Live AI Q&A + Crushing it in Chess at the Same Time - Come Hang Out!

Welcome everyone to the AI Q&A and chess live stream. And uh this this is actually going to be me be me playing chess. I've never done it before in a live stream. I've never done anything where I am like bringing a hobby into a stream. So this is like stretching outside of my comfort zone for sure. But it's it's going to be fun. So Oh, there's an echo. Shoot. Let me Okay, there we go. Sorry, I had to echo or I had a little bit of an echo for a sec, so I had to stop that. But anyway, yeah, so I am I'm not at my usual place right now. I'm actually traveling. Uh well, not not really traveling. I'm at a hotel, but not leaving the state. So, uh my my wife and I are just like going to hotels sometimes. And it was our fifth year anniversary, wedding anniversary last night. So, we're just out and about a little bit. So, streaming from a hotel room here, so the lighting might be like a little weird. I'm trying to adjust it and see what the best fit is. But yeah, also thought it was a good reason to do a bit more of a casual live stream today. Um, so yeah, I'm I'm here to just answer all your guys' questions and just chat with you guys about AI. I've got my coffee as well and then uh try to play some chess. I don't know how good of chess I'll play if I'm trying to talk at the same time. I haven't done that for a really long time where I've done any kind of like multitasking with chess. Uh because usually you're not supposed to, right? Like bad idea. But I'm going to try it here. So, we'll see how it goes. So, yeah, I've got my chat up on the left uh monitor here, and then I've got my recording on the right monitor, and then uh yeah, of course, got the game up on my center monitor. So, when you see me looking around, that's what I'm doing. But yeah, good to see all you guys here. Um, all right. I have no idea how many people are going to tune in for today, but uh I'm just going to have fun with it. So, all right. It's going to be fun. I appreciate it. Sean, thanks for being here. Hey, Scott. Hey, uh, Eric, good to see you as well. Hey, Sydney. Hey, John. Happy to have you guys here. I'm here for it. And, and if you don't play chess, like, it's all good because we're gonna be just chatting and it's just something interesting going on in the background. And, um, maybe it'll be fun to see me try and play well when I'm talking, too. So, yeah. Yeah, I think what I'll do is I'll I'll just kind of kick off a game here and um then if you guys have any any questions you can feel free to put in or just like anything interesting you want to hear my thoughts on or you want to talk about. There's definitely a lot of things going on in the industry right now. Uh so I can even just kind of like spitball random stuff as I'm talking. I have no idea how this is going to go. I just think it'll be fun to try it. And um I I have thought about like going so far as to like like because like chess chess is definitely a step into the informal compared to like just me doing Q&A uh talking and that's it. But if I wanted to make it even more informal, I could like play a video game and and do Q&A with you guys. Uh because there's actually another YouTuber couple that like mentioned wanting to do that and like do it as a group. I don't know. It's like so informal that like I don't know if I'd want to like actually do that like play a video game. At least chess is like still like an intellectual pursuit, you know what I mean? Uh but like that could that could be fun. So I might like do more of these. We'll see just like how how fun this ends up being. Um and uh yeah, if I want to keep doing these, but uh yeah, I mean like for the most part the live streams I do every Saturday, I want to keep it to like, you know, very educational, like very very much like focused on just AI. But doing these once in a while, I think will be pretty fun. Like I'm actually really excited for for this today. All right. Um what's my rating in chess? Yeah. Yeah. So, let's see. I Well, at least for this account right here, I uh haven't played on it in a while. So, right now, my rating and blitz is 2116 and then bullet is 1948. Uh, but it's still kind of in like the calibration phase, so I don't know like what it would really end up at. Uh, my rating in in real life, like my USCF rating is like um I think like 1850 or 1900, but I haven't played in a tournament since um oh man, it was 20 2019 is the last time I played in a tournament in North Carolina. Uh that was really fun. I actually did like really well in that tournament. But anyway, I think you can you can look people up if you go to USCF uh ratings. You can look people up. Let's search me quick. Let's let's find me before we get started here. Uh impressive. Yeah, I appreciate it. Thank you. I chess was a big part of my life uh at one point. Um okay, ratings, player lookup. So, I mostly played in uh junior high school actually. So, it has been quite a while, but if I just search I think you it was like last name, first name. Yeah, here we go. Okay. Um, okay. So, I I definitely overstated it. So, my my rating in real life is 1,700. Um, but it kind of like I never really got to the point where I got to actually climb, right? So, like this this is me in in like 2014 2015 playing a ton of tournaments and then this was um I guess like this I think this was North Carolina right here and then I I don't actually know what this second tournament is because I don't remember playing in 2020. I don't Anyway, so like that's the last time that I've had any kind of rating uh change at least. So yeah. Um, yeah. Last time I played was this one right here, 2019 Pan-American Interccogiate Team in uh in Charlotte. That was really fun. I had some really good wins here. But yeah, anyway. All right, let's uh let's go ahead and play a game here. Um, let's see. I'm going to take a look at the chat quick and see if there's like something I can talk about as I play my first game. So, give me a second here. Um, thoughts on Anthropic's latest paper. H um, which one? What? I don't know like what their latest paper is actually. May maybe we'll we'll pop this up quick and then Oh, is it this one right here? When AI builds itself. Okay, may maybe we'll take a look at this and then we will uh play a game of chess as we talk about it. U, someone asked about A2A protocol as well. Any companies actually building it for their products? Okay, we'll talk about that too. Good. Okay, cool. Cool. So, we got we got a little bit of fuel going into the first games here. I like that. Okay. So, uh well, anyway, let's let's take a look at this article first. Actually, this was put out. I don't know if this is what you're referring to because there's also one from three hours ago. Or wait, no, that's not a blog. I think this this is the blog. Making claimist teaching. No, that's not new. They've got a lot going on actually. Um, how we contain call out a class project. So, this this isn't new. This might be new. Is this what you're referring to? Let's see here. I'll zoom in a little bit for you guys. There we go. That's way better. Um, code per person by quarter. Of course, they have to advertise that claude mythos is a 8x. Of course, lines of code is an imperfect measure. It measures quantity over quality. So, 8x lines of code per engineer per day in the second quarter of 2026 is almost certainly an overstatement of true productivity gains. Well, at least I acknowledge that cuz yeah, that there's no way you're actually eight times more productive just because you have cloud mythos in your hand. Hm. Claude code session success rate trivial routine substantial open-ended problems. I mean it it is really cool. Like one thing that I have noticed myself. Okay. Honestly, I feel like I have enough uh fuel to go into a chess game here. Uh I'll I'll keep chatting about this as far as like how fast I I might do like five plus three so that I can at least like pause once in a while to think about uh what I'm saying. But I I really do want to play blitz chess because I think that's that's more fun. I don't want to do like a 15 or 30 minute game or anything. So, let's go ahead and do this. I'll do uh actually maybe I should do casual. Yeah, whatever. All right. So, let's get started. I'll full screen as well. So yeah, one one thing that I've been doing a lot of testing on right now is uh benchmarking using different large language models for different steps of an AI coding workflow. So you have your exploration like typically when you're doing any kind of agentic engineering, you're going to be doing uh exploration and then planning and then implementation and then validation, right? Like those are like the classic four stages. No matter what your process looks like, what framework you're using, whether you built one or you're using BMAD or GSD or whatever, like you're almost always going to do that combination like those four stages. And especially right now because of how bad the rate limits are getting and like Frontier models getting more and more expensive, um it's it's getting to the point where like it's not always realistic to use the best of the best large language model for every single step. And so I've been doing a lot of exploring like can we get the same results or or at least only a little bit worse if we use the most powerful model like opus or GPT for planning and then use the weaker model for uh maybe even everything else like exploration doing the implementation and the validation right because like as long as you have a good spec for that that like you created with the more powerful LLM then you can hand that to a cheaper model and it has such a good set of instructions to follow that you don't need it to have as good of reasoning power to still get the same results. And so I'm not on my desktop right now, so I can't I unfortunately can't like show uh there's like a really cool graphic that I built for the Dynamus workshop that I did yesterday because I do one in the community every single Friday. And um well, this position is really boring by the way. They were kind of just getting locked down. Um but anyway, so um geez, what Hold on. I got to think of what I actually do here. Um yeah. Okay, this is going this is going to be kind of weird. I'm going to play a little bit weird here because I want to get something moving. Um but yeah, so like I I had this graphic that showed through like a ton of benchmarking that I did that if you have this explore um plan implement validate workflow and you use opus for the planning but then Kimmy K 2.6 for everything else. It's pretty much as good as if you are using um as if you're using Opus for everything. like it's it's obviously a little bit worse because of course if you want the best of the best results, you're going to be using like GPT 5.5 or Opus for everything, but you can get you can save like hundreds of thousands of tokens with the most powerful model and um still get really good results. Like it's actually pretty crazy. Uh I think I'm losing a pawn here. I'm not uh playing the best. Let's try this. Yeah, I'm I'm losing a pun. So, yeah, not like you have to use Kimmy K 2.6. There's a lot more exploration that I want to do with other models. Um, like just looking at um like Gwen models or um what is it like Miniax M3 came out just last week and that one is pretty good. It's It's not like nearly as good or not quite as good as Kimmy K 2.6, but it's like a fourth of the price, I believe. So, like for routine and this this is kind of what I'm getting at here, like when we're talking about this article for trivial and routine tasks, like it's gotten to the point where you can use smaller models to knock out these things like really reliably. So even for like the entire explore plan implement validate workflow, you can do that full thing with a smaller model now and then save the larger work for these problems where like the only reason this curve is going up is because we're relying on the really powerful models these days that do have that deeper level of reasoning. So yeah, I'll talk about that in a sec, but I want to see if I can recover this position. I I I might just be kind of losing this game now, but that's okay. Um, yeah, because I'm just going to lose this pawn. This was this rook move is just dumb. Um, yeah, I'll just I'll lose the pawn. At least he has an isolated pawn, so I might be able to recover something. All right. Um, John said I can't keep up with both the conversation and the game. It's all good. You don't have to. It's it's uh it's just casual. Um, yeah. you don't really have to like follow the game, but uh if if we want to slow it down, like I could try playing a a slower game, too. That might actually be a good idea because this is pretty, you know, um let's try. I don't think I want to lose the bishop pair, but I actually Well, actually, I don't have a choice. I have to lose the bishop pair, too. So, yeah, not doing not doing too hot in this game. Um, I'll just do this. That was really Yeah, it's all good. All right. Oh, I got some sparkling water, too. So, drink cuz I just finished my coffee already. It's such a tiny cup that they have in like the breakfast downstairs. So, want something else to drink, too. I'm not actually that big of a fan of sparkling water, but it was just like in the it was free with the room, so I'll take it. All right. Um, yeah. So, let's see. Going Let's go back to the article. I don't have that much time left, but I'm going to kind of accept that this game isn't going the best. Anyway, I'm trying to think. There was something else that that I wanted to talk about based on this article here. Um, let's see what it was. Oh, yeah. Productivity gains. I I want to talk about this because there's been a lot of of um I've just seen like a lot of people talking about how AI is overhyped because you you get the sense that like large language models make you 10 times as productive as a developer or there's like all these these claims out there and um and then people are kind of asking the question like if it really is a 10x productivity game, how come businesses aren't 10xing their revenue? Right? Like if if you're a SAS platform and all of your engineers are able to move at 10x the speed and using cloud code like your product managers and QA engineers like if you have that too like theoretically they can move at 10x the speed. How come you we aren't seeing 10x profits? And I think that's like it really comes down to um the the numbers aren't as good as they seem, right? Like just because you can code faster doesn't mean you can solve problems faster. Like there actually is a big distinguishment there. As much as like I would, you know, I hate to admit that. Um, hold on. I I feel like I might I might actually have something coming in this game. Like I'm I'm not I'm not winning, but I feel like I could put some pressure here and make something happen. We'll see. He'll probably just defend with rook C8. Um, yeah, I probably don't really have anything I'm trying I'm just trying to see if I can like be annoying because then I can Oh, okay. Interesting. That's a decent move. Can I just uh cuz at least I have like I'm down a pawn but at least I have the open files. Okay, I got Shoot. I got to think fast. I'm gonna take I'm gonna do that and then do that because I might actually get my pawn back. This might not be a lost cause of a game. Actually, I just I only have 38 seconds now. So, I might take a pause answering questions or like chatting about AI just to like so I can actually think because now I have the time pressure. Um, yeah, my knight doesn't really have anywhere good to go though, so that might be a problem. I'm just going to do this. Well, yeah, he's going to be able to defend now. Um, yeah. So, it might not Yeah, I probably am not actually that good here. I'm going to try this. Actually might be kind of sneaky here. Okay. Well, if I go bishop here, then he can go knight here. I I might not um I might be able to. Okay, cuz if he takes I think I'm okay actually because then I take his queen then he takes my queen and then I take his rook. Yeah. Okay. Uh oh wait. Oh, he just threw his queen. What? I can't believe it. He just threw his queen. I was getting Okay, I wasn't getting smoked, but I wasn't doing that well. And then he just Wow. Okay, we'll take that. Holy cow. I haven't had like someone drop a queen in like forever. That's crazy. He just What are the chances that my first game that I live stream it he just uh why would he have resigned? because there's no coming back from this. Like, yeah, I only have 18 seconds, but the game mode is 5 + 3. So, okay, let let me explain this for those of you guys who don't know how uh time works in chess. The first number is the amount of time that you have on your clock for the entire game. So, like five minutes, if I spend four seconds on the first move, then I have four minutes and 56 seconds for the rest of the game, right? The the second number here is the increment. And so every time I make a move, I get three more seconds. And so even though I only have 18 seconds, I have such an advantage now that I have his queen that like I can move pretty fast and I don't have to be extremely careful because even if I slip up a couple of times, I'm still going to have a big advantage. Like even if my advantage goes from a queen minus a pawn to like just a rook or just a bishop or knight, like I'm still going to be crushing him. And so I can move pretty quick and like each move I get 3 seconds, 3 seconds, 3 seconds, then before you know it, I'm above 30 seconds, above a minute. And um I can bring it home pretty easily because he doesn't really have much counterplay. He doesn't have much he can do to really make me think um very hard because there's no really big risk when I have a I have a queen and he doesn't, right? So yeah, I mean in in his seat like resign, I would have resigned there as well obviously. Um, even though like even if like I was the one that had three minutes and not 18 seconds. So there we go. I I'm Okay, I gotta be honest. It is very cool to have a a rating in uh Blitz of 2172, but I wouldn't I I don't think I'm actually that good. Like if I if I were to like really rate myself right now, I would probably be like, 1900 or 2,000, I think, is like an honest estimate. But it's fun. I've never been this high in a online chess account ever. So, all right. Never surrender. That's right. Yeah, I just kept all Well, I guess he surrendered. Well, okay. Here's the thing. They tell you to never surrender, but like in chess, it's sort of like an expected thing. It's like the culture of chess is like when when you like really don't have much of a chance at all, you just you just surren. You just resign, right? Like it's not really a sign of weakness in chess. It's almost a sign of strength where it's like at least I can like maybe I screwed up, but at least I'm good enough to know that I can't win, right? So, it's sort of like a you're you're saving face a little bit. I don't know. May maybe most people don't actually think like that. That's just me. But yeah. All right. If you are a 2067 player, you wouldn't have given up that game. I think that's true. when you lose your queen, like when you're that when you're rated at that level, like you you just kind of know that like they're probably also not going to drop their queen and give you a chance to win. Yeah. Yeah. You don't you don't count on mistakes. You never that in in chess that's called hope chess. When you make a move hoping they make another move and you're kind of like hoping they make a mistake, even if you think it's like a tricky one they might not see, you still don't want to play that way, right? Like you always want to play chess expecting them to make the best move. All right, cool. So, anyway, I think that's most of what I wanted to say for this article here. So, maybe before I get into the the um next game, I'll I'll chat a little bit more about this. So, yeah, the productivity gains. So, so what I was saying before, I wanted to really hone in on the game is um there's a big difference between like code output and productivity gains. I think so like when the your typical engineer starts using an AI coding assistant and they are blazing through a lot more work a lot of the work that they end up doing that they couldn't before is picking up a lot of the things that are just in the backlog. So, it's like those lower priority items that uh the team kept shoving to the side and like it's things that have to be handled, but it's also things that aren't going to 10x the revenue or the the apparent productivity of the engineering team. So, if anything, AI coding assistance, I think like this this kind of explains the disconnect while still presenting AI as being pretty important. The way that I put it is AI coding assistants enable engineers to handle a lot of the grunt work that used to be shoved to the side. So you're handling those refactors that you put off for a while, adding testing into the codebase, improving your CI, improving your security, like they're like really important things, but it's not the things that like immediately drive the needle forward. So if you're handling 10 times more in your backlog, it's things that are in the backlog for a reason. because even if it's like kind of important there, you have a reason like you're kind of able to just shove it to the side. So now AI enabling to you to do those things, it's not like you're doing like an adex of of uh new features. It's more like an adex of at least if you have the right system and your team is using the tools, right? It's an adex of like improve hardening your security and and reliability. And um obviously there's kind of a double-edged sword there if you where if you rely on AI coding assistants too much then you might be hurting your reliability because you if you're like vive coding but if like you have a really structured approach and you're taking advantage of coding agents to like really be more thorough in your work because you can move faster and test more and validate ideas more and and like build out different versions of a single ticket or feature really quickly. Like that's also a really powerful thing. So like another yeah kind of like going off of that another important point is um when you're using AI to work on things because you can write code so fast you can ideulate faster like now thanks to AI ideation isn't just let's sit in Figma or let's create a sequence diagram it's like you can actually write the code and you can write multiple versions like different ideas of how to implement a feature and then have those as like initial smoke tests to figure out the best way to go about solving a problem. So yet again, we're not building eight times the number of features, but the features that we do build, maybe we got to build eight versions of it so that we can really make sure that what we ended up going with is the one that it's best. So I think you can use AI to I mean like in the end you can build more features too, but like an 8x in uh being able to code might be more like a two or 3x in like how fast you're moving the needle forward, but then if you're doing it right, everything's more reliable, too. And uh yeah, so that that's kind of where I stand with it right now. So I think that like it's good that they admit that it's almost certainly an overstatement on true productivity gain, but it doesn't mean that it's like not eight times. Like the engineer itself, I really do think is like, you know, five to eight times more valuable. See, I'm curious what your guys' thoughts are on that, too. And then yeah, um we can talk about the Sorry, I put up the wrong thing. Where' my Where's the comment going? There we go. Yeah, we can talk about the A2 A2A protocol a bit as well. Maybe when I get into the next game, I'll chat about that. All right, let's see. That was a pride resignation. Yeah, maybe it was. All right. Uh, you could have easily blundered like him and we don't know how strong you are yet. Yeah, fair enough. I mean, I could have with 18 seconds left. Like, maybe I would have ended up dropping my queen, but I think that is unlikely. So, again, may maybe you guys don't think he should have resigned. I think it makes sense, but I'll take the win either way. All right. Uh, Queen's Gambit. I actually do like playing the Queen's Gambit. Maybe if I get white next game, I'll play Queen's Gambit. Um, well, I guess I was white last game, but he he just played the uh Kings Indian defense. All right, let's see. All right, I'll read a couple more things in the chat and then let's uh get into the next game here. Uh, since you're playing chess, I built a platform that generates chess animations via prompts. Can you please review it? That's cool. Yeah, I mean, feel free to share it if you're in the Dynamis community or if you just want to shoot me an email. My email's in my bio. I I'd be interested in taking a look. It's very cool. Um, and then yeah, Scott, this is a good point. Coding is just a fraction of what a business needs to 10x, right? And like that that's why like even if an engineer is writing 10 times the code, they can't push the business forward 10 times because they're not the only thing in a business. Even if you are a SAS platform where like you are a group of founders mostly as engineers like you still have to make so many highle decisions and and be in meetings and work on other things like marketing and go to market strategy like like in the end like if you're writing 10 times the code a lot of that like I said is going to hardening and refactoring and backlogging and backlogs and testing and like things that matter but like it isn't like you're able to just all of a sudden ship 10 times the features because of everything else that goes into each feature before And after the actual development, you have to plan it and and no, not just talking to Claw to plan it, but like actually like plan it as a team, build it and then market it and uh actually like integrate it in your platform and uh make content about like whatever you have to do. There's so much that goes into it outside of like just the code 100%. All right. What's the most productive way to spend time for an engineer who has uh limited time to keep up with AI and its effective use besides being a member of Dynamus? Well, I appreciate the Dynamus mention there, but yeah, most productive way to spend time um for an engineer. So, uh yeah, I mean like that's what I try to do is help you guys stay up to date with all of my content. So, kind of like have to mention that, you know, like of course because that that's what I spend my life doing, but like outside of outside of like my content um really the most important thing honestly like the anthropic blogs are really good. Like I know that like anthropic is very controversial right now, but their their blogs are extremely good with keeping up with the industry in general. And then the other thing that I would highly recommend doing is having some part like almost everyone at this point has or at least is interested in building a second brain like building it themselves or using something like open claw or Hermes. And um actually maybe I can talk about this when I get into my next game here. Uh let's do that. Um I might actually try a let's try a bit of a longer game. I think that'd be fun. Let's try let's just try this. I'll try a 10 plus 5. I'm rated 2,000 for rapid. Okay, let's chat about this. I'll close down the comments so we can see the game as well. So, um yeah. So, anyway, talking about second brain. So, almost everyone has or is interested in building a second brain. I have one myself. It runs with cloud code. Well, actually, I recently migrated it to use PI with my codec subscription just because um Enthropic is making some changes. you can't really use the client agent SDK with your subscription anymore. So, uh yeah, Pi with Codex. But anyway, I have a second brain and uh my most important component to my second brain honestly at this point is I have it curate all the news and latest trends for me. So, there's a ton of sources that I have it um basically look through every single day. So, I have like a bunch of YouTube channels that I have at follow uh from people that a lot of you guys are probably familiar with like you know Matthew Berman and um I don't know Nate B Jones like there's a lot of really good channels out there that are just like covering the latest news and trends. So, that's a really big one. And then I also have like RSS feeds like the anthropic blog is of course something that I have injected in my uh second brain and like the openi blog and everything like that. Um what else do I have? Um I hacker news is a really good source. So like a lot of good articles on hacker news that my second brain is digesting every day. Um let's see here. And I think that's mostly it. And then I I have like Perplexity built into it as well. So it does like Perplexity web searching. Um so yeah those are all the sources that it crawls or not like crawls but ingests I would say. So every single day it's constantly reading in that information. And then because the system is a part of my second brain, it's also able to um it's also able to like figure out like based on the things that I just pulled from the internet, here's what Cole probably cares about. Like it knows that I have a YouTube channel where I focus on building AI agents and leveraging AI coding assistance. and knows the things that I'm working on in the Dynamis community for like upcoming workshops that I have planned and content that I have planned and so it's able to really like curate like okay cool here are the things that you probably care about that have happened today and so like that kind of thing I recommend for everybody. I know that like some of what I was talking about there might have seemed kind of specific to also I'm not playing that well in this game either. Uh gez. Um so I know that like some of those things might seem like they're specific to like being a content creator, right? Because it like knows about like the content that I want. But even if your second brain knows more just like the projects you're working on, like here's the tech stack that I'm using, here's the tools that I'm using. So you can like find news on like evolutions of those tools or how to use those kinds of architectures. like that kind of thing can also be really useful. Um, so yeah, like that that system is my biggest recommendation because then you're you're building a second brain to help you keep up with things so you're not always relying on yourself. Like I call it my co-founder. I think that's kind of a cool name for it. Um, okay. Let me let me spend a sec to think about this game now because I'm in a little bit of a predicament here where I I kind of let my bishop get put in a stupid spot. I probably should have just moved it back here or Whoops. Probably should just moved it back here before I pushed the pawn up. It's It is really hard to talk and play at the same time. Like I'm I'm consistently making mistakes that I wouldn't normally. Yeah, we'll see because I I might just have to kind of let my pawn structure get messed up. because I I don't know. Like I could do this, but I think he's going to sacrifice a knight and it's not going to look actually I really really can't do that because he's going to take here. Take take and then my knight is uh pretty much Yeah, it's gone. Okay, I can't do that. So, I'm going to think this is my best bet. Yeah, I'll do this. All right, let's see. Move pawn forward. If you're talking about this one, I don't I I can't do that. Uh what is my rating? Uh so well right here in the rapid version elite chess is 2,000 and then I just got up to uh it's like 21 something in blitz chess. In real life it's 1,700. Like my USCF rating, but I think that would definitely go higher if I kept playing tournaments. Good move. I appreciate it. Yeah, I think I might be kind of saving myself here. We'll see. Cuz I'll have to I'll have to take this and then um yeah, like I think I'll be okay because I I should be I I actually should be pretty fine here. The only the only concern is Oh, no. No. I'm good. I'm actually I'm actually doing pretty okay here because I I was worried if I like that he can't do this push. Yeah, he can't cuz then after I take and he takes then I put my bishop here and his queen is actually pinned. So, uh yeah, I think I'm doing this because he can't take my knight because then I pin or I uh skewer the queen to the rook. Uh, so I'm actually I'm actually looking okay. Like this this is a little uncomfortable, but I'm doing pretty good. I'm getting baby brain. [laughter] I think I think it's just Sydney that gets the baby brain. I don't think that applies to to me. Hopefully not. I got a business to run here. All right. Appreciate the news aggregation idea. Yeah, you're welcome. It's really powerful and the best part is it evolves itself over time, right? Like if your second brain is responsible for pulling the news and understanding the news you care about, it's understanding is better over time and then you can tell it like, hey, this news, this piece of news you gave me, like actually I don't really care about that at all. And it's like, okay, let's let's go ahead and try something different or let I'll remember that for next time. Okay. Um, let's I don't think I want to take this. Um, because it's going to open up a like if I if I take and he takes and he's got to line my king. That's pretty uncomfortable. So, I might play it safe here because I could do one of one of these two. I'm thinking um yeah, let's do Well, I guess I can't spend too long on this. This is probably better because if I went here, I was just thinking I could go here. So then I'm just a bit more protected and then I have then I could maybe Yeah, this is probably fine. I don't know. I was a little stuck between those two. All right, put pressure on the bishop with the knight. Yep. Well, that's what I'm doing. That's what I'm doing. The problem is though, I I don't think I can take because that that looks like bad news. It might actually be okay because then I play king up and there's not too much he can do. That might actually be fine. But like my knight is kind of sitting fine as it is. So I might even just kind of like Well, no. I can't do that. I might just do this. I'm trying to think if he has any kind of counterplay. But his bishop's kind of stuck here and he can't move his pawn up because then I have this space open again and he can't move pawn here. The only problem is if I move this pawn up then he can shove me here. So I'm left like a little stuck. But this this is looking actually kind of okay. All right. You got kavade syndrome both times. Gained 50 pounds. I actually don't know what kavade syndrome is. I I assume it's like like a sympathetic pregnancy kind of thing. Kavad syndrome. Oh yeah. Yep. Sympathetic pregnancy. Yeah. Uh proposed condition where an expectant partner experiences a real physical and psychological pregnancy symptom. I mean, I've heard about this. I don't think I've really had it yet. Um we'll see. It it might happen. Or maybe it has already happened. I haven't quite realized, but Yeah. But yeah, it's a it's a real thing. Yeah. All right. Yeah. So, one thing I wanted to talk about is uh the A2A protocol because someone mentioned that at the start of the stream here. Sorry. I mean, I'm going to be a little all over the place for our stream today. Hope you guys don't mind. But yeah, so someone talked or asked about A2A at the start. So, the A2A protocol, I mean, I guess this is going to be kind of a quick thing. Um, there's really not much that's happening with the A2A protocol, um, unfortunately. So, um, oh, wait, hold on. This is might be an interesting opportunity. I'll talk about I'll talk about that in a sec. But hold on. Because if I do this, I can if I boom my pawn here and then he pushes, I can pin him or not pin him, I can fork him. That's interesting. I might actually do that. could be good. Or I could take here and then I can push this pawn. I'm actually I might do that. Yeah, let's do that. Because then I'm so I'm protecting this pawn so that I can take and then I can push this guy so that I'm pressuring him here. We'll see. He might just take me though. Oh, interesting. Oh, okay. H. Well, yeah, that kind of changes that plan because now now I don't really have a way to push this pawn up. Um, let's see here. I think I might have to take this. Just makes me nervous. Or I could retreat. Honestly, retreating might be the best move. Oh, yeah. Actually, yeah, it is the best move because now now I can still play this. So, I'm getting ready to make a push. His bishop is still kind of stuck. Like, if he now now he can move back here if he really wants to, but it doesn't really help him. Okay. Anyway, A2A protocol. I've really not seen any businesses adopt it. So, for those of you guys who don't know, the A2A protocol is um it was developed by Google last year. It's kind of like the MCP protocol. MCP was Anthropic's way to connect any AI agent to any tool and it blew up like super successful, widely adopted. It's like everyone is like MCP is one of the things you have to know when you're working with AI coding assistants just AI agents in general because you can connect MCP servers to any agent to give it a set of tools to work with any service like you have the the GitHub MCP which that one's actually probably a bad example because the GitHub CLI is so good that you don't need it but you have like the Atlassian MCP server for Jira you have the superbase MCP server or Neon so it can manage your database like there's so many capabilities you can give your agents through it it's really good and so the A2A protocol is like MCP, but instead of for connecting agents to tools, it connects agents to other agents. So, it's a standard of communication. So, it was actually pretty cleverly designed where essentially every agent has what's called an agent card. And so, it's just like JSON configuration that defines like here's what the agent is able to do and the different capabilities or like here's a description of its role and like the tools that it has, like the capabilities that it has. And so that way an agent can communicate to another agent very quickly and like in a standardized way like hey here's what I can offer. And um it like really seemed promising like oh finally we have a way for uh interconnected networks of agents like they can communicate with each other and they can uh rely on each other like this agent can do this thing this agent can handle these tasks. But the problem with A2A is the only way that it's successful is if a lot of people adopt it at the exact same time. Otherwise, there's no use for it because if you don't have other agents to talk to with the protocol, then what's even the point, right? And so, unfortunately, um there never really was that mass adoption event. So, at least to my knowledge, like there really isn't anything where people are using it regularly. Um, yeah. So, that's kind of what it comes down to is like for a protocol like that, you need a lot of people to use it for it to be useful. MCP was kind of different because with the model context protocol, you're just giving capabilities. So, even if you don't have a ton of people adopting it right away, but some people are packaging things up like here's my superbase MCP server. So now next time when Johnny goes and builds an AI agent, he doesn't have to write the tools for Superbase like he might have before. You can just leverage this MCP server and then boom, your agent has access to it right away. So it's like really valuable right out of the box. But A2A is different because the whole idea is like swarms of agents can from different companies can work with each other so that that they can sort of like there's even talks of building some kind of like payment system where you like pay the agent before it does an action for you. So you like have this network of agents paying agents almost like a blockchain kind of thing. And um like yeah that just yeah never happened. Uh okay. What is going on here? I don't think I'm really threatened by this move. Should just be able to go back or even here. Or I could just do this. Um, I think that might be fine. I don't know. I'm not really I don't think he has that much counterplay. Like my my pawn structure is just bad, but it doesn't really matter. Then he'll move back and then I'll go here. Yeah, I think I'm doing okay. I guess the main thing is I got to figure out this knight. I do this or put more pressure. No, because this might be good, right? He can't do anything. Yeah, because I I think this knight is actually in such a bad spot that I don't want to go here and and propose the trade. So, I'll just get my way over this way. This game's going all right. White pawn aligned with rook and queen. We'll move forward. Um maybe it's not I'm not too worried about that. Well, now I am because he put the rook here. So, I think I'll just have to move my queen here and I should be Well, wait. Oh, wait. Hold on. Um, that's a bit of a problem actually cuz he has two attackers on it now. should be okay to go here. I don't have that much time anymore. So, I got to like kind of move fast now again. But he can't take cuz I would win all this trading here. Yeah, I think this is probably his next move. But I should be fine because I'm pinning the pawn. So, I should be able to just side step. Yeah, I think I'm okay. Only problem is I actually rely on both these pawns to protect him right now. This this actually made me a little iffy now. Yeah, I don't think this is ideal anymore. Yeah, that's not good cuz now he'll just take the rook. So I'm down a pawn now. Don't really have any counterplay. Yeah, I'm so low on time now. The only thing that like really I have hope for here is just that his knight's so bad. Like I don't really have a minor piece that's like like actually bad. Uh that actually kind of stinks. Yeah, he's up two points now. Yeah, it's not going too well. All right. Oh, shoot. And that pin makes it even worse. Dang. Yeah, that's rough. I don't really have any counterplay at all. But see, this is where like I think I will resign. I know you guys said to never give up, but it's like I mean there's not really a chance here. Just because you guys said it earlier, I will play the game to completion, but it's not looking hot. So, we'll see. I have like a tiny bit of counterplay down the file here, but it's going to be squashed very fast. because he's going to uh pawn here. Knight's going to go away. Bring the queen in. I mean, he's got things to do here. All right. One thing that would be fun to do at least once for the stream now is play a game where like I actually focus on the game. We'll see. I mean, I actually probably won't do that because like that's the point of this is to chat with you guys as we're playing here. Well, yeah. Uh, another problem with A2A was the network latency added on top of the interface latency, right? Yeah. So, it was slower in the first place. Yeah, it's just not the best. All right. Uh, any advice on which cheap model is reliable right now? Miniax M3 is being hyped. Neotron, I tested it and it's not good. Yeah, Neotron isn't really ideal. Um, so M3 is pretty good. There's we actually so we did some testing with Archon and M3. So like we built some harnesses, AI coding harnesses with M3 and did some testing and uh it's kind of like Okay, I can't really show the article. I'll show it really. I know I have like two seconds here, but the game's lost anyway. It's kind of like this where it's like for trivial and routine tasks, M3 actually does pretty good. Like you can use a cheaper model for simpler things in your codebase. But where it really falls apart is anything that's like probably at like this level like substantial and open-ended problems like don't use M3. Maybe you can use it for like a bit of your workflow like what we were talking about at the start of the stream using the cheaper model for some of the implementation or exploration like mini Miniax M3 is good for that. Um I would say that like Kimmy K2.6 is a step above M3 when you're going for those like open source cheaper models. And then another model that I haven't tried yet like I have M3 and Kimmy is um the newer Quen models. Those I think are pretty promising as well. And really like what it always comes down to is like don't use the more powerful or don't use the cheaper models for everything. Just use it for part of your workflow, right? So like that that's what I've been teaching a lot recently. Like I actually did a workshop in the Dynamis community just yesterday where um I I taught like how to mix providers. Wait, hold on. I might have some counterplay here. What if I do this? I know this is kind of crazy, but I'm gonna I'm gonna ignore No, I can't. Oh, no. I can't ignore. I have to take Okay, I got six seconds, so I need to I need to focus up here. I mean, this game's lost. I can't There's not really much I can do. I think he's gonna he's pro He probably will nab me here. I mean, I just have like a little bit of counterplay, but I mean, even if he moved his king out, he'd be fine. So, not looking good. All right. Um, yeah. Oh. Oh, I ah [laughter] I I ran out of time. But yeah, if you look at the analysis board here, not good. At least I think. Let me give it a second to load. I I mean maybe I could like get away and then somehow have a way to push this, but no. Yeah, it says mate and seven. He has a way to check if he if he I mean I don't think he would have missed it. So all right would recommend trying the Quen 3.6 27B 35B models. Haven't any good results for the size of them? Yeah, and but that's the important thing there is for the size of them you're getting good results. like they're still going to be significantly worse than Kimmy or and then you know especially like GBT 5.5 Opus 4.8 like the frontier models but something that like a lot of people in the Dynamis community have been sharing that they're getting good results with is using something like a smaller Quen model uh for the exploration phase of an AI coding workflow. So you can you can uh rely on the larger model using less tokens overall like significantly less if you have a process up front using a cheaper model to basically set the stage for the planning. Right? So like you have some kind of spec that for something you want to build or you have a Jira ticket or a GitHub issue and you can have a exploration step where you take that context and you sprawl out into the codebase to figure out the parts of the codebase that have to be analyzed more heavily for the implementation. So it's basically like the leading indicator for the stronger model like where it should actually pay attention to. And that might sound like it's overengineering, but it actually makes a big difference for your cost and just the overall effectiveness of your workflow. Like the pure output quality is goes up when you have that exploration stage. And you might as well explore with a cheaper model because you don't need that much reasoning power. Like at that point, it's it's almost like just evolved pattern matching where it's like here's the problem I'm trying to solve or the new thing I'm trying to build and now like in my brownfield codebase, here are the parts of the code that I have to integrate with. And so exploration and then like even if it's green field development, you're building something from scratch, that exploration phase can certainly just be a lot of web research, right? Like best practices for the tech stack and architecture for this application I'm working on. Like that's something that a smaller model can definitely drive as well. So all right, uh let's see. How do I uh make money for a living with Dynamis or Consultancy? Do I have a full-time job? So, I do not have a full-time job. What I do with with my YouTube channel and Dynamus and the like AI coding enterprise trainings that I do, all of that is my full-time job. So, I I left I I used to work as a software engineer at a wealth advisory company called Prize. It was actually a pretty awesome job. Like I I appreciate a lot. I worked there for many years. And uh I left that three months after starting my YouTube channel because the income already replaced that my software engineer role there. So, yeah, it's definitely a big blessing. like I I get to do what I've always wanted to do full-time, which is content creation and education. And really, like above everything, education is the most fulfilling thing to me. Like even if it was something other than AI, like actually um believe it or not, I used to be a chess coach in high school for elementary kids. It's not like I was like some hot shot chess player. It's just for elementary kids, but like I actually really really enjoyed that. Nothing to do with tech or anything, just coaching chess. But um I so I was um yeah at that time I was like taking college classes for like my last two years of high school. It's called PSO if anyone in the states is familiar with that. And so like after my PSO classes I would just like go over to the elementary schools for most weekdays and just coach chess. So I got paid like you know like it was like $8 an hour. It was so bad. I got paid like $8 an hour in high school to coach chess. But it was just so fun to me and like really fulfilling. So, I've always been an educator at heart. So, it's really really cool that like that's just what I get to do full-time now is educate like you guys on my channel in the Dynamus community for the trainings that I do for companies like helping them adopt AI coding assistance and creating a team standard for that. It's pretty awesome. So, yeah, that's that's how I make a living. All right. Uh after three months of YouTube, I was making decent money. I was. Yeah, cuz there was like, believe it or not, it was I I think I only had like 5,000 subscribers when I like I already got to the point where like my inbox was getting pretty full with people that were looking for a lot of different things like just contract work or um even like coaching calls like, "Hey, like help me build this agent for myself that you showed in this video." Um, so it obviously started like pretty simple and like it wasn't like paying a ton, but like there was a good amount of opportunities and then um started getting some sponsorships for YouTube. It was pretty menial at the time. Like that plus ad revenue was pretty small at the time, but like that started to to increase and it just became obvious that like I could definitely take this a lot further. And so like even though because like a salary like if you're making the same amount of money in entrepreneurship that you are in a full-time job, usually that's a good indicator that you should now go down the entrepreneurship route full-time because it might be more unstable, but there's just so much more opportunity. Like I just I really saw the opportunity at that point. Like I I only started my channel as a hobby because like I said like doing chess coaching and I was like a teachers assistant in college for uh uh like a C and assembly course. Like I' I've always loved teaching and so I'm like okay I'm going to keep working as an engineer. I'm going to continue to build AI things at my company and help my friends with their startups but I'm going to teach as well because I'm learning so many cool things and I haven't gotten to teach in a while and I really missed that part like being an individual contributor at a larger company. I just didn't get to teach. And so, uh, I yeah, I was just having a blast with that. But, but then the channel blew up really fast. So, I was like, "All right, hold on. Wait a second. I need to start paying more attention to this because this this is like not it can be a lot more than just a hobby now." So, that became apparent very very quickly. Yeah, you got paid uh $8 to coach chess. Yeah, appar unfortunately coach Chesting does not pay extremely well. U unless you're like really really high up there. So, um yeah, I don't think this guy minds if I call him out here. John Bartholo Mu. So, um Oh, geez. Apparently, there's a lot of John Bartholomew. This guy, he's a a a chess coach and u he's like really good. He's pretty cool. Um he he charged like at the time when I was in high school, he charged like $80 an hour for chess coaching, which is pretty good. Uh definitely like less than an AI consultant can make still, but like it's good. $80 an hour is very good. And so like yeah, you you can you can make more money, but like John Bartholome like this guy is legit. Like he is so good. Um I don't know. I don't think he's a grandmaster, but I believe he is an international master. I mean, we could we could look him up here on USCF. Let's let's search him really quick. Um, Artholomew John. Uh, oh wow. Yeah, I mean he's he's 2534. So, um, is he okay? He's he's an international master, so he hasn't reached grandmaster status, but anyway, you have to be like very very good to make decent money in chess. Like, it wasn't really something I ever considered pursuing as a career, but yeah. Anyway, so there can be some money, but yeah, usually don't make much. Job at a large company? Not for me. That's just such an ancient business model. Yeah, I mean the corporate red tape is pretty frustrating when you're at a larger company. That's for sure. All right, cool. Um, oh wow, you guys got a lot of questions in the chat that I haven't gotten to yet. Um, I think I also missed some as well, unfortunately, because like when the when there's more chats that come in, like the ones at the top go away. Let's see. So, I'm going to I'm going to knock out a couple of uh other questions quick here and then we'll get into another game. So, as a beginner learning panti lang graph, will these frameworks remain relevant longterm against native provider SDKs? Uh, the answer is yes. So um there's a lot of SDKs out there like Pi, Codeex, Claude, these Asian SDKs that are going right on top of the harness that people are using for AI coding. And you can use these SDKs for a lot more than just coding. You can build a lot of agentic workflows with them. But you're always going to have more control, scalability, and it's going to be cheaper to use a more, you know, quote unquote classic framework like Padanti or Langraph. These agent SDKs are expensive and slow. So they're really really powerful. Like the harness gives a lot of capability to the large language model out of the box. And so like that's why I built my second brain on top of the cloud agent SDK and now the PI SDK. But for any kind of agent that you're going to deploy to production and have hundreds or thousands or even millions of people using like you can't really use an agent SDK because you're you're going to have a platform where people expect speed and you have to be able to scale and it gets so expensive like these these SDKs the coding agent SDKs they are built they're optimized for output quality and so like yes you want your agents to always have good output quality but they're optimized to the extent where they're slow and expensive like you have bigger system prompts and it have more like reasoning chains and um the tools aren't optimized for speed. Yeah. So, generally the split that I like to say is when you're building a more of like a personal agent like a second brain or like what I built with Archon, then using a coding agent SDK makes a lot of sense. Or if it's just like a simple internal agent for a team of just like a few engineers, like if you're building like a GitHub issue triager or something like just use an agent SDK for that or a provider SDK. But if you're building anything where it's like it becomes a part of a plat a larger platform like that should definitely be using a framework like panticai or lang graph or like using both of these together like I teach on my channel or I have in the past and there's other really good frameworks out there now like these uh frontier model providers they they also have like separate agent frameworks like Google also they have their anti-gravity SDK but then they also have the Google agent ADK for building these more production AI agents. OpenAI has the OpenAI agents SDK. I actually have a YouTube video coming out uh this Wednesday here on the Google ADK because they have like a really really good skill that makes it so incredibly easy to build AI agents with cloud code and even deploy them to production. So you don't even have to think about that anymore. It's pretty neat. All right, cool. Uh yeah, so I'm going to go ahead and I think there was a question on the second brain. So, I'm gonna pop that up here and then I'll get into the next game. I got to see if I can find that maybe. Oh, that might have already gone out of the comment list, unfortunately. Someone was asking about my second brain. Oh, here it is. Andrew is asking, "In your second brain course, will we learn how to build that personalized news YouTube aggregator system that you built? Seems pretty compelling." Yeah, I appreciate it, Andrew. Um, and so it's not a part of the second brain course itself, but there is a follow-up workshop that I did in the community where I show you exactly how to build it. I call it my AI signal engine. Maybe I could pop that up really quick. Let me open up my uh profile here because I have to go into the Dynamus community to show you that u because I I just this Chrome profile is not my usual one. Let me open this up really and then we'll get into the next game. All right. AI signal engine. Well, here I'll just go to the I'll just go to the past workshops in Dynamus. So, I maintain this list of all the workshops that I do in the community and uh yeah, here it is. Build your own AI signal engine. So, I actually did this. This is like a special workshop because it was the one-year anniversary of the Dynamus community. And so yeah, this is like a one and a half hour session where I basically give away my entire system and I um have it like all creat like all documented. Oops, I'm not in my GitHub here. I got to sign in. Oh man, it's taking me to do my 2FA. Hold on. Am I Am I signed in um on my other profile here? I might have to do the 2FA just to show it quick. That's annoying. Um, hold on. It's private because I'm not signed in here. Ah, shoot. Okay, I won't do my 2FA right now. Anyway, so that repository, I'll just say like it has like my full AI engine implementation so that you can um just copy it yourself, right? Like you can incorporate it yourself or ask your coding agent to look at it as a reference. Like I give away the whole thing, easy for you to bring in and you can customize the different sources that it pulls from. Like I like I said, I have like hacker news and some RSS feeds and um yeah, like YouTube. Yeah. So I actually list it here. So GitHub repos, archive, hacker news, RSS, YouTube, and then it filters that from like it takes out the noise and it gives you the signal based on the things that your second brain knows you care about. And then it can help you like create scripts for content if you want as well. But that's an optional part. Like even if you just want to use it to help you keep up with the news, then you definitely can. So yeah, I'll put a actually I'll put a link to Dynamus in the YouTube comments here if you guys are interested in building your own second brain or uh just this content engine specifically. I would encourage you to check it out. Love to have you in the community. I do weekly workshops like this, like always giving away a ton of value. And then as far as courses in the community, I have the second brain course and then I have a course helping you master AI coding assistants and then one for building AI agents and deploying them to production. And so that's kind of the split that I was talking about where it's like if you want to build something for a full platform like padantic AI langraph that's what I teach here. If you want to leverage AI coding assistance and these coding agent SDKs that's what I teach in this one. And then using uh an agent SDK for the second brain as well. So yeah. Yeah. I don't need to do the 2FA right now. Uh I mean there's not like too much to show if unless you want me to like click around all the files, but I also want to get into the next game of chess here. So, let's go ahead and do that. All right, let's see. Someone asked about convex. Okay, man. Maybe I'll chat about that a little bit. Um, I'll start up a new game here. Let's do another five plus three. I want to see if I can actually get to 2200. That would be nuts. And I do have to check out of my hotel within the next like 45 minutes here. So, I can't, but I was planning on doing this live stream for maybe like hour and a half, hour 45. We'll see. Um, yeah. Okay. Oh, wow. C4. I haven't had someone play this in a long time. I actually don't even remember what this opening is called. Um, let's see here. What do I want to do? Yeah, I'll probably do that. Honestly, a lot of my theory of chess, I I've retained a good amount of it, but I used to be one of those people that was super tryh hard with my opening theory. Like I would study the book, like I had so many books, especially for like the Sicilian because I'm a big Sicilian player. Unfortunately, we don't get to see that today, I guess, but I I love the Sicilian. So, I study the heck out of that. But, yeah, unfortunately, I've never really gotten to uh or like I haven't really retained 100% of it. I'll just say. Uh, okay. What am I doing here? Let's play it safe. All right. Boom. Let's get decent start. And then the thing that's really fallen for me is endgame theory. I've forgotten so much of my endgame theory because I haven't studied I didn't really study it as much. It just wasn't as fun to me, right? And then there's like so many games where it really doesn't get to the endame theory because you have some at least one person has an advantage earlier on. So it doesn't like come to that being as important like the Lucina position and things like that. For those of you who know chess, you know what I'm talking about. I I just I don't really remember. Like if you put me in a Lucina position right now, I I don't think I would be able to get the pawn promoted, I would Yeah, that whole like rook on the fourth or fifth rank thing and like I don't even remember when or how to do that. Yeah. But like sometimes I think about getting back into chess again. Like it's such a fun hobby. Like if I if I were to like really go crazy uh like I did back in junior high school, like that would be fun. I would enjoy it. Yeah, this this position is pretty solid. I don't have any complaints yet. I'd hope by moon by move nine I don't. But you never know when I'm trying to talk at the same time. Sicilian is a very deep opening. Uh yes, there is a lot that goes into it. Um oh, you're 2,000 ELO. That's awesome, dude. Cool. C4 I always mirror. Yeah. Okay. I I respect that, but at the same time, in my experience, when you mirror a C4, the game is just kind of boring, right? So, like maybe if I really wanted to play optimally and play it safe, I would would have mirrored. But like since I'm doing a live stream here, I don't really want to just like get into some super locked down position where I I'm like kind of fumbling like, well, what do I even do right now? You know what I mean? So, I don't know. But like, yeah, mir mirroring is definitely a decent idea. All right, let's see. Nothing too crazy happening yet. Um, I'm going to go to elite chess really quick. So, I'll show you guys the Sicilian opening that I like playing usually. Oh, hold on. I got need I need to actually make my move here. Um, I don't know. There's not too much craziness going on here yet. So, I think I'll just do that. That's good. All right. Um, chess practice, I think. Wait, I I don't actually know how to open just a chess board. Oh, no. Yeah, here it is. Analysis board. Okay, then. Can I Can I flip the board? How do I flip it? Flipboard. Yeah. Okay. So, uh my favorite Sicilian defense is um wait, hold on. Oh, is this one right here. So, instead of uh like most people will do like this, right? Like this is the most common Sicilian. And then um well, actually what I really used to like playing is the uh Najdorf defense, right? Isn't this the Najdorf? Why is it Why did Why' the wiki book go away? Did I miss a step? Man, my theory is Oh, shoot. My turn. Hold on. Let me Let me play this game, then I'll go back. My theory is getting rusty now because I have Well, I haven't played the Najdorf in a long, long time. That used to be my favorite opening. I'll go back to the board after I uh make a move here. Um Okay. Yeah, I must have missed a move. Anyway, so I like doing this and then um this is this is the most common and then take take and then this right here. I forget what this opening is called. I thought it was called like the old Banani defense or something, but that might be something else. But it's crazy because like the best move here is actually knight here. Like it's really random. But yeah, that that's like the newer version of the ceiling that I played more recently that I've been pretty fascinated by. All right, looking good because he can't take or no he can but it's just a trade. That should be good. Wait, if he takes me Wait, does he actually win a pawn from this? Hold on. No, we're good. We're good. Okay. I just have to figure out which direction to take. This could be interesting. just do this. Yeah. So, I take here. I think I just risk this pawn getting kind of isolated if he plays here. This This is probably better. Sacrifice the rook. Should should I have sacrificed it somewhere or you just me? Where could I have sacrificed a rook? I've never played go. Um I have I have not played Go before. Never. It's just uh not something that I've never heard of anyone playing really because it's more popular in China, right? like a really really popular game in China. And I I know that like back in the day with like um Deep Blue like they were teaching it to play Go and chess, right? That's probably why you're thinking about it, Sean. But no, I haven't um I haven't played Go before. Gotham meme quote. I I haven't seen Goth. Are you talking That's like a TV show, right? Sorry, I'm like really ignorant there, but I I'm not uh familiar. Sorry. I I probably should get the reference, but I don't. All right, let's see. Proposing a pawn trade. Um, let's start with this. Yeah, I'll start with that. Get my queen out there and then kind of think that there's some way I would want maybe want to like shove get my rook in there. This I don't think makes sense. Well, actually this might make sense. This looks kind of annoying for him. You know what? Screw it. I'm going to do it. He has to take the rook. Well, actually, that's not bad. Just to take the rook. Maybe that wasn't the best move. Hard to say because I do I do have to take now. Yeah. I don't know. It's all right because Not really sure what's the best option now cuz now that his bishop my my queen's aren't really doing much here anymore. So I might just bring her back honestly. But yeah, this this doesn't look ideal now that he has open file. I would say he probably has a slight advantage now. It's nothing rough, but yeah. All right, so 30 seconds left. I'm going to focus on this game now. Oh, you're talking about Gotham, the big chess crater. I know who you're talking about. Yeah. Yeah. I'm sorry. I know Gotham. That's my bad. Yeah. The Yeah. Yeah. Yeah. I don't know why I thought you were talking about a TV show. Um, let's see. Try actually. No, the pawn's protected. There's not really much I can do there. think I'm okay. Is he going to force a draw? I mean, he probably I think this game is going to end in a draw, though. Like, obviously, I have a time disadvantage, but with the 3se secondond increment, I'll probably be fine. I just have to focus here. But I don't really think there's a way for either of us to do anything. I mean, it's not opposite color bishops, so there's definitely more of a chance of one of us doing something, but boy, does this look like it's going to be a draw. The one thing that like I think might actually be kind of nice for me is, you know, I I can keep pressure on this pawn. He can't pressure my pawn. So, I might actually want to push this Actually, I might do this. Or wait, no, that was Sorry, that was Damn, that was dumb. Cuz he's just going to go here. I was thinking like, let's get ready for an endgame and push my king up, but I think I just kind of blundered. It's not that bad, honestly, cuz Yeah, I'll just do that. This actually it worked out. I guess maybe maybe he saw that and didn't want to do it because then then I'm just getting my queen up in his business over here and he's kind of getting it off to the side. I I I think he probably still did miss that as an ideal move. Oh, interesting. Uh, sure. I'm going to take that. I don't have too much time to think. That might not have been the best, but I like my position more, I think. I guess the one problem is I don't really have a way to get my king up in here. Yeah, I got to move fast. And then he he creates another pass pawn. So I might have messed up. Yeah, that actually that does look kind of bad unfortunately. I don't know if you can push it though. I think I'm I think this is actually a draw cuz now he can't he he can't push it. There's no way to to get his bishop to defend. His king is stuck there. And I'm going to bring my bishop around. looking like a draw. But I still have to pay attention because I only have 10 seconds, right? All right. We might just be shuffling bishops now. Like if he goes back here, I'm just going to do the same thing. Well, here I'm going to go Wait, this might actually be uh Okay, no, it doesn't really help. This might not be ideal. Take. Take this guy. Oh, wait. I messed up. Dang. I think I just threw it. Oh, that's actually good. Shoot. Yeah, I lost the game. Oh, that's too bad. He's going to be Well, actually, this might not be It still might be a draw, but I definitely messed up there. Might still be a draw. Move fast. All right, it's a draw. Jeez, that was an intense game. I didn't play the best there, but I'm happy I at least ended with the draw there. All right. All right. Uh, in Dynamus, what do you mean we're using Dino Chat? Um, yeah. Okay, let's let's chat about Dino Chat really quick. So, this is something. Let me pull up my other here. Whoops. Chat dynamus.ai. So, diet chat is something that I worked on for both my YouTube channel and the Dynamus community. So, it is a an agentic rag application where it's just your AI tutor. You can ask it any question about AI and it automatically searches through my YouTube content to find the answer for you and it'll even cite it the sources so you can like watch the videos and then if you're in the Dynamis community the agent is evolved to not just search through my YouTube videos but also all course and workshop content in the community. It's pretty cool. So, if you sign up for Dina Chat in the same with with the same email that you use to sign up for the Dynamus community, then the agent automatically unlocks this next level of search. So, I'll put a link to this in the chat. If you haven't seen this yet, especially if you're in the Dynamis community, you should definitely check it out. But it also is free to use. If you're not in the Dynamis community, you can still use this to ask questions over my YouTube content, which is pretty cool. Like right here, I just asked like what does Cole recommend doing to create AI layer for existing codebase? Um, and let's see, uh, like how do how do UI use sub aents? What rag strategies does Cole cover in his content, right? And I can like find the rag strategy like the references here. Like here's a workshop that I did in Dynamis a while ago on knowledge graphs. Uh, so that's pretty cool. So yeah, it's a really neat application and the the responses are a bit slow because I prioritize it being very comprehensive and it searching instead of being like really fast. So you have to be patient with it, but you get some pretty good results. So yeah, that is uh that's Dino Chat. It's uh pretty pretty worth checking out there. So all right, cool. Well, I'm going to go ahead and play one more game here and uh then I will call our stream because his time's flying by. This is pretty fun though. Um, I appreciate all you guys being here. Like definitely like a bit of a smaller stream than I normally have, but uh, makes sense because we're just chilling and playing chess, but like this is pretty fun. So, yeah. Let me let me go ahead and we'll do one more game here. I want to see if I can push to 2200. I've never been to 2200. I've never been to 2100 actually. Usually for online chess, I sit around 2,000. So, this is pretty neat. Um, yeah, definitely prefer complete over speed. Did Cole win at chess on Saturday? Well, it doesn't ingest my live streams. Uh because that's like just so long. But um I I guess I might want to add that. I don't know. Uh Whoa, that is weird. Okay. Also, this guy is 2,300, so that's a little scary. All right, 2300. If I win against the 2300, I'll be pretty excited. I've done it before, but it's definitely rare. All right, he is putting the pressure on. I guess he knows what he's doing with this opening here. Um, all right. So, he's sacrificing a pawn, but I feel like that makes sense. I don't think I want to accept the sacrifice. But what can I really do? Yeah, whatever. I'll accept it. Or no, I guess it's not if he takes. Okay. Yeah. So, I I I'm a pawn up, but uh it looks kind of not the best. He's going to castle. Yeah, I mean, that actually looks pretty good for him. He's going to castle um queen side here. So, I'm I'm just going to let him have the pawn if he wants. actually looks way better for him. I think I like pretty I played pretty poorly here. So I'll let him take it. And then unfortunately because of the pin threat I I just have to Well, I guess I can do this. If he plays that, I'm just moving over here. Yeah, I mean it definitely looks better for him. I I honestly don't know how to handle that opening. Um I guess I'd be curious to go to the analysis board now that Well, okay, I should wait until the game is over or I'd be tempted to like Yeah, I'll wait till the game's over. But I'm curious what the optimal way is to play that. Oh, interesting. Yeah, that's a good move. Respect that. Let's see if I kick him out. Yeah, I don't want to screw this pawn over. Yeah, this is a bad game to end on. [laughter] Um, maybe I will do this. Doesn't seem too bad. Probably won't take, but premove just in case. All right. Gonna get smoked on this one. I might get smoked on this one. We'll see. Uh oh, he actually did take. Okay. Oh, but he has a fork, so he's get Yeah. Yeah. All right. All right. So So he'll probably I think he can just take here. Yeah, this is probably a lost game. Uh I mean, I don't think I'd win to a 3330. All right. I also should ingest the live streams because they have the most diverse variables and mentions of thoughts and assessments on wider software. It would say Colon talk about X, but I mean that's that's fair, Stuart. Yeah, I respect that. Maybe I Yeah, maybe I should. Maybe I really should. I'll have to I mean there's obviously a lot of of different ways I can improve the AI tutor application like Dino Chat as Sean called it. I guess that is like the official name for it. So yeah, Dino Chat there there are a lot of ways that I can improve Dino Chat. Um so I'm continuing to work on it. Yeah, adjusting war content. It Yeah, that could definitely make sense. All right. Uh let's see. Knight. Which way do I want to move my knight? Probably this looks nice. just kind of just perch it here. It's just kind of unfortunate that he gets to double up his rooks. And um yeah, so it's hard it's going to be hard for me to contest. So I'm just going to play here because he can't fork me because of my knight. So I'm going to end up Yeah. The only problem is like Yeah, it's not looking that great. because he's actually threatening check. Wait a second. I think I'm actually toasted. Like, not yet because I can play here, but it's looking really bad. Can't do Yeah. Dang. Yeah, that's looking rough. I'm not doing the too hot here, that's for sure. See, I think he's just gonna shove pawns up my butt. Or or just take some pawns. Sure, that works as well. Um, yeah, not good. And then, uh, Yikes. That is rough. That really rough. We just going to take this. Yeah, take that way. Oh, man. May Maybe I'll end with some bullet chests. That could be fun because this this game is pretty hopeless right now. Uh if I take here I can't that checkmate. If I take here, then he's going to take. That forces me to go here. And then he moves his rook there. And that that actually goes checkmate. So, I can't even do that. This game's lost. It's lost. Can we get a tutorial on how to combine Archon with Hermes? So I did actually do a workshop in my community on uh incorporating archon with your second brain and so definitely like taking that doing that as a YouTube video as well. I would be interested in doing that. Um, so yeah, like obviously like second brain that that includes Hermes, right? It's kind of the same process like you have archon as a skill that you give to your second brain like Hermes, openclaw, whatever. And then the thing that I taught in that workshop that was really valuable is showing how I built a system for my second brain to have an index of all my repositories and an understanding of uh how I like to actually oops how I like to actually use archon on each of my repositories. Um, and so like that way it knows like custom workflows that I use and like how my development flow typically looks per project because it is actually different depending on the project. So yeah, that was that was pretty cool. Um, uh, wow. I'm I'm like totally destroyed here actually. At least I can do this. I mean, I have like something here, but it's barely anything. like he's clearly won this game. At least I can win a pawn. He doesn't have any immediate threat of checkmate anymore, I don't think. Yeah, you're done. Yes, I am done. I I understand. Oh, wait. This bottle's I'm I'm out of liquid, unfortunately. I down the whole coffee and sparkling water. All right. What else do I have here? I can take this, but then this pawn's getting too far. So, I'm going to do that. But then, I mean, he's just he's just going to clean house, right? Like take and then I mean, he might do something else, but like take take and he's still up a ton of pawns and I don't have any counterplay. So, we'll see. I'll I I'll play this out a little bit, but maybe I want to end with like a game of bullet as well. Um, all right. Let's see. I'll just do that. I think he's planning on just taking all my pawns. Yeah, go for it, dude. All I got I mean I got a couple threats like Yeah, but it's I mean this guy's not He doesn't have a queen to even drop anymore. Oh boy. Let's see. So, yeah, I know there's like a lot of you guys throwing some other questions in the chat that I'm not getting to here. So, definitely can't get to everything right now in the chat as much as I wish I could. But, uh for everyone that's like throwing some things in and uh chatting with me, I appreciate it a lot. I hope that you guys have gotten something out of the live stream today just like talking about about AI in general. This is this is really fun. Like I I enjoy playing chess even though I can't play as well nearly as well like this, but it's still it still is really fun. Um yeah, I mean like this point I can't do much. I don't even know like where I could bring my knight to do anything really. Um man, let's put it back here. I don't know. I'll play a game of bullet after this. All right. Or shoving it up my face. That is fair move. Yeah, I guess I lost this pawn as well. Yeah. All right, I'll go ahead. Well, okay. I won't I won't click resign. I'll I'll But I'm gonna run out of time here. Or Yeah. All right. I ran out of time. All right. Bodhija. Bodhija. Good game. I I'll say GG. All right. Cool. Let's play a game of bullet to end things off here. Uh let's do bullet. All right. So yeah, actually this will be pretty good timing. Cool. Let's see. All right. Bullet game. Let's go. Wish me luck. Gotta move. I won't be able to talk for this one. I'm gonna move fast for this one. Oh, they're going lightning speed. You don't have to play that fast for bullet. It's going decent. It takes some uh Okay. I'm going to be kind of silly here. I shove pawns up his face. Random. It's kind of all right, though. I think he actually he has to play rook here or I checkmate him. This game's going much better than the other ones. I should just play a bunch of bullet chess. What? Okay, I only have 10 seconds, though. Losing on time is is actually a risk here. All right, I'm gonna wipe the board and make sure that I at least get a draw if I run out of time. Oh, shoot. a stalemate. That's so unfortunate. I I didn't have time to think about the stalemate and I stalemated him. Man, that would have been such an epic win if I had actually pulled through. Okay, I got to do one more. I got to do one more. Dang it. Okay, one more. One more. I gotta win. Just one more. That's what they always say, right? Just one more. Okay. All right. Yes. I actually I actually know this opening. I know how to play this one. All right. That's good. Oh, that's wild. Okay. I don't think that that F4 is usually common there. Not really threatened by that. Let's do that. I'm going to slow it down here. All right. Not worry about that either. Oh, that was kind of a dumb move because I can't Okay. Anyway, this is bullet. But I got to play fast. Can't really do anything with that. He brings his rook up. Yeah. So, I'm going to have to That was dumb. I'm making dumb moves, but it's bull. It doesn't matter. It's all good. Force his queen out. Um, that's actually fine. Oh, shoot. I could have just taken that. That's too bad. Yeah, I'm falling apart. Oh, what am I doing? I can try to win on time. I won on time. Oh, bullet chest looks so bad at the end. Oh, boy. All right. Well, there you go. That That's a That's a dumb win, but I got another win there. So, cool. Got my bullet up to almost 2,000 there. [laughter] Oh, man. Just throwing random stuff at the end. Sometimes it just works. I I guess I pulled through. All right, cool. All right, so yeah, I'm going to go ahead and end the live stream here so I don't get in trouble for not checking out of the hotel on time. But, uh, yeah, this was fun. It's such a random stream, but like I en I enjoyed doing it and I appreciate all you guys tuning in just uh chatting about AI as much as I could. I I think I focused more on the AI chat than the actual games. So, we got a lot out of that and I I played okay. I actually played okay for how much I was um multitasking. So, not bad. [laughter] Yeah. Appreciate all you guys being here. Hope that you guys have a fantastic rest of your weekend. I'm going to be working hard on some big videos I got coming out for next week. Um mostly actually tomorrow. I'm going to take some of today off, just relax and continue to enjoy the anniversary celebrations. But yeah, appreciate you guys being here. Hope you guys have a great uh weekend, like I said. And uh I'll see you guys