foundation, Jim Zlin. We can't talk about open-source and AI without talking about security. I'd love for Jim to share some of the amazing things they're doing. All right, I'll go be very brief. First of all, I want to thank Peter in particular for saying how difficult and even more difficult than a regular company it is to start a nonprofit. You know, 30 uh 22 years ago, I met my wife who's a Harvard Business School graduate, tech exec, blind date, saw her, she asked me what I do for a living. I'm like, I work at this open- source nonprofit and we give everything away. The look of disappointment, it's like palpable. But uh today the Linux Foundation's home to Kubernetes, PyTorch, MCP, you know, all of the technology that's used to build all of this super cool stuff. And the good news is that, you know, we sort of have won in the sense that the entire AI stack essentially above those GPUs is open source. You know, Kubernetes, VLM, like think about it, everything all the way up to the agentic layer, open claw. It's it's great. The bad news is is the technology that is changing all of our lives that is making things like Open Claw possible is uh causing us to look back at all this stuff and realize that it used to be an adversary that wanted to find a vulnerability and exploit it to attack somebody had to have a bunch of time, a bunch of skill, and a bunch of money. And AI has compressed that all into minutes and hours. And what is happening is 30 years of open-source technical debt. How many people here are maintainers? I I know the open claw folks. Who else are maintainers here? How many of you have foregone test coverage in order to create a new feature? Do you always fuzz your code? Are you always threat modeling? are you? These are the least popular O'Reilly books, you know, like um and the the bill for this 30 years of all of these uh you know, sort of technical debt will get around to security has now come due because these attackers now have a tool that they can use to exploit that technical debt. And so the question is what are we going to do about it? And the way that we need to think about it is answer three What is the world's most important open source software package version number in rank order? We figure there's about 700,000 if you think about starting with Linux, obviously openclaw, and then all the longtail dependencies out than any normal enterprise would count on. It's about 750,000 projects. That means there are more than 750,000 maintainers like the folks here who are struggling to keep up with what they're doing right now that now have all these folks who are powered by AI tools, things like Mythos, the Linux Foundation's a part of it, other tools who are in goodnatured way finding vulnerabilities, submitting to the maintainer and the maintainers heads are exploding. And so what do we do for these maintainers, right? Hey, what do we do for the XZ utils kind of attacks where, you know, you didn't exploit the maintainer's software weakness, you exploited his loneliness and vulnerability. People don't just need a helping hand by giving them intelligence. They need a companion. That's what these maintainers need. And that's what at the Linux Foundation we're working with and what you saw last week with IBM, how many people saw the lightwell announcement? $5 billion and 20,000 engineers that IBM is contributing to open- source projects to go look at that footprint of technical debt and help pair up a maintainer with an engineer, maybe stop working on a feature at IBM and go fix some of the dependencies that they use in their products. The good news is if we can get this intelligence into the hands of maintainers and we can collectively have the willpower to work together like we do great in open source to give the maintainers at least a short-term companion to retire a bunch of this technical debt using AI tools. We can get ahead of these adversaries and that's what the Linux Foundation is really going to be committed to. But it is a huge endeavor. And so if you are out there working as a maintainer and need help, let us know. If you're a company that took a dependency on an open- source project and you're trying to push something upstream and you're complaining that, you know, no maintainer is responding, the burden of maintenance is not on the maintainer who may not even be around anymore. It's on you. It's time to pitch in everybody. And so that's what we're thinking about at the Linux Foundation. That is what we need to do about AI. And uh I I I think it's a pretty daunting challenge. Good news is we won. Bad news is we also won. And as they like to say in my favorite movie, sometimes you eat the bear. We did with open source. But sometimes the bear eats you. Let's stop that from happening and all work together on open source security and AI. Thank you. Thank you, Jim. All right. Thank you very much, Jim. All right. Next up, we'll have to rearrange some chairs here. We're going to have a panel with some of the maintainers that you saw earlier on stage. I'll invite them to join me up on stage while we add some chairs. Brad, Vincent, Sally. We'll spend a few minutes doing a panel discussion, but after we want to take some questions from you all. So, if there's anything you've been wanting to ask people that are behind the project OpenClaw, it seemed like maybe very many people aren't maintainers here. So, you may not even know what a maintainer is or what it takes. We can chat about that a little bit or you can take some questions after. All right. Thank you. Are these all live? What? I I was just wondering if this was Yes. All right. Everyone have a mic? Yep. One more. So, first I would love for you to introduce yourself. Talk a little bit briefly about what you're maintaining on OpenClaw and how you came about the project. I'll start with you Val. Okay. You said talk about myself. my favorite topic, but I'm going to um just say I do control UI design UIUX as well as the um some of the iOS stuff, not all of it yet, but hopefully more. Uh and the Mac OS app and um also Claw Hub uh V2. So the V2 for the the the UI for both the claw hub and the control UI. Um but it's mostly me, but also opinions by other people. And can you all hear? Is the mic close enough? Okay, a little closer. And then um how did you come to join OpenClaw as a maintainer? Oh, as a maintainer, I had to beg and plead and email and email and email and impress people and make multiple dashboards that were very very nice looking only to be told that I have to write it in a new framework. So, it's not going to be as nice looking, but I tried very hard and you made it and I hopefully made it and if I didn't, I'm open to suggestions. Excellent. Think that's all. Cool. Okay. Uh I'm Jacob Tomlinson. I work for Nvidia. Um we're mainly maintaining uh like providing security assistance to open claw. So we spend a lot of time going through the security advisories that are being opened against open core and finding out kind of what's noise, what's relevant, what do we need to fix, what can we harden, what can we change within within open claw. Um we also kind of maintain all the NVIDIA integrations and stuff but like that's kind of like a side activity. Um we got involved because Peter originally reached out to Nvidia to partner with us and and see how we could come in and support the project. Uh, and there was a few folks like everybody has got a bunch of people working on open source libraries in in a bunch of different places and so a few people were pulled together and uh I I just I love open source projects and this seemed like a great opportunity for like multi-institutional open source getting loads of big companies working together to help uh a nonprofit just seemed like uh exactly the right thing for me to work on. So that's how I I got involved. So that's one way begging and pleading. Another is getting invited. Excellent. Sally. Yep. I'm Sally. I work for Red Hat. Uh, and how I got into OpenClaw, I I went rogue. Weren't we're definitely weren't supposed to run OpenClaw on our laptops at Red Hat, but I was on a few days off and I installed it and I instantly felt that this was the most important project I've ever experienced. And so I went back into Red Hat. It was it was everybody's day off on Red Hat. We have these recharge days. And I went into Slack and I said, "Guys, have you uh tried this new project, OpenClaw?" And instantly a few people were like, "Don't run openclaw. It's so insecure. Do not. What are you doing?" And I was like, "Hello, what have I been working on for the past 11 years? I've been working on our product that is supposed to run any application securely. It's open shift, right?" And I'm like, "If we can't run this properly, like what what are what are we what are we doing?" And then I see the three dots dot dot dot. And it was one of the it was one of the executives that read dot dot dot dot. And I was like, "Oh yeah, this is on." And uh and he came and he's like, "That's right, Sally. That's exactly what I'm talking about." I'm like, "Oh, yeah." And from there, I just, you know, yeah, just emailed as everybody else did and uh here I am. Excellent. Brad, how can I follow Sally? Like she's You can't top that. That brings up all of us negative nies every single time. So big ups for Sally. Um my Brad grew MCO of Digital Mel. I I I focus on technology innovation for mid-market companies and help help them punch above their weight class. I focus on AI and automation and so I came across this um from the Microsoft Teams perspective which I found today uh I didn't know Peter told a very funny story today that like he almost didn't publish the plug-in for Microsoft Teams and if uh he hadn't done that I don't know if I'd be here right now. Um so I'm glad he did. Um that got some intros into Microsoft and so a lot of what you see today came from that. So thank you to Microsoft for everything they've done. And so um my goal is to reduce the friction between the Microsoft software and services and I think that Microsoft is uh the I I I say open call has been enterprise ready since day one because of intra ID because of graph API because of all those sorts of things and it's great to see that uh Microsoft is is given that backing and I can't wait to see where we go from here. Hello Vincent. Hello Ashley. Um so I'm Vincent for those of you that don't know me. Uh chief architect at OpenClaw Foundation. I actually joined as a contributor pushing PRs in like December, January, but I was obsessed. Like I literally had like my clawfield moment. I was like I couldn't sleep at night. I was like this this thing's going to change the world. I was going around some of the people in this room I know will remember me actually saying I think this project's going to change the world and I thought I was probably batshit crazy. Um so I get the last laugh now which is great. Um but what became of interest from that was like I needed in like I just needed to be part of this project. Um, so I hit Peter where it hits the most cuz I couldn't get his attention and I started doing security advisories. So I created such a mess of a problem that he's like, "You just got to come and join and fix the problem you started." Um, and here we are today. So that's my story. Love it. Hi, I'm Josh. Um, I'm with the OpenClaw Foundation, too. Um I uh I so I followed Peter on Twitter before uh Open Claw like took off in January and uh I saw it when it was his like little side project and he was in Marrakesh and uh took some time off in you know did the December holidays came back took some time off from my job to start playing with AI discovered openclaw in like January and Peter's like help I need maintainers um and I just taken time off my job I was like well I'm not doing anything and it's been like full throttle since. So here we are. Amazing. Well, thank you all for being here. We heard a little bit about security. Security is a hot topic with OpenClaw. So, uh, some of you actually participated in the GitHub secure open-source fund and program. I'd love to hear from some of you about your security journey and the project, a little bit about the program if you can share. I can jump in. Um, wow. Uh we're at a time right now and I think we're probably like the biggest one of the biggest targets of a new wave of like AI bug reports. Um we it's really easy to find vulnerabilities in open source code right now and we're one of the bigger open source projects at the moment. So we get a lot of security advisories. Um it's an especially hard problem because some of the magic of openclaw is the certain yolo mode about it. So, we want we want to enable a permissive environment. We want it to be like you at your computer doing whatever you want to do, but everybody's definition of safe and secure is different. In an enterprise context, that's like not acceptable. Um, so we're like getting a ton of bug reports and breaking in new and interesting ways every day. Um, we've been, it it was very hard for a while getting tons of bug reports every day and then Nvidia came along and has been a invaluable help for us to help wrangle those. But yeah, it's been it's been a ride. I think like 85 90% of the ghsas, the GitHub security advisories that we received were like slop like actual slop. Agree. Yeah. But like we still treat every single one of them super seriously because we have to like we can't not take people's security seriously. But you know it it does drain on you a little bit. Yeah. It's just that dealing with the volume of stuff that's coming in. Like I've worked on open source projects before as a maintainer you know kind of 10,000 star big open source projects where you're getting like single figures per year security advisory CVS. You come in you go through the whole process triple digits over here each day sometime. Yeah exactly. you make you make the private pull request, right? You fix the thing, you like time the release so that you put the the bug fix straight in so that everybody can pick it up straight away. And then how do you scale that when you're getting like 60 bug reports a day, right? Like you need to treat everyone with that same amount of seriousness. You need to look at it and say, is this a real bug? Is this by design? Like people are kind of freaking out around just the whole agentic thing because it can call bash, right? Like it can do all sorts of things. And so the whole class of bug of like oh you can accidentally call exec or do this other thing it's like no it can call exec by design like how do we actually mitigate that right like how do we we can't change the the the philosophy of open claw otherwise it's not openclaw anymore right we need to mitigate these risks in other ways and so you can't treat securing these agents in the same way that you would treat securing like traditional software you almost have to treat it like securing an employee right like if you hire someone into your company you can't like control everything they do there there is a certain amount of risk to having someone working for you and giving them access to systems. So you have to set up this whole like layered architecture of kind of tracking what they do and giving them single sign on and access to all these different applications. We need to treat agents in that same way and have that layered architecture um to actually secure these things. We can't just make like a pull request to like fix the security stuff. You have to like design a whole system and a whole stack to deal with it. Yeah. And the the the more eyes on OpenCloud, the more people using OpenCloud, that's the fastest way to make it the most robust uh project ever. Uh that that's that's the story of open source. That's what makes it so powerful. I have one quick thing on that. Please don't be delusional and think that it's a good idea just to not accept PRs from agents cuz eventually you're going to have to get over it. It's going to happen. It's going to be the future. And if you want people to use your technology, you should expect that to happen. And fortunately, we have a magician on our side who knows the defense against the dark arts. So, you can use agents against agents. They can be great at bastion as can you. So, you can play on a more equal playing field. Granted, there's like, you know, money comes into play. Not all of us have buou bucks. But, um, for those who don't, there are sponsorships or maybe your project isn't worth that much. I don't know. But what you can do is use your resources too to combat that and try to arrive at the best solution. And and there are a lot of open source tools like that that that we have on our um on in our repo on GitHub and you can fork it or play with it, do whatever you want to do with it, make it into your own. I think that they're really helpful and finding and identifying like this is slot, this is not slot, this is a duplication, uh doing all the triages, etc. And as does GitHub as well. Uh and I want to dig in a little bit there on what you touched on it. There are more contributions on GitHub than ever before. Maybe some of it thanks to OpenClaw and the rise of AI, but we have seen phenomenal accelerated growth on the platform. Users, contributions, more people doing open source than ever before. Yes. First, let's celebrate and appreciate that. We love to hear it. Um, so what are some of the tools or signals you're using as you're looking at tens of thousands of issues and PRs? Can you share some of the tools you've already made, some of the quick things you're seeing to evaluate to help you make good decisions about what's coming into the project? So, one of the tools um one of the things that most maintainers try to do when they first join was like, "Oh my god, there's so many issues. There's so many PRs. I'm having a panic attack like this." And they're like, "I'm going to automate this." And what we ended up doing eventually was simpler than what most people think, but we essentially take every single issue in PR and we feed it through an LLM, ask it three questions. The output of that then gets clustered and that cluster is like immutable and every single PR and issue gets like bundled together and it's relatively cheap, relatively quick. And we quickly found out that people's clankers were essentially like raising issues and PRs for the same issue. Lots of duplication. And you would have seen one night it went from 10,000 PRs down to like 5,000. Uh me and Peter probably closed half of them in like a span of like 70 hours because it was just like most of it was like just heavy duplication and then within that there was a lot of it that was already pre like already solved, already resolved. Um but it was just like a lot of noise. So that was like one of the solutions that we did is like how do we combat the noise? How do we know which issue PR is like a regression and I think subsequent to that like everyone started chipping in and we started creating like more and more automations around like how do we run agents on top of PRs and things like that and that's kind of the factory stuff that Peter was mentioning earlier. Yeah. Um I'm I'm currently unlearning everything I did over the past 12 years as a software engineer and that's that's really what we all have have to do. Um and yeah these tools are making it possible to and before we turn it over for questions uh on that what should people do how should they craft their contributions in a way that adds value that will be seen that you want to reply to that you want to work with. What do you suggest to all the folks out there that want to get involved in openclar any open source Either do our job very well for us like identify all the duplications, identify the issues that you're closing. Um prioritize them, say why you're better than the other ones that are out there already, the things that we're going to have to do oursel. If you can do that for us, obviously that's going to help us out. Another thing is if you can't do that, then maybe try to just create an issue and then let us come up with a solution because you may not come up with the best solution just because your agent says it and you really trust your agent because your agent is the best coding thing you ever seen in the world. But really, it's just saying things that's going to make you smile because you smile cuz you don't know what it even says anyways. So, we have to realize that our agents aren't perfect. They make mistakes and but very convincing mistakes. Don't be fools to those mistakes. Um, be aware that if you don't know the topic, probably don't make a PR about it. There's some audio feedback. That's my stomach. Sorry about that. Oh, there. All right. Thank you. Thank you. I think Peter I've heard Peter say this exactly what you're talking about calling it uh prompt requests, right? Like we want your prompt requests. That's what these issues are. If you don't want to bring code, you can bring code to it, but bring your ideas. Anyone else on what a good contribution could look like now with AI and open? I mean, I think overall like one of the big things we found is like open source is is a a community activity, right? It's humans working with other humans on stuff and we're all using agents to do this and to like accelerate ourselves and it's building like a buffer between us. So if you look at the issues, you look at the pull requests, they're all like agent written, right? So like come to the discord, come to the community places, come and join the like claw contributors channel in Discord and talk to other humans about what you're working on, what you want to do, what problems you want to solve, and then together we can like handle the like code side of things and dealing with the agents, dealing with all the stuff in in GitHub, but ultimately you can still have like humanto human communication. And I think that's like the next thing that we need to figure out collectively is how do we like avoid losing that? I think that's like the biggest threat to all of this is like we still all need to work together. Like we hang out in like a a call like all the time. Peter is just basically always in this like call and we can just like drop in and talk to each other and like there's that still that huge human side to it and I think if you just look at the PRs and you just look at the issues you can't see that like interaction and so we need to find other places to have that we do that in our discord. I think the only other one was like if you can attach proof like as a screenshot or video like me as a human I tested this I validated this is my experience like it's going to jump straight to the top. Yeah, Jacob and Vincent covered the two biggest parts, but I would elaborate on what Vincent said, like talk to your agent more. Uh, the best PRs consider the conventions of the project. They're not like a oneline like I have this problem. Fix this thing. Okay, push. Like, find out the conventions. Make sure what you're doing is not going to cause regressions. Uh, ask your clanker, is this the most robust way to do this? Um, the biggest thing, the biggest reason why, like I would opt to not use a contributed PR is because it goes against the conventions of the project. It's a fragile fix. Um, so talk to your clanker. Uh, spend a little time with it. Yeah. And when you talk to your clanker, Peter added um a a tool that you can upload the transcript with your agent and that tells us um how much time you've spent trying, you know, thinking about the pro uh the the problem and trying to solve it. Um and that's yeah, that that's very valuable. Yeah, that's a huge thing cuz all PRs look the same now, right? And you can't tell how much effort's gone into them. And utilize the history of the problem you're trying to solve. something that had been, you know, if it's you for a specific plugin, uh, check the history of that plugin and see what what PRs were pushed through and merged through and it'll kind of give you a road map of of what we're look the the maintainers are kind of looking for. And I really like the suggestion contributions look different now or they might and they come in all shapes and flavors. So, it could just look like joining the community and listening and participating. I hear there's some interesting music sessions and some multi music dance parties that are happening on the Discord. So, hey, if anything, you can join in on that. But I have a slightly contrarian view. So, I do see some people writing in the PR's comments basically saying please somebody please tag Peter tagging other people in the team and then they have other bot that's like please stop tagging everybody please don't do this anymore. So, sometimes whenever you do see humans in there it's usually a very sad story. So don't be that sad story. If it's really really important, like trust me, we'll know. And if it's not, then we don't need you to keep telling us over and over again. Like we also have lives. We're human beings. We also, most of us have jobs, too. So sometimes be a little bit patient and understanding. And um I feel like the team in general has this intentionality to do the absolute best for y'all. And we have a lot of weight on our shoulders. And some of us become extremely overly productive and obsessive about that. Um, I don't think my friends like me as much anymore and I'm sure that that's the case for other people and I'm sure they have to love me otherwise they I I won't let them be my friend anymore. But like we have people I was going to say before you say anything like Ashley was that like this doesn't apply just to open claw this applies to like any open source project and I'd almost go as far as like probably applies to like other repositories within organizations that are like adopting aentic like this level of like speed and noise and AI and like security and like all this stuff is is impacting everyone. He heard my speech. He heard the speech. I was practicing earlier. That's how he knew that. I was going to say that next. Oh, excellent. I I know there there are people behind projects, people with agents maybe in in a lot of them, but we all work together in open source and we want to create welcoming environments. So yes, be nice to the maintainers, especially the maintainers receiving piles and piles of contributions that they truly want to go through. Um, but together we can continue to make OpenClaw amazing and successful. So please do consider getting involved. This project by the way this is what 6 months of open claw at this level of growth. I think we're this is incredible. We're not talking about a year really exploded in like January sort of like mid January. So like what four and a half months. So you're still a young project. There's plenty of time for people to get involved. Um all right with that I hear the bar opening pretty soon. So, we want to be able to take a few questions and then have some time for you to chat even more with the folks that are here. All right, PJ's going to help. If you have a question, please raise your hand. Oh, one hand. Fantastic. That's an easy job for me. Yeah. Hi. Uh I along with Max and Lexi was here earlier, we run the AI agent meetup in San Francisco for the last year and a half. We've had Vincent speak. Um, and we run lots of workshops and I particularly run a lot of workshops where I help people install OpenClaw. And the number one complaint I get over and over and over again is it's not secure. What do I do to install it? And it creates a lot of unnecessary hurdles. And I know the answer is just go ahead and install it. But that I don't think that's quite enough for those folks who've heard it over and over and over again that it's insecure. What do you what would you recommend other than you know you know yolo? What's insecure about it? That's what I ask them usually. I I you know what there's so much of that so much. So it's got to be more than just the quick answer. So So I I I install OpenClaw in businesses for and I I'm seeing it transform actual businesses today. And I think you we one of my customers said it really well. The CEO said like we're we're juggling knives now. Nobody really knows what the future holds, right? And we are Dave mentioned it earlier. We're in like the late 90s, right? We're in the AOL days of the internet. And just think of all those of you are are old enough to remember that. I remember that. Um, just think like what you missed out on. You had an idea or something, you didn't see it through. So I I I tell people that it's risk versus reward. Like we everything's dangerous. Open call is just a tool. Um, but you have the ability to lock it down and give it the access you you do like and you have to take um you give it agency. And so whatever agency you give it, that's that's on you. It really is. You kind of have to take that ownership. in a in a business perspective, I think you show the value of of the ROI of like, hey, if I can automate someone's, you know, 80% of someone's menial task to focus on the meaningful task to grow the business, it's it it sells itself. Um, so I think I focus on on the positive on, hey, here's what people have built whenever you have taken these steps more than because we can have that argument all day whether it's whether it's safe or not safe and those things. So I that's how I would kind of tackle it. And hey, this week at Microsoft Build, we heard a big announcement. Microsoft's releasing scout. So that's OpenClaw backed by Microsoft. Thousands of employees at Microsoft are using it. So if you need to prove point about a huge enterprise, there you go. I think this is the thing is having all these different companies come in and add these layers of security around like OpenClaw is still OpenClaw, right? And like the scariest way to run open cclore is to install it as me on my laptop with access to all my stuff, all my credentials, everything, right? Like that's terrifying. But like if I put it in a contained environment, a sandbox environment, I'm going to run it in the cloud. I'm going to put these layers of authentication around it. I'm going to give it its own identity with its own like SSO and its own tokens. You look at what Microsoft's been doing with all the like ident um authorization and identification type stuff. You like you're going to run that in a container. You're going to run it on the cloud. You're going to limit what it can talk to and what it can access. You audit things. You log things. This is what I mean about like you need like a layered architecture around it so that like if you go and take open core and just run bare open core like you're kind of on your own but you can also go to Microsoft you can come to Nvidia you can go to whoever else is building these these structures around it to get the kind of enterprise packaged solution to building these things. Yeah it's a it's a it's a thousand small things and it starts from the lowest level of the operating system up. It's not um all the burden of openclaw. It's how you configure it and where you run it and what you give it access to. And the bottom line is that AI is super powerful, like incredibly powerful. Um, OpenClaw is so fun and whimsical, but it's not a toy. It's really, really powerful. And so, you have to give it that respect and configure it properly. Saw your hand first. Me. Hey, I'm Baron. um PM at Google. I actually help run the uh OpenClaw community at Google. Um it was pretty amazing seeing um you know the January moment we came from break and like in classic Google fashion seven openclaw variants spun up and and and tried to be uh the winner. Um but in any case u you know the community we run at Google really wanted me to ask how do you guys run your memory systems? I think openclaw sort of pioneered a very simple one but it was very effective and I think everybody sort of outgrew it and there's a lot of you know community discord on what's great is it gbrain is it other things um yeah so kind of just curious how you run your own stack um I can take this at least a little bit because again I do this for businesses and small teams I think the memory system that exists now it can scale for a medium and small team I think scaling it to an enterprise perspective that's a trillion dollar idea that Microsoft's trying to solve and others are trying to solve How do you share these markdown files from these personal agents back to a centralized place and understand how it works across your business? Um, so I think the ability of utilizing these tools to share the beauty is they're markdown files. So you can use GitHub to share those with your teammates um, securely um, or Azure DevOps or any of those other types of solutions. And that's what I do internally and that's what I do for my customers. And so like if my if my if my fellow employees have a meeting with a client, I have access to their memories and and then they have access to mine. And so over time again I think you start start a little slow kind of build those things out. I think things like Gary's brain are great but the way I see open claw is you need to make it yours and by yours we say for you or your organization. So I would point your open claw to Gary's brain and say you know my workflow I've been working you with you for three or four months. What from this repo do you think we should utilize and then build it your own. And I think that's the real power of open call. You make it it's the foundation for everything you build. And I think you take whatever other people do as inspiration, but then you also put your own spin on it cuz your secret sauce, that last 20% is what makes you special and that's what makes your agent special. One of the things that we are working on uh in the background is releasing benchmarks around this stuff. So Scott's another maintainer here who we're working with and a few other people around sort of evaluating like long run long long running tasks and various like personal agent tasks just to understand like which configuration is like most ideal. But this is going to change right? Something's going to work this week is going to change in 3 weeks and your model's going to come out. It's going to change. So it's like how do you adapt to this? So I can give you an answer now but it's only going to be based on what's good today. Be be comfortable being uncomfortable. Like I think that's the future we live in. Hi, my name is Abriel. I work in financial services. I am a founder and my question is I have three but I will hope you can answer at least one of them. What do you think is still fundamentally missing in agent execution infrastructure today? Have you seen any adoption in any industry in particular? And how do you approach identity consent and auditability in agents? What was the last word? And uh consent and identity. Sorry. And auditing. How do you audit? How do you audit what's going on with an agent? So sorry because I'm using this in industry. I I focus on mainly blue collar works. I'm from Houston, Texas. So oil and gas, construction engineering, those types of guys. We call it redneck rich where I'm from. So, they wear coveralls, but they make good money and they have big trucks and big boats. Um, you know, and and you would think that these people would not would not be interested in these tools, but once you show them the possibilities, I tell people all the time, you're the AI expert, you just don't know it. And this is people that aren't technical because you have 20 or 30 years of institutional knowledge and experience that makes you that makes you indispensable. And so, I know the tools. I'm an AI expert or quote unquote AI expert. Nobody is, but like I understand these tools really well. My goal is to share these with you so you understand how better to unlock your potential. And nobody in Silicon Valley can has has the institutional experience to build something for someone in working in the oil fields of West Texas. So the goal is for those people to learn these tools and build this build the systems that transform their businesses. And I challenge that for all of you here. 30-day challenge is hey, how can I make upskill myself in my space and uh and all these full potentials from an auditing perspective. The beauty is again markdown files are auditable. Yeah, I found that we're going to have to treat agents like we treat humans and I think you touched on that. So the layers that we have for um auditing and tracking and identifying humans working on systems, those are the same things that we'll need for agents. Um these identity providers. Um uh yeah. And just something on that because we are a Microsoft related event. agent 365 if you haven't heard of that yet. Basically, you're creating an identity within Microsoft 365 and Azure that is fully audible and controllable by things like Microsoft Sentinel and things like that. So, it's making that's one reason Microsoft rolling is still up. So, we're going to close things out. I want to say thank you to everyone from OpenClaw. You've actually I I want to give a another shout out to the other maintainers because I'm by far not the um most productive maintainer of the bunch and these guys uh just work 24 hours a day and they they Yeah. So I just really want to make sure they they have Thanks. Yes. So thank you all for joining us today. We're so happy that you all came to the GitHub office. We want to hang out and celebrate with you the success of open source, what's been happening on GitHub, tremendous growth. If you're interested in chatting more with people from openclaw, we'll be hanging out here for the next bit. Thank you all for joining us. By the way, the website openclaw.ai/cats what? Cats. Cats. Is that a special little Easter egg for us? Is it like some random Easter egg that got published? I don't know about Thanks for watching. Oh, dabs. Knuck one by the clankers. Yeah. Thanks all and thanks everyone who joined us online. Thank you all for joining. Just as a quick thank you to everyone here. We had people viewing the stream from across the globe from Iceland, Nigeria, uh India, Poland, New York. Incredible attendance. So, thank you all. Stay tuned here. We're going to be in the back doing uh demos with Nvidia. We would invite you to do some community ships and showcase your claws in the back. Uh and more importantly, just hang out and get to meet people. So, thank you all. Have a wonderful evening. Close down. It's done. The clos grilled cheese in the back. Grilled cheese. Are we good? Can I go home now? Yeah, you can. You can do whatever you want. Um well so when you get you dinner