Pełna transkrypcja odcinka Local Models, Foreign Affairs | The Case For Private AI Archives

to keep up in a perfectly straight line. But but if we compare, you know, October even back to July and August, we're considerably outpacing it still. So still, you know, start measuring whatever, a six... Sixty day increment, and there's considerably more throughput, which just ultimately means more activity. 31 00:02:25.819 --> 00:02:29.660 On to other things. Let's also just take a look really quick. 32 00:02:29.900 --> 00:02:36.325 And, actually, I'll share my screen here for this one. Sorry. I forgot for a second that we have the technology. 33 00:02:36.645 --> 00:02:40.805 We're actually live and live and in person. Yes. Yes. We can do that. 34 00:02:41.605 --> 00:02:43.845 Activemora.active.mora.org 35 00:02:43.845 --> 00:02:47.045 is where I'm looking right now. Just checking out what is 36 00:02:47.620 --> 00:02:50.740 routable. So nearly 240 37 00:02:50.740 --> 00:02:56.340 routable models, a 110 of them being redundant, meaning that at least two 38 00:02:56.820 --> 00:03:00.580 compute providers, which there are 20 online currently, 39 00:03:00.660 --> 00:03:02.260 at least two are providing 40 00:03:03.235 --> 00:03:05.635 that same model, which is what makes it redundant. 41 00:03:07.235 --> 00:03:07.875 So 42 00:03:08.275 --> 00:03:26.380 I saw some... Actually, you know what? We'll we'll save that for a little later in the show. I have I have some comments here, but we'll we'll come we'll come back to it. But, yeah, this is awesome to see. Tons of redundancy, tons of models. This number should just continue to grow as well as people build things and they want access to various models and compute providers, 43 00:03:26.620 --> 00:03:27.740 you know, can obviously 44 00:03:27.980 --> 00:03:30.300 see that, take take advantage of those opportunities. 45 00:03:31.025 --> 00:03:33.665 And and this is, in my opinion, like, 46 00:03:33.905 --> 00:03:35.505 the ethos of 47 00:03:35.825 --> 00:03:37.905 Morpheus the Protocol because 48 00:03:38.065 --> 00:03:39.025 I want 49 00:03:39.185 --> 00:03:40.305 permissionless, 50 00:03:40.385 --> 00:03:41.505 perpetual 51 00:03:41.585 --> 00:03:43.265 access to intelligence, 52 00:03:43.820 --> 00:03:48.460 and, like, that is literally this. Like, this is, you know, what that looks like, basically. 53 00:03:50.380 --> 00:04:00.425 Not sure if you guys have any comments there. Otherwise, I would like to hear what's going on in your guys' world and, you know, anything new that be shipped or is about to ship that we can we can talk about from the community. 54 00:04:01.145 --> 00:04:01.945 Well, 55 00:04:01.945 --> 00:04:04.825 just on that point, you know, there was a 56 00:04:04.985 --> 00:04:08.505 realization by some people this week that 57 00:04:08.665 --> 00:04:11.385 if Navidea changes course on open weights, 58 00:04:11.829 --> 00:04:14.470 and I believe they purchased Hugging Face. 59 00:04:14.790 --> 00:04:20.710 They did. So They did. Now Hugging Face has started removing models. People are like, maybe 60 00:04:21.270 --> 00:04:24.150 we need a permissionless way to access models. 61 00:04:24.665 --> 00:04:26.425 And little known fact, 62 00:04:26.665 --> 00:04:28.905 there's an IPFS node 63 00:04:29.225 --> 00:04:30.425 built into 64 00:04:30.505 --> 00:04:32.505 every single Morpheus 65 00:04:32.665 --> 00:04:33.865 inference node. 66 00:04:34.105 --> 00:04:37.145 And so I've got mine running. I 67 00:04:37.545 --> 00:04:43.640 would encourage others. If you wanna be able to download and persistently store in a decentralized way 68 00:04:44.040 --> 00:04:45.000 the models, 69 00:04:45.240 --> 00:04:55.015 it's a great thing to do is to contribute to the network in that way as well. So just highlighting that, just kind of putting that out there, you know, in the off chance 70 00:04:55.175 --> 00:04:57.495 that there's some feature censorship of models. 71 00:04:57.815 --> 00:05:05.495 Yeah. It's it's funny you say that, David, because that was exactly the thing I was about to comment on, but I said I would wait, but now I'm not going to. Yeah. This... 72 00:05:06.070 --> 00:05:10.390 There seems to be I scooped your bow. I scooped you. Here we go. I'm 73 00:05:10.870 --> 00:05:21.405 glad you did because, yeah, there seems to be some waking up to the fact that this is a possibility and and if not, you know, a likelihood that something like this could happen. And there there 74 00:05:21.405 --> 00:05:27.245 are some big players on X asking this question, like, has someone built this yet? 75 00:05:28.044 --> 00:05:28.764 Yes. 76 00:05:29.164 --> 00:05:33.405 Yes. It exists. Yes. We have. Yes. Yes. The community has. Yeah. We have built this. 77 00:05:34.660 --> 00:05:35.300 So 78 00:05:35.620 --> 00:05:49.685 definitely check that out. Think about that, what the implications are. And, like, it just highlights the importance of Morpheus. Like, this is why it exists as a... You know, owned by no one, no central 79 00:05:50.325 --> 00:05:52.565 entity. Like, it's it's permissionless. 80 00:05:52.565 --> 00:05:54.485 Everyone has access to the contracts. 81 00:05:54.485 --> 00:05:56.565 The the node can be used. 82 00:05:57.045 --> 00:06:00.325 It's a public good, ultimately, is what I'm trying to say. And, like, this is exactly why. 83 00:06:01.770 --> 00:06:09.610 So, yeah, very... Not cool in one sense, but cool in another sense that, like, yes, people thought of this, which is which is nice. 84 00:06:10.090 --> 00:06:14.890 Alan, do you have anything from your side? I know that you're always doing stuff. Like, what's what's the latest? 85 00:06:16.195 --> 00:06:16.915 So 86 00:06:17.875 --> 00:06:33.050 on the note, we... I think we talked about this a couple weeks ago, and and David mentioned it briefly on video. I do not. I'm terrible. My my goal is to at least have one good release a week, and so this week, I've I've failed. But I think a couple weeks ago, had two, so maybe that that makes up for it. 87 00:06:33.930 -->

00:06:36.010 But we are working hard on 88 00:06:36.330 --> 00:06:46.575 making sure the video stuff is routing well. So not just to the node, but to the API gateway and then the app as well. And that's a little bit heavier lift than your normal LLM and LLM and chats. 89 00:06:46.975 --> 00:06:52.015 So we want... Again, we wanna make sure we get it right. We also have a other bundle of fixes 90 00:06:52.575 --> 00:06:54.175 from the community as well. So 91 00:06:54.970 --> 00:06:59.210 no three... No dot seven dot two zero today, 92 00:06:59.930 --> 00:07:06.650 but soon. And that's gonna have some of the video stuff, and I think we'll be talking about that in the coming weeks. A lot of good stuff there. 93 00:07:09.335 --> 00:07:10.695 On Node. Neo, 94 00:07:11.735 --> 00:07:15.335 three dot eight zero was released a couple days ago and 95 00:07:16.215 --> 00:07:17.015 per 96 00:07:17.015 --> 00:07:18.775 community request on Android. 97 00:07:19.255 --> 00:07:23.015 So that's up on the Play Store. Look for node neo, all one word, no spaces. 98 00:07:24.750 --> 00:07:35.470 That's that's easy. I need to figure out how to get the the searches right to... You know, when you search it for Morpheus or something like that, it comes up as well. But Node, n o d e n e o, 99 00:07:36.405 --> 00:07:39.685 if you want the decentralized AI in your pocket 100 00:07:40.245 --> 00:07:43.365 in either Android or or Apple devices. 101 00:07:44.565 --> 00:07:50.965 And by all means Alan, I know that was one of the big requested things is get it on Android. Now it's on Android. 102 00:07:51.740 --> 00:07:59.020 It was it was, I think, milliseconds after the the announcement post, someone, you know, the usual, you know, 103 00:07:59.500 --> 00:08:00.780 when droid. 104 00:08:00.940 --> 00:08:04.780 And so I had I had fun answering them yesterday. It's like the time is now. 105 00:08:05.865 --> 00:08:10.265 So that's that's good. I'll be interested to see how that uptake is and some of the feedback. 106 00:08:10.665 --> 00:08:15.225 And see... And it's funny. It may... That may drive different model behaviors as well. 107 00:08:16.105 --> 00:08:27.870 You know, what do people wanna have on deck on hand when they're not doing big development work? Just normal everyday chat. How's my finances? What's my health look like? Any of those private conversations that you'd rather, 108 00:08:28.190 --> 00:08:41.485 you know, not have as training material for someone else. So as usual, Node neo, your AI should not take notes for somebody else. Oh, I like that. Thank you. Also, Morpheus Asia, if you're listening, it's on Android. 109 00:08:42.045 --> 00:08:43.005 There you go. 110 00:08:43.565 --> 00:08:44.285 There you go. 111 00:08:45.470 --> 00:08:47.630 So exciting. Asking any promises made. 112 00:08:47.950 --> 00:08:50.350 Yes. Yes. Promises kept. 113 00:08:50.510 --> 00:08:54.270 Well done. Yeah. Those are... So those are those are the software releases. 114 00:08:54.670 --> 00:08:55.870 Got a couple much 115 00:08:56.510 --> 00:09:04.904 talked about some of the the upcoming smart contract changes as as tightening up. We've got some epics that we're kinda trying to lay out over the next couple of months. 116 00:09:05.865 --> 00:09:11.144 But, again, no shortage of changes and needs and feedback. So keep building 117 00:09:11.580 --> 00:09:13.340 as the man says. 118 00:09:14.700 --> 00:09:15.980 Yeah. Absolutely. 119 00:09:16.620 --> 00:09:21.580 David, anything change on your side of things or any... Anything else that you're aware of that ship this week? 120 00:09:22.540 --> 00:09:39.565 You know, I think this week has been sort of dominated by the the drama the last few days. So Yeah. You know, I think the question is, what is the effect on us? What is the effect on the Morpheus community? You know? The the the sort of sad thing is, you know, Jensen was 121 00:09:39.725 --> 00:09:40.365 the 122 00:09:40.600 --> 00:09:42.120 the supporter 123 00:09:42.360 --> 00:09:55.320 of open weight models. Right? Famously came out with his letter and encouraged everybody to support open weight models. So I think he officially resigned an hour ago Yeah. If I'm not mistaken, if you guys have checked under. So 124 00:09:54.834 --> 00:09:58.035 what happens now? Like, does NVIDIA 125 00:09:58.274 --> 00:10:00.835 change course? What's the new leadership like? 126 00:10:01.235 --> 00:10:06.195 You know? But it it kind of, again, confirms this is why we need protocols 127 00:10:06.595 --> 00:10:08.755 that aren't 128 00:10:07.880 --> 00:10:13.560 susceptible to leadership change and the wins of politics and all the rest. 129 00:10:14.040 --> 00:10:16.680 Yeah. No. Completely agreed. Although 130 00:10:17.000 --> 00:10:18.040 Spinzone, 131 00:10:18.200 --> 00:10:22.519 maybe Jensen takes a full dive into decentralized AI 132 00:10:22.755 --> 00:10:25.715 now that he's he's got some free time on his hands. 133 00:10:28.115 --> 00:10:47.850 So maybe, yeah, it's it's possible that he, you know, even further solidifies his stance to promote open source. Maybe there's this arc where... And I don't know who the new leadership is. I I have no idea. I don't even know if they've announced it or not. I'm sure they haven't. Oh, they have? Okay. Yeah. One of the one of the interim c suite is take a... The CFO, I think. 134 00:10:48.890 --> 00:11:02.225 I could be wrong. I saw that a couple of a couple minutes ago, but... While they do a CEO search. But... So they've they've got... It was interesting. I saw they didn't have a real succession plan. It was interesting that some of the write up talked about some of the culture that he brought. 135 00:11:02.625 --> 00:11:11.450 And it was like knowledge sharing at a broad level, but no true kinda succession planning, which will be interesting to see how that flows out in big corporation like that 136 00:11:11.930 --> 00:11:19.130 because of all the cascades. So the the fallout both

in in the AI, the models, the open source, 137 00:11:19.595 --> 00:11:23.035 the chip manufacturing in general, the relationship 138 00:11:23.275 --> 00:11:24.235 overseas, 139 00:11:25.115 --> 00:11:30.155 and their corporate structure will be very interesting. It's it's going to be interesting. 140 00:11:30.715 --> 00:11:31.035 Yeah. 141 00:11:31.690 --> 00:11:33.930 At the very least, it'll definitely be interesting. 142 00:11:34.250 --> 00:11:35.450 I 143 00:11:35.930 --> 00:11:47.290 wonder... Yeah. Like, maybe this is an arc where NVIDIA takes a super dark turn into, like, centralized megacorp AI ville, and Jensen tries to be the, you know, hero out of the ashes 144 00:11:47.585 --> 00:11:49.585 to become the open source, 145 00:11:50.065 --> 00:11:56.785 you know, like, champion of the world, basically. One of the immediate rumors was is Elon gotta hire him 146 00:11:57.105 --> 00:11:59.025 to build the TerraFab. 147 00:11:59.025 --> 00:12:00.065 Oh, yeah. 148 00:12:00.720 --> 00:12:05.600 They immediately released an open position to lead the TerraFab, 149 00:12:05.600 --> 00:12:26.875 which, to be fair, will be the largest chip foundry on Earth with the combined capacity of all other chip factories times 10 or something. Like, it's this absurd scale. So if there was anything on earth he's gonna do next, Elon immediately replied. Like, it would be on... An honor to work with him. An honor. So Yeah. Maybe a $100,000,000,000 comp package 150 00:12:26.954 --> 00:12:31.410 upon the, you know, completion of the world's largest chip factory, 151 00:12:31.570 --> 00:12:36.530 and that's his that's his new arc. I don't know. We're gonna find out. Man, that would be really What 152 00:12:39.730 --> 00:12:46.205 a day to wake up. And, you know, if that does end up happening that way, man, what a pickup for Elon. Like, I can't even imagine 153 00:12:47.645 --> 00:12:52.445 waking up and being like, oh, man. That's... It's like... I'm trying think of an equivalent. I mean, it's like... 154 00:12:54.125 --> 00:13:02.610 I don't know. If, like, all of a sudden, ten years ago, LeBron James just, like, became available and was like, yeah, sure. I'll join your team. Like, I have nothing else to do. 155 00:13:03.730 --> 00:13:07.810 Wow. That's crazy. Yeah. Well, I mean, we'll see how it plays out. Obviously, 156 00:13:10.065 --> 00:13:12.465 this is another instance of 157 00:13:12.865 --> 00:13:14.625 poor decision making 158 00:13:15.665 --> 00:13:20.145 being, like, a massive downfall to someone and, like, so avoidable. 159 00:13:20.225 --> 00:13:22.465 Incredibly avoidable. But, you know, 160 00:13:23.699 --> 00:13:24.660 here we are. 161 00:13:25.459 --> 00:13:34.740 Yeah. It's it's too it's too bad. The interesting thing if we go if we go down sort of the the the conspiracy theories is, like, other people in that position have weathered those storms 162 00:13:35.945 --> 00:13:45.705 better or... You know? So it's interesting that all of a sudden that came to a head and and it... And he's out. So you wonder, you know, what else. Right? What else is going on 163 00:13:46.505 --> 00:13:50.425 and where else? You know? Is it is it some of the noise and cry of of... 164 00:13:51.310 --> 00:14:06.110 From the political fallout from from a couple weeks ago or something else? So it's... Like anything else, this is why I can never get out of x is is all of a sudden, they they pull me back in, and there's something else like, well, that's interesting. Wanna kinda gotta see what happens after that. 165 00:14:08.575 --> 00:14:10.255 Maybe maybe maybe 166 00:14:10.255 --> 00:14:14.015 maybe Jensen rides off, buys a yacht, and is never heard from again. 167 00:14:14.575 --> 00:14:15.295 Or, 168 00:14:15.615 --> 00:14:16.335 you know, 169 00:14:16.735 --> 00:14:20.415 any of the spectrum. So it's always... It's it's like the the best and worst 170 00:14:21.129 --> 00:14:28.649 drama TV show you could imagine on X these days. I don't know. He... He's wired for it at this point. You spend three decades 171 00:14:28.970 --> 00:14:45.445 building. He's addicted to You're a builder. Like You're a builder. A serial entrepreneur that's addicted to it. Like Yep. You know, what's he gonna do in retirement? Like, play golf? Come on. Garden? No. That's... Yeah. Well, that's the thing that sad. That would just be sad. I always think that's funny because builders, 172 00:14:45.524 --> 00:14:47.285 you know, even flawed ones. 173 00:14:48.100 --> 00:14:51.380 I always think that's funny. You're exactly right. That's one of those things... 174 00:14:52.740 --> 00:14:59.700 Hey. What's going on? It's Bronze Buddy doing Timekeeper. Awesome. Glad you could join us, man. Welcome. Welcome. Hey, guys. 175 00:15:00.260 --> 00:15:04.555 Hello. I I just wanted to make one one point on that, David. That's one of those things where... 176 00:15:04.795 --> 00:15:30.190 It's exactly right. People are always like, oh, if if I, like, had this successful startup and made this big zag zit, I would never work again. And it's like, well, the people who do that will work again because that's how they got there in the first place. Like, they're not... It's generally not the type of person who's just gonna pack it up and be like, yeah. You know what? That's fine. I'm just gonna golf for the next thirty years. Off off to the wilds to fish somewhere. No. It... You gotta keep... Building is addicting. 177 00:15:30.805 --> 00:15:32.165 Yeah. You gotta build. 178 00:15:33.285 --> 00:15:34.325 Exactly right. 179 00:15:36.805 --> 00:15:37.925 Well, 180 00:15:37.925 --> 00:15:50.610 welcome. Why don't... I think it's your first time on the space. Why don't we get an intro? And building with Brian about how cool it is, and glad that you carved out time to join us. Yeah, Matthew. Thank you, man. Yeah. 181 00:15:51.170 --> 00:15:55.250 Thanks. I I jumped on early, I got the link, so I'm I'm happy to be here. 182 00:15:56.735 --> 00:16:05.135 You

might have gotten bad data this week. So we usually do start at one, but we started at 12:30 this week. So that's why we were, like, already rolling. Nothing you did. No worries. 183 00:16:05.615 --> 00:16:16.230 But, man, it's awesome to have you here. We'd love to hear a little bit about Timekeeper and what what you guys have going on. Yeah. Matthew Johnson here, and really excited to be enjoying with you guys today. 184 00:16:16.550 --> 00:16:19.270 About a year and a half ago, I got into this space 185 00:16:19.910 --> 00:16:25.589 basically as a passion for preserving history. And, man, it's been an awesome ride. 186 00:16:26.245 --> 00:16:29.205 Just learned a lot, loving the technology. 187 00:16:30.005 --> 00:16:34.805 So what we're what we're doing is working to use technology, 188 00:16:35.125 --> 00:16:36.085 AI, 189 00:16:36.725 --> 00:16:40.245 to help people preserve history, and that's that's the Timekeeper app. 190 00:16:41.230 --> 00:16:46.990 And then, you know, running into to Brian and a and a few of you guys over the the past year has been really 191 00:16:47.310 --> 00:16:48.590 amazing to learn 192 00:16:48.910 --> 00:16:52.270 what we can do with AI to curate data 193 00:16:52.670 --> 00:16:53.710 and 194 00:16:54.455 --> 00:16:55.335 also 195 00:16:55.335 --> 00:16:59.975 using the open source tools in the Morpheus network. That's been amazing to have 196 00:17:00.615 --> 00:17:06.055 compute at our fingertips, basically, whenever we need it and getting incredible 197 00:17:06.055 --> 00:17:09.415 results. So that... Yeah. That's that's what we're working on right now, just taking 198 00:17:10.000 --> 00:17:16.720 taking historical data and bringing it to life with the power of AI while preserving its original authenticity. 199 00:17:16.720 --> 00:17:17.760 So TimeKeeper, 200 00:17:17.760 --> 00:17:18.239 we... 201 00:17:19.039 --> 00:17:22.080 Any data that comes into the system is time stamped, 202 00:17:22.240 --> 00:17:22.400 and 203 00:17:23.145 --> 00:17:31.065 we know that that's the original source data. But then afterwards, you can do, you know, a lot of stuff with it. You can clean it up with the with the AI. You can 204 00:17:31.305 --> 00:17:37.945 share it with your friends and family. And we... You can even make little video clips out of your ancestors' photos and... 205 00:17:39.570 --> 00:17:48.370 Which is just incredible. So the goal the goal with TimeKeeper is to to bring history to life for people and to do it in a way that's safe where, one, their data stays 206 00:17:48.850 --> 00:17:55.654 theirs. You know, it doesn't belong to anyone else, And that's why we're using open source tools, using the Morpheus network for compute 207 00:17:55.815 --> 00:17:59.014 so that the data stays in the hands of the individual. 208 00:17:59.255 --> 00:18:02.615 And from the beginning, that's kinda been my premise to build this system, 209 00:18:02.935 --> 00:18:05.895 is we live in a we live in a age in a world where 210 00:18:06.340 --> 00:18:08.900 everything belongs to to the big corporations 211 00:18:08.900 --> 00:18:17.540 and everything's gone to a SaaS model. And my goal is to give people a tool to work with their history that kinda reverses that. I I know it's 212 00:18:17.940 --> 00:18:20.500 different from all the models out there where 213 00:18:21.235 --> 00:18:29.315 everything's going big and corporate, but to give people a tool they can use that's powered by distributed compute, that's open source, that's safe, 214 00:18:29.715 --> 00:18:35.740 and in the... At the end of the day, the data remains there. So that... That's what we're working on. Does answer your question there? Yeah. Absolutely. 215 00:18:36.300 --> 00:18:40.220 And and is there is there a link or anything I can show on my side, 216 00:18:41.420 --> 00:18:45.420 Matthew, that just kinda shows off, like, your guys' UX or anything like that? 217 00:18:46.385 --> 00:18:48.624 Yeah. Perfectionimage.com. 218 00:18:51.585 --> 00:18:54.304 And then there's actually a demo archive on there. 219 00:18:55.105 --> 00:18:59.664 Perfect. Hang on one second. Check out And by the way, to everyone watching, I... Right as... 220 00:19:00.720 --> 00:19:25.955 Sorry. Do you go by Matt or Matthew? Matthew. Matthew. Right as Matthew was jumping on, I received a text from one of our community members saying that Matthew was hopping on with us, but was jumping on early so we could, like, run through a test. He did not get the opportunity to do that because we had started early this week. So Okay. That is not on him. That is on us. Just wanna put that out there. Although, Matthew, if it makes you feel better, we... A couple weeks ago, we did this live stream as part of a conference in Provo, 221 00:19:26.550 --> 00:19:34.710 and we did do a full run through and test. And then on the time we were supposed to start, all of the tech broke. Literally, everything that was I 222 00:19:35.270 --> 00:19:48.214 conference. Yeah. Okay. Yeah. It was a nightmare. I wanted to die. And I had just told Alan, like, literally ten minutes prior that I hate... Like, I'm so nervous to do it with a live audience because that's when stuff goes wrong and... You know? Sure enough. 223 00:19:48.774 --> 00:19:52.134 But, anyway, here, let me let me share. Also, Christopher, 224 00:19:52.294 --> 00:20:00.220 look at this guy. While you're while you're bringing that up, Matthew, nice nice to meet you. Yeah. And welcome. I'm glad to meet some of the other... The the Morpheus 225 00:20:00.220 --> 00:20:01.019 fanatics. 226 00:20:01.100 --> 00:20:03.419 It's interesting, and and I hope this is not 227 00:20:04.380 --> 00:20:09.739 too out of out of whack, but what you described, especially with family history, the image that I got initially was 228 00:20:09.900 --> 00:20:11.100 the old family bibles. 229 00:20:11.785 --> 00:20:15.945 Right? Where people would pass this down, and it was theirs. It was a protected... 230

00:20:16.105 --> 00:20:20.424 It had history. It had notes. It had people. It had that lineage that was theirs. 231 00:20:20.825 --> 00:20:23.304 And I don't know if that's 232 00:20:23.705 --> 00:20:34.909 not the theme, but boy, did that pop right into my head when when you talked about it was this is the resurgent. It's ours. It's not a big corporation that doesn't live on any of the big guys. It's something that is... 233 00:20:36.750 --> 00:20:42.485 It can be passed down and not messed with. And I I love your... Love the idea. I love your idea. 234 00:20:43.205 --> 00:20:50.165 Yeah. We're actually... One of the things we really focus on preserving is is family bibles and family journals and family records and 235 00:20:50.325 --> 00:20:56.989 and keeping them preserved in their original state, but then using the Power BI to extract names and dates and, 236 00:20:57.870 --> 00:21:04.029 you know, find information about your family. Yeah. If you scroll down a little bit through the web page there, you can see some of the the UX 237 00:21:04.190 --> 00:21:04.909 screenshots. 238 00:21:09.414 --> 00:21:15.494 Oh, yeah. Sorry. You said hit explorer demo archive? Yeah. If you do the demo archive, then that actually... 239 00:21:16.934 --> 00:21:21.174 It it takes a minute because it's loading a... It's actually preparing a 240 00:21:21.760 --> 00:21:23.119 user experience 241 00:21:23.840 --> 00:21:26.159 that anyone can test out the 242 00:21:26.799 --> 00:21:30.240 the archive live. At least we hope they can. That's the goal. 243 00:21:31.760 --> 00:21:38.654 Nothing like a live nothing like a live demo. The live demo the live demo is what it... It goes. Right? It's... It can work out. 244 00:21:39.055 --> 00:21:49.295 The moment you stick it on a live feed is when you're gonna have an issue. By the way, this is why I'm I'm so glad we changed formats. So we used to do this call strictly with audio just through xSpaces, 245 00:21:49.295 --> 00:21:56.089 but a month or two ago, we switched to video. And I'm so glad because, yeah, now we can jump on, try to do stuff in real time. And, 246 00:21:56.490 --> 00:22:04.009 also, I just love seeing another face pop up. Like, Matthew jumps on, Christopher jumps on, and all of a sudden, it's like someone walks in the room. So it's great. I love it. 247 00:22:05.130 --> 00:22:06.409 While this loads 248 00:22:06.945 --> 00:22:13.104 Yeah. It's not loading on my end either, so maybe just go back to the main page and just Sure. Go through a couple of the features real quick. 249 00:22:14.865 --> 00:22:18.224 Yeah. If you just scroll down there on the... On that page. So the... 250 00:22:18.465 --> 00:22:20.144 You know, it's a full archive 251 00:22:20.770 --> 00:22:28.450 suite, has all the tools to curate your data. And then on top of that, we've built in all the AI tools that are running on the Morpheus compute 252 00:22:28.850 --> 00:22:30.370 to work with your data. 253 00:22:30.610 --> 00:22:35.410 So when you send a request out, it's staying on that distributed network. 254 00:22:35.794 --> 00:22:38.754 It's it's not going to, you know, one of the big providers. 255 00:22:41.075 --> 00:22:43.955 And then the the... A real exciting thing that... 256 00:22:44.914 --> 00:22:49.154 One of the features at this point... Yeah. So the restore photos. That... That's my grandpa right there. 257 00:22:51.480 --> 00:22:57.240 And you can see how we use the the AI to to bring to bring it back to life. 258 00:22:58.120 --> 00:22:58.679 Yeah. 259 00:23:00.600 --> 00:23:09.174 It's pretty powerful. Yeah. That's wild. And then, obviously, we can't demo it here on on the website, but the tools for data extraction and 260 00:23:09.895 --> 00:23:14.695 finding insights on your data is is just powerful. You you can find things that... 261 00:23:16.669 --> 00:23:32.315 You can sort through thousands of documents in in minutes and find connections and then share those with family members. And we now have a live feed in your archive when you first log in. It just shows you a live feed of, like, what happened on this day, 262 00:23:33.435 --> 00:23:34.955 what was going on with your... 263 00:23:35.435 --> 00:23:42.395 And your family history on that date, and and stuff like that just to kinda give you a good a good way to work with your data. So is it almost like... 264 00:23:42.635 --> 00:23:46.929 That part of it is almost like a a social media dedicated 265 00:23:46.929 --> 00:23:52.769 to, like, your family, basically? Like, within your family unit, you have your own record of just kinda everything. 266 00:23:53.010 --> 00:24:03.325 It's like can invite friends and family to join it. There's three different levels. You can give them full access. You can give them read only access. You can give them comment access. And so then you can collaborate with all your friends and family 267 00:24:03.565 --> 00:24:04.605 on the data 268 00:24:04.845 --> 00:24:06.205 within Timekeeper. 269 00:24:06.445 --> 00:24:20.210 This is like the natural progression of the family group chat. You know? Like, you have the family group chat. This is like endgame of that. I love it. That's the goal. And that... I think you actually said it so well there. That's kind of the end goal is to give families the ability to 270 00:24:20.690 --> 00:24:25.890 interact in a in a way that brings the past into the present and makes it fun 271 00:24:26.385 --> 00:24:27.825 and makes it 272 00:24:28.145 --> 00:24:32.625 visual because we're all visual these days. Everything's gotta be visual and quick and nice. And 273 00:24:33.185 --> 00:24:37.985 my my goal was to bring the next generation into this love of history. 274 00:24:38.510 --> 00:24:42.110 And I think the best way to do that is visually. So it was very photo centric. 275 00:24:42.830 --> 00:24:48.910 You know, we started the app working with photos, and we've moved into documents and audio and video. But photos is kind of the the 276

00:24:49.550 --> 00:24:52.190 goal... The first thing that started out was bringing photos back to life. 277 00:24:53.435 --> 00:24:58.394 I lost my train of thought there for a second, but I was No worries. I was gonna say... I just... Yeah. Oh, yeah. Bringing families together. 278 00:24:59.515 --> 00:25:02.315 What... Know, when when it's in a... An app on your phone, 279 00:25:02.795 --> 00:25:06.875 then everyone can share and connect. So if it's sitting in a box in the attic or in the basement, 280 00:25:07.380 --> 00:25:09.940 then it... It's it's lost everybody. 281 00:25:09.940 --> 00:25:24.135 Yeah. So the the goal of Timekeeper is to bring those boxes to life, bring them into the age where it's on your phone, and yet keep it yours because I think that's really critical. There's a lot of other people out there that, you know, you scan your data, and then the... It belongs to whoever... 282 00:25:24.535 --> 00:25:27.015 What... Whoever's app you're using. It's their data, 283 00:25:27.975 --> 00:25:33.815 and it's monetized and advertised to. And our whole goal is to flip that around and keep the data private. 284 00:25:34.309 --> 00:25:39.590 It belongs to you, and AI is just a tool that you use to help bring it to life and curate it. 285 00:25:40.070 --> 00:25:45.749 Yeah. No. I mean, I think that's fantastic. And and you're exactly right. I mean, who wants, like, 286 00:25:46.309 --> 00:25:47.910 x y z megacorp 287 00:25:48.105 --> 00:25:54.424 profiting off of your dead grandparents or not... You know what I mean? Like, it's just one of those things that's, like, that's just crazy that that's even 288 00:25:54.905 --> 00:26:08.470 the reality of what we're facing right now, but it is. And not and not even that. Like, to your point, the intimate documents. Like, you might have a journal from, you know, somebody a 100 years ago or, like, something like that that's very intimate to the family. 289 00:26:08.870 --> 00:26:12.309 Have you guys talked to or done any exploration with, like 290 00:26:13.350 --> 00:26:22.385 like, museums or, like, other archives or, like, libraries or things like this where they might have, like, a private collection? So we have over 1,500,000 291 00:26:22.385 --> 00:26:26.065 scans that we've done with local museums, 292 00:26:26.065 --> 00:26:27.664 local county libraries. 293 00:26:28.465 --> 00:26:32.544 Our largest collection is over 200,000 scans with one museum. 294 00:26:32.544 --> 00:26:41.129 Wow. And then we've curated the data sets for them. I've been doing that kind of just as a a pilot, you know, test program to really beta test the system, 295 00:26:41.370 --> 00:26:42.889 you know, fill it with information, 296 00:26:43.049 --> 00:26:48.650 and it's been very effective. You know? I I I sorta... I talk about the power of using AI to to work with your data. 297 00:26:49.825 --> 00:26:51.424 You know, we had 87,000 298 00:26:51.424 --> 00:26:54.144 files come in from a Iowa library, 299 00:26:54.304 --> 00:26:56.784 and they were just in the most random 300 00:26:57.345 --> 00:26:58.065 order. 301 00:26:58.784 --> 00:27:06.730 There was no order... No structure to it. And so I just used the built in AI within Timekeeper and told it, hey. I wanna organize these into yearbooks, 302 00:27:07.210 --> 00:27:22.264 family histories, local histories. Basically, I gave it the criteria I wanted, and it took forty five minutes to go through, you know, 87,000 some files. And it came back with a fully curated tree. And I wish I could share that with you if I could share my screen, but I guess... Yeah. Hopefully, we can do another 303 00:27:22.505 --> 00:27:37.620 demo in the future that we're more prepared when I'm more prepared for it. But No. No worries. Yeah. We can definitely have you back on. We do these every Thursday. So Nice. But, yeah, and that was all powered on on open source compute. And I think that... That's the goal that I wanna share with the group today is that if you have 87,000 304 00:27:37.620 --> 00:27:46.474 files and they're your documents, you don't wanna send those to any of the the large LLM. You don't wanna send them... I won't name all the names. We all know who they are. But, you know, you send all that out there. 305 00:27:47.034 --> 00:27:49.674 It it... It's used as training data. It's used 306 00:27:50.715 --> 00:27:51.274 in... 307 00:27:51.595 --> 00:27:53.354 For advertising purposes. 308 00:27:53.515 --> 00:28:07.240 But if you're using open source distributed compute, then it's it's yours. It's safe. You don't have to worry about it getting out there and being used. So that's... That was kind of the the motivation for us to go open source with the compute was because the data we're dealing with is 309 00:28:07.640 --> 00:28:12.520 personal and often private. Like you said, a personal journal. We've done a lot with journals. You know, we could OCR 310 00:28:12.975 --> 00:28:16.495 the original cursive handwriting and have had phenomenal 311 00:28:16.495 --> 00:28:20.575 results with that. Mhmm. A lot of... Even my kids complain they can't read cursive. 312 00:28:20.735 --> 00:28:31.950 And so I can take my grandpa's journals, and I can scan them in here to Timekeeper. I can turn OCR them into text, and then they can read them and enjoy them. So that... That's the tools we're giving people to help them 313 00:28:32.350 --> 00:28:34.830 interact with and and connect with their history. 314 00:28:34.990 --> 00:28:36.349 This makes me think of... 315 00:28:36.990 --> 00:28:43.149 So a concept that we've kinda had around the Morpheus community for a couple years now is this idea of nonfungible 316 00:28:43.149 --> 00:28:43.549 agents. 317 00:28:43.975 --> 00:28:52.374 So, like, I... I'm I'm assuming, Matthew, you're familiar with NFTs, nonfungible tokens. Right? Uh-huh. So so it's a similar concept, but basically this idea that you can have, 318 00:28:52.695 --> 00:28:59.174 like, an agentic AI or product

that has a limited number of copies that you can verify is the real original. 319 00:28:59.620 --> 00:29:02.340 And I could see a world where, like, 320 00:29:02.740 --> 00:29:12.820 it... I don't know. This is may... This is an extreme, so this is definitely an edge case. But let's say, for example, you had a family member who was, like, an author or an artist or had some type of, like, 321 00:29:13.705 --> 00:29:21.305 life and history and experience that other people would wanna interact with, you could make a nonfungible agent with their 322 00:29:21.705 --> 00:29:42.110 private works in this way, and that would be, like, a perpetual royalty resource for your family. Like, in other words, like, the value and, like, the intelligence that you create while you're alive can be something that your family benefits from long after you're gone in a way that's not possible today because there is no way to, like, 323 00:29:43.475 --> 00:29:47.555 well, one, like, organize that information, keep it private, 324 00:29:47.555 --> 00:29:51.315 verify that it's from the actual source that you're saying it's from. 325 00:29:52.195 --> 00:29:55.235 And and, like like, I perceive a time where 326 00:29:57.260 --> 00:29:58.140 someday, 327 00:29:58.780 --> 00:30:02.220 who's like a, you know, a great thinker of the day, 328 00:30:04.060 --> 00:30:10.125 you know, whatever, just use Elon as an example. I perceive a time in the future where there is an Elon nonfungible 329 00:30:10.125 --> 00:30:20.124 agent that's owned by the Musk family or whoever, and I can have a Zoom call with literally all of the intelligence that was recorded, private and public, of Elon. 330 00:30:20.284 --> 00:30:28.810 Right? But, like, you could only do that if you had access to all of that information and data and if you, like, as the family, can keep it private and and 331 00:30:28.970 --> 00:30:30.730 retain ownership of that, 332 00:30:31.050 --> 00:30:33.210 which something like what you guys are doing 333 00:30:33.450 --> 00:30:37.290 is like a a baby step in the in the direction of that being possible, 334 00:30:38.065 --> 00:30:40.785 which is which is cool. Not baby steps maybe underselling. 335 00:30:40.785 --> 00:30:44.945 It is a step, you know, in that direction of, like, that real real life. 336 00:30:45.745 --> 00:30:50.945 But you you read my brain, brother, because that's actually the direction we're headed. And what got me started in this is 337 00:30:51.730 --> 00:30:55.249 allowing individuals and families to preserve their data 338 00:30:55.490 --> 00:31:01.009 for generations to come because we... We're moving into a world where data is the only thing that's gonna matter pretty soon. 339 00:31:01.890 --> 00:31:03.010 Data is everything. 340 00:31:03.625 --> 00:31:08.585 And so we want... We built Timekeeper to give people a place to store their data. 341 00:31:08.904 --> 00:31:11.224 So we're there, and like you said, people come to you 342 00:31:11.705 --> 00:31:14.424 for access to it rather than the other way around. 343 00:31:14.985 --> 00:31:18.985 With all the social media platforms to date, the data belong to the corporation. 344 00:31:19.840 --> 00:31:23.519 And I I kinda like to compare it to your belongings in your home. 345 00:31:23.840 --> 00:31:26.959 You know, when when you move, you take all your belongings with you. 346 00:31:27.440 --> 00:31:38.495 And right now, data isn't managed that way. When you move and leave a platform, they keep your data. Mhmm. And going forward, Timekeeper, the the premise will be that, yeah, you have this set of data, 347 00:31:38.815 --> 00:31:39.855 and it's yours, 348 00:31:40.255 --> 00:31:41.855 and it could be authenticated. 349 00:31:41.855 --> 00:31:47.535 Even a hundred and fifty years from now, it could be authenticated. This is original source information. This is original data 350 00:31:48.030 --> 00:31:56.030 and can be used by the by the family. If it... If it's a community library, the the same concept that same concept that the data... 351 00:31:56.270 --> 00:31:58.910 Authenticated data can be preserved for generations 352 00:31:58.990 --> 00:32:00.830 and actually be an asset 353 00:32:00.910 --> 00:32:01.550 rather than... 354 00:32:02.255 --> 00:32:08.655 It's, yeah, it's it's a real digital asset. Yeah. Yeah. Sure. Totally. So so an idea... Think... Oh, I 355 00:32:10.175 --> 00:32:16.895 was gonna say the other part of this is, like, it it... There's an asset on on multiple fronts here. Right? One is the one that conveys 356 00:32:17.160 --> 00:32:28.280 that you're able to, you know, you're able to sell and someone can, you know, someone can tap in, you're able to monetize it. But I think, you know, there's a real asset in having those memories because I I think this is probably the first time... 357 00:32:28.600 --> 00:32:32.280 You know, we're we're hitting the first generation of people who have, 358 00:32:33.425 --> 00:32:42.784 you know, the the... That have pictures, vast amount of pictures. I know when I grew up, you know, it's like... I think if you look... My guess is, Matthew, is you, like, go back through families. 359 00:32:42.945 --> 00:32:53.230 Some families will have a handful of photos. Some will have... You know, they'll they'll have, you know, 20 or 30 photos, but it's like this will be the first generation where there's the the... You... You're able to have, 360 00:32:53.870 --> 00:33:07.744 you know, a lot of images, a lot of photos of what happened and what families went through. And then it's like keeping you... Keeping your legacy. Right? Is is Yeah. I think that's, like, knowing... People knowing where they come from. And we look at we look at the culture today, 361 00:33:08.145 --> 00:33:08.705 and 362 00:33:09.345 --> 00:33:17.089 it gets, like, it's like fast food. So many so many people really don't know where they came from, their family background is, 363 00:33:17.250 --> 00:33:22.849 or or their...

Like, their community is kind of the same thing. And that's that's... It's somewhat of a Chesterton's fence 364 00:33:23.169 --> 00:33:29.645 from a from a culture perspective of knowing where you came from, what your... You know, what what the value sets were. 365 00:33:30.605 --> 00:33:34.765 And so, you know, you're you're getting an... You're building an asset on both sides on it. 366 00:33:35.405 --> 00:33:42.365 But I had I had one question, like, looking at this. Where did you... Where have you seen... You... Obviously, you have the open source tools kinda powering 367 00:33:42.510 --> 00:33:44.909 on the background as people are are using it. 368 00:33:45.230 --> 00:33:51.470 Where did you see the biggest jump where you were using open source models and you looked and you said, 369 00:33:52.110 --> 00:33:58.750 wow. Okay. This is this is really this is really hitting. Right? You know, this is this 370 00:33:57.955 --> 00:34:01.715 is this is the intelligence level. Right? The the the 371 00:34:01.875 --> 00:34:05.394 the experience with open source where you just were, like, kinda blown away by it. 372 00:34:06.674 --> 00:34:10.515 Yeah. I mean, from the beginning, I wanted to work with open source, so I built myself 373 00:34:11.070 --> 00:34:15.790 a a beast of a machine. I mean, I put $50 into my computer that's sitting right here in my office, 374 00:34:16.510 --> 00:34:20.990 and I was just blown away with what I could do with data. I'm complete open source models. 375 00:34:21.550 --> 00:34:26.270 And I was like, I need to share this with the world. And that was kind of the catalyst to start building the Timekeeper app 376 00:34:26.655 --> 00:34:28.895 that anyone should be able to do this 377 00:34:29.215 --> 00:34:32.015 with the power... Because AI has opened up doors 378 00:34:32.575 --> 00:34:41.855 that weren't there before when it comes to working with data and processing it and finding connections. So I'm so excited about that. I think from the beget... That's what happened for me. It's just on my own machine, 379 00:34:42.480 --> 00:34:46.160 you know, building something that could run the models and then saying, wow. 380 00:34:46.880 --> 00:35:01.795 The the whole world needs access to this. And so that's what... We started building Timekeeper out with that same premise that not everyone can afford to build a a server in their in their home. Right? But they should have access to this technology. And that's where Morpheus came in. It was like, okay. That provides the tool 381 00:35:02.275 --> 00:35:05.395 that the same concept is running it in your home, 382 00:35:06.675 --> 00:35:10.515 but you have access to that and and incredibly powerful compute 383 00:35:10.860 --> 00:35:12.220 to work with your data. 384 00:35:14.380 --> 00:35:18.540 Was there was there a specific was there a specific... I appreciate the the Morpheus 385 00:35:18.540 --> 00:35:37.695 the the Morpheus plug there. But was there a specific model that that that you you saw and you experienced that you're like... That, you know, this is it? Yeah. The the Quentin models. Model that was? Yeah. The the Quentin models I've been Okay. Very, very pleased with. And since we're working a lot with photos, you know, that... That's a very good multimodel. 386 00:35:37.695 --> 00:35:44.410 And I'd say most of what I run on my home machine are all some version of Quinn. 387 00:35:45.770 --> 00:35:51.130 That's either the original open weight models or community modified models for specifically 388 00:35:51.130 --> 00:35:52.490 for photo editing. 389 00:35:53.290 --> 00:35:55.130 There are some out there. That's what I 390 00:35:56.525 --> 00:35:57.805 that's what I run on my 391 00:36:03.725 --> 00:36:07.965 you spell that, Aaron? Very cool. Is there... And then is is there a is there a tool that... 392 00:36:08.685 --> 00:36:14.610 Looking at from an AI Workbench perspective. So you've got the model that's kinda powering it on the backside. 393 00:36:14.930 --> 00:36:21.570 Why... Is there a is there a tooling set that you work with that you've really you've really found that's 394 00:36:21.730 --> 00:36:24.690 that's been a a... An unlock for you as far as 395 00:36:25.455 --> 00:36:28.575 from execution... Your be... Ability to execute and leverage? 396 00:36:29.535 --> 00:36:31.135 Using AI models? 397 00:36:32.335 --> 00:36:41.440 No. Yeah. No. Like, on the workbench side, are there any is there any tooling? Is there, you know, the brain structure? Are there, you know, the... Your your CLI? 398 00:36:41.440 --> 00:36:48.960 I'm getting you behind the nitty gritty here a little bit. But things as as a builder that you found that are underappreciated 399 00:36:49.280 --> 00:36:54.720 that a lot of... You think a lot of builders would benefit from if they were to integrate into their workflows. 400 00:36:55.415 --> 00:36:59.255 Well, I'll just be up front and say that this is a totally new field to me, and I have 401 00:36:59.575 --> 00:37:02.215 built all of this not even knowing what I'm doing. 402 00:37:02.615 --> 00:37:03.255 So... 403 00:37:04.535 --> 00:37:08.375 Yeah. No. I'm serious. I've just scraped together Get in and ship it. Yeah. 404 00:37:09.900 --> 00:37:16.940 I've... My whole career has been in construction, just so you guys know. And a little over a year ago, I'm like, I'm gonna build an app. And, 405 00:37:17.500 --> 00:37:19.340 yeah, if it weren't for AI, 406 00:37:19.660 --> 00:37:40.330 it wouldn't... If we... We wouldn't we wouldn't be here where we are. So I've been... I've used a lot of different tools. I... I've got on... Like I said, I've I've I've found GitHub, found all the open source tools, downloaded them, started with that, and then realized really quick I could just modify that and build it for our own purposes and integrate it. So I'm still learning on that front, all the stuff that's available. 407 00:37:40.970 --> 00:37:42.410 It's a work in progress. 408

00:37:45.050 --> 00:37:46.730 That's that's 409 00:37:46.730 --> 00:37:47.050 great. 410 00:37:49.065 --> 00:37:57.625 What do you think... What do you... Based on based on that, what are what are missing... What are the Lego pieces that you wish kind of existed right now that would help you... 411 00:37:58.345 --> 00:38:02.025 That would be... That would help you get get more done, you think? 412 00:38:05.369 --> 00:38:06.650 I... Yeah. More 413 00:38:07.530 --> 00:38:12.490 access to open source compute for coding. So right now, I do most of my coding with codex 414 00:38:12.490 --> 00:38:13.930 just because it's so easy. 415 00:38:15.865 --> 00:38:18.105 I would love to build a plug in to an 416 00:38:18.585 --> 00:38:27.704 open source compute for that, you know, like, run GLM 5.1 or something easier, which kinda like you demonstrate that one button download 417 00:38:27.704 --> 00:38:30.750 for an agent, right, for an OpenCLaw type agent. 418 00:38:31.550 --> 00:38:38.030 For the community in general, I think it'd be really helpful if we were to develop a better access to compute for coding 419 00:38:38.430 --> 00:38:49.265 on on a distributed type network. The the same thing where where a layperson like me can come in and in a few minutes start coding using, you know, using those open weight models 420 00:38:49.505 --> 00:38:52.945 and not have to spend hours or even months 421 00:38:53.105 --> 00:38:54.625 on the education 422 00:38:54.625 --> 00:38:55.585 curve behind it. 423 00:38:56.619 --> 00:38:59.340 Can Perfect. That's that's a that is a... 424 00:38:59.580 --> 00:39:03.260 Well, no. Not an Odeo, but we do have that built in, 425 00:39:04.060 --> 00:39:09.740 and I can get you the information. But instead of using something... Yeah. Instead of using something like 426 00:39:11.185 --> 00:39:13.505 cursor, your your IDE, 427 00:39:14.545 --> 00:39:15.825 you switch to ZEV, 428 00:39:16.145 --> 00:39:21.585 and you can plug in our Morpheus API gateway Okay. Just like you would any other API, 429 00:39:21.744 --> 00:39:26.700 and you're into the Morpheus network. So there are frameworks. And if you go to node docs, 430 00:39:27.340 --> 00:39:33.660 there's is... There's documentation on that. So that was one of the first things because that was my thing is I wanna be able to, you know, get out of 431 00:39:34.060 --> 00:39:34.620 the 432 00:39:35.580 --> 00:39:42.685 the cursor products and and codecs and all that and do daily driver coding through there. And and so Zev plus 433 00:39:43.805 --> 00:39:46.685 the API gateway and Morpheus 434 00:39:46.925 --> 00:39:52.925 works a treat. And you can do mixture of models too. Right? I can do Quen for a little bit or GLM five three. 435 00:39:53.680 --> 00:40:12.075 So switch back and forth. So that's that's awesome to hear you say that, and I'd love to be able to highlight that in the future too so... If other folks need to do that. No. That's good to know. I'm I'm excited to set up on that. That's the direction I wanna go with the coding end. Mean, thankfully, on the the tooling end of Timekeeper, 436 00:40:12.075 --> 00:40:15.195 we're able to run all of that, you know, easily on the open source 437 00:40:15.515 --> 00:40:17.275 models, which has been really fun. 438 00:40:18.155 --> 00:40:22.475 But the the coding side of it, I've still been using, yeah, the big commercial 439 00:40:23.390 --> 00:40:24.510 models for. 440 00:40:25.070 --> 00:40:33.550 And just because my machine, even as big and powerful as it is, I have not been able to load any of the GLM models on it. They just need too much VRAM. 441 00:40:33.550 --> 00:40:36.750 I mean, I've got a 196 gigabytes of VRAM, but that's not enough. 442 00:40:37.375 --> 00:40:37.695 So 443 00:40:38.255 --> 00:40:43.295 One of one of the issues that I do have, and I I will say that I'm not pure into 444 00:40:43.535 --> 00:40:48.815 full open source coding as much as I'd like. I really wanna get there. One of the... And one of the issues is 445 00:40:49.135 --> 00:40:56.650 I take screenshots a lot, and I add that as context. Like, I'm sure if you do on the website, like, hey. This doesn't look right. This doesn't... So that visual 446 00:40:56.970 --> 00:40:59.370 Uh-huh. Some of the models that are out there 447 00:41:00.250 --> 00:41:03.850 in the open source may not know how to digest images. 448 00:41:04.835 --> 00:41:08.835 And so you send that same rich context, that same multimodality, 449 00:41:09.234 --> 00:41:12.595 you may not get the same response. So we're... I'm... 450 00:41:13.154 --> 00:41:25.910 That's my last kinda Rubicon before I'm all in and and, you know, kill a bunch of subscriptions, but I'm so close. But that's that's the only... That's the only... I don't wanna Yeah. I don't I don't wanna say that's the... It's the be all end all, 451 00:41:26.470 --> 00:41:32.070 but it's close. It's so much closer for for being able to have the, you know, the the the development 452 00:41:32.070 --> 00:41:37.425 and the image and the... All that stuff is... We're real close. Well, and it's an inevitability. 453 00:41:37.425 --> 00:41:50.320 I mean, we saw this with every other step of the large language models. Like, if... And and I I somewhat agree, Alan, that, like, anything strictly chat based, I've been able to move to open source. No problem. When it starts to become multimodal, 454 00:41:50.320 --> 00:41:56.400 that's when it gets a little more complicated. So for example, when I'm walking my dog, I often talk to Grok. 455 00:41:56.480 --> 00:42:02.000 Like like, physically talk to, like, through my headphones. I'm sitting there having a conversation with it, just going through whatever. 456 00:42:02.484 --> 00:42:11.045 And there isn't really a good open source alternative for that yet, but it's it's just a matter of time. And then there is, and then, you know, we can 457 00:42:11.365 --> 00:42:12.485 keep keep 458 00:42:12.805 --> 00:42:22.780 moving on it.

But, yeah, that is that is awesome feedback, though, Matthew. And I love... Listen. I love that your background is not in software development. Mine certainly is not. And our... I want... 459 00:42:23.100 --> 00:42:42.435 Well, you know, I would love to have a community also full of, like, super seasoned builder devs, but I personally really enjoy the monkey with the AK-forty seven meme and just giving the ability to build crazy stuff to people who couldn't have done it previously. But now it's like it's like a superpower. Like, go for it, You know? I love that. 460 00:42:42.755 --> 00:42:44.515 Well... And the window between 461 00:42:44.595 --> 00:42:46.035 ideation and actually 462 00:42:46.510 --> 00:42:51.310 implementing it has shrunk so small. It's Yeah. It's microscopic almost. 463 00:42:51.550 --> 00:43:04.945 And and, yeah, I don't have a coding background. And yet, I've had experienced people in this field tell me that it's some of the best software they've used for working with historical photos and documents. They're like, wow. This is really cool. You know, I've had some beta users from museums, 464 00:43:05.185 --> 00:43:08.145 even some very large family history institutions, 465 00:43:08.145 --> 00:43:16.545 you know, test it out and be extremely impressed with it. So I'm really happy with where it's going and the fact that we based it on, you know, open source technology. 466 00:43:17.430 --> 00:43:42.515 I... You know, I'm really excited about that direction, and we're hoping to actually launch it to... I I I mean, anyone could sign up now. It's still... I consider it the beta phase, but going into the first quarter of next year, of as my Christmas gift to the world, I want to start, yeah, just open this up to the whole world and say, hey. Come use this tool. Use this for your... Preserve your history and start creating an asset that is... As the world changes and data becomes more and more valuable, 467 00:43:43.315 --> 00:43:50.690 start creating your family heirloom, your family asset. It needs to be created, and that's the tool we wanna give people to use to create a secure, 468 00:43:50.849 --> 00:43:59.090 safe archive for their family for generations to come and a community or, as anyone, any group of people could use it for that same end. 469 00:43:59.810 --> 00:44:04.555 Yeah. Oh, absolutely. I think that's fantastic. That's great that you guys are doing that. You are... Matthew, 470 00:44:05.275 --> 00:44:08.395 you're you're a a testament, as Bowen said, to the... 471 00:44:08.555 --> 00:44:10.715 Any monkey with an AK 47. 472 00:44:11.035 --> 00:44:12.234 But but 473 00:44:12.700 --> 00:44:15.260 you know the problem intimately, 474 00:44:15.500 --> 00:44:22.140 and the monkey knows how to ask the question well. I think that's that's really the that's really the edge right now. 475 00:44:22.380 --> 00:44:25.724 With... If AI takes away the requirements 476 00:44:25.724 --> 00:44:27.244 for coding knowledge 477 00:44:27.325 --> 00:44:31.325 is do you actually understand the problem set that's being tackled, 478 00:44:31.644 --> 00:44:36.444 you know, at a at a first principles level? And then you know how to ask questions 479 00:44:36.444 --> 00:44:44.380 to kinda get to the other side to, like, layer on. So I think that's that's that's really it is is anybody that builds something where 480 00:44:44.700 --> 00:44:55.465 someone comes in and is like, wow. This really fits all of my needs. It's someone who spent a lot of time thinking through what what are all the problems and and how to resolve that. So that's that... That's impressive. 481 00:44:55.625 --> 00:44:58.345 Love love what you're doing here. Is there anything that you... 482 00:44:59.065 --> 00:45:03.945 Anything that you're shipping right right now that you're not allowed to talk about that you have to... That you 483 00:45:04.345 --> 00:45:05.865 can can somewhat leak here? 484 00:45:07.280 --> 00:45:13.360 So on my machine, yeah, the capacity of Timekeeper is way beyond what's running in the live 485 00:45:14.160 --> 00:45:15.440 app online. 486 00:45:16.240 --> 00:45:18.240 And what I'm building there 487 00:45:18.960 --> 00:45:23.805 is a way to take... To look at someone's life and take all of their data and create 488 00:45:24.045 --> 00:45:30.205 a graph of their life. And it is really, really cool. I don't know how many of you are familiar with, like, 489 00:45:30.605 --> 00:45:31.565 mind mapping 490 00:45:32.020 --> 00:45:33.220 software. Yes. 491 00:45:33.620 --> 00:45:41.060 I love mind mapping software. I have a lot of irons in the fire as an individual. I run multiple businesses, so I've used that for years. So 492 00:45:41.300 --> 00:45:44.180 what what I'm doing with Timekeeper is building out 493 00:45:44.500 --> 00:45:45.460 a feature 494 00:45:45.815 --> 00:45:49.255 that uses an agent to look at all your data 495 00:45:49.655 --> 00:45:51.975 and also look at everything coming in 496 00:45:52.215 --> 00:45:55.415 and build out a mind map of that person's existence. 497 00:45:56.055 --> 00:46:10.980 And without any prompting... I mean, it's trained. And so without any prompting along the way, it'll build out a branch for your child, a branch for every relationship, a branch for every business, and basically build your life model based on the data that it has. And it is it is cool. 498 00:46:11.299 --> 00:46:18.025 I'm really excited about it. It's got a ways to go, and I'm still working on it. But the end goal will be to use Timekeeper 499 00:46:18.025 --> 00:46:20.185 not only as a historical tool, 500 00:46:20.425 --> 00:46:23.545 but as an arc of repository for your living... 501 00:46:24.025 --> 00:46:27.065 Your... A history of your life, your living legacy. 502 00:46:27.385 --> 00:46:35.950 And so everything will go into it, text messages, emails. I'm basing on everything. All my photos go into it. Everything I generate on a daily basis goes into Timekeeper, 503

00:46:36.350 --> 00:46:37.630 and then the agent 504 00:46:37.790 --> 00:46:44.350 models it onto my graph. And it... It's incredible. I'm really excited about that. I didn't expect to ask that question. So 505 00:46:44.825 --> 00:46:51.224 that's where it's going kinda long term there. And I I have a ways before we release that, but that'll be the 506 00:46:51.464 --> 00:46:54.265 kinda living version of the archive 507 00:46:54.984 --> 00:47:00.580 where it just taps into everything you generate on a daily basis and turns that into your... Part of your living... 508 00:47:01.540 --> 00:47:22.135 I I call it a living... Your living history, so to so to speak. Imagine when I I think that's awesome. I can take, like, all my records from age, you know, whatever, 15 and earlier and then create an avatar of myself that I can talk to myself in the past, and that avatar will only have the information available to them that they would have had on, like, whatever date I give them. 509 00:47:22.455 --> 00:47:28.360 Be fun fun experiments to be had. I don't an avatar I think that's on my machine, yeah, for that very purpose. 510 00:47:29.560 --> 00:47:36.840 And I've got a ways to go. But, yeah, it it uses only the dataset to communicate. So anything it answers, it has to answer from the dataset. 511 00:47:38.145 --> 00:47:52.705 It's pretty... Yeah. It... That's Yeah. Like, I'd love to be like, here's a snapshot of the Internet when I was, you know, whatever, a teenager, and then this is all the information that I would have had and then, like, talk to yourself. And you can... Like, that... That's that'd be crazy. That'd be a really wild experiment. 512 00:47:54.270 --> 00:47:57.950 That's that's awesome. It's funny, Matthew, you talk about that. One of the... 513 00:47:58.670 --> 00:48:03.710 I'm I'm I'm in. I'm ready I'm ready for that one, both both the the current one and the future. 514 00:48:04.030 --> 00:48:19.745 At one of my milestone birthdays, I sat down... My wife and I sat down and kinda mapped out, you know, all the decisions along the way and kind of what you were describing. You know, how did we get here? And it was interesting to see all those decisions as one of those, you know, butterfly effect. It's like, I had chosen differently, 515 00:48:19.825 --> 00:48:24.900 would you have ended up in the same place? And so when you map those decisions and that timeline out, 516 00:48:25.460 --> 00:48:35.700 it's really interesting, not just for you, but for other people that come after you to tell that story. Tell me about grandpa. Tell me about your life. Tell me more about where were you when. 517 00:48:36.695 --> 00:48:45.655 We've we've lost some of the storytelling aspect in our culture with this, you know, fifteen second brain span Uh-huh. Attention span. Mhmm. I think that is 518 00:48:46.535 --> 00:48:50.520 absolutely wonderful. I'm so excited. So excited. Yeah. 519 00:48:50.680 --> 00:48:55.880 Well, and and I think this would be really appealing to your audience. I feel like times are changing, 520 00:48:56.440 --> 00:49:08.265 and we can either embrace them or you can get crushed by them. Maybe that's a bad metaphor. But I feel like rather than being afraid of AI, we need to embrace it and use it to empower the individual. 521 00:49:08.984 --> 00:49:17.680 Because those that have the ability to use AI with their data and on a on a private level are gonna be empowered in the future of our society. 522 00:49:18.079 --> 00:49:23.600 And those that don't learn how to do that and don't get in on that are are gonna be at a disadvantage just because 523 00:49:23.920 --> 00:49:31.895 of the the dynamics that are gonna come into play. Those who use AI will be more effective and more efficient in their day to day life, in their 524 00:49:32.135 --> 00:49:33.975 goals. And so knowing... 525 00:49:34.295 --> 00:49:47.140 Having your own timekeeper vault that helps show you show you where you've been, where you are today, where you're going will be a huge advantage. And so that... That's kinda... That's what I want to build for people is give them access to these tools 526 00:49:47.380 --> 00:49:54.340 and keep their data private. You know, that's a big part of this for me. Keep the data yours. I've been journaling since I was 11 years old. 527 00:49:54.819 --> 00:49:57.619 And, yeah, I want those journals to be freely accessible 528 00:49:58.235 --> 00:50:08.395 in my digital vault and nobody else have access to them and still be using the power of AI to interact with them. So that... That's one of the driving passions for me. This is a passion project for me 529 00:50:09.115 --> 00:50:19.360 of bringing history into the into the... Into the day in a way you can interact with it. I love your avatar idea. I've been messing with that. I think the day will come where we all have a personal avatar, 530 00:50:19.760 --> 00:50:22.880 and and we interact with our data using that. 531 00:50:23.200 --> 00:50:30.785 And then our kids will interact with it. It's gonna the... It's gonna change... The future's got a lot of exciting stuff in it. So that's my philosophy. Embrace it, 532 00:50:31.185 --> 00:50:33.585 but let's find a way to make the average 533 00:50:33.585 --> 00:50:35.745 individual or every individual... 534 00:50:35.745 --> 00:50:38.865 Forget the word average. Every individual have access to this technology 535 00:50:39.650 --> 00:50:43.810 at their fingertips and not have to go through a corporation to get it. 536 00:50:45.090 --> 00:50:45.730 Yeah. 537 00:50:46.050 --> 00:50:50.290 Couldn't agree with you. You you hit the nail on the head regarding 538 00:50:50.290 --> 00:50:51.010 the, 539 00:50:51.490 --> 00:50:55.995 you know, the changing times right now and the solution for... 540 00:50:56.795 --> 00:51:01.755 You know, if the you... If the youth of today do not have access to opportunity, 541 00:51:02.395 --> 00:51:05.595 we're in for a world of pain. And so it requires 542 00:51:05.595 --> 00:51:08.235 people to be able to build build 543 00:51:08.420 --> 00:51:22.180 a bit like... This is gonna sound a little nuts, but a bit like Prometheus bringing light down from the light light down from the heavens. Right? So how can we do that? And then your your practice of the journaling, 544 00:51:22.895 --> 00:51:30.655 if people can look at this and have tools that's, like, common frameworks for success, and then they can they can actually look at kind of what's happening 545 00:51:30.895 --> 00:51:37.775 and and get that nudging, you give the opportunity to someone that wants to improve, that wants to lean forward, 546 00:51:38.460 --> 00:51:43.340 and and and get that opportunity, the the access to do so. So, you know, 547 00:51:43.980 --> 00:51:45.020 yeah, I do... 548 00:51:45.500 --> 00:51:50.859 Inspired. Appreciate the appreciate the the work that you're doing here, Matthew. Yeah. Thanks for having me, guys. 549 00:51:51.835 --> 00:51:55.435 I I appreciate the time and look forward to, yeah, spending more time and 550 00:51:56.395 --> 00:52:05.755 diving into... Also diving into the tools that you are making available to the world. So thanks for the great work. Yeah. Well, the community appreciates it, and it is a massive, 551 00:52:05.480 --> 00:52:11.000 massive effort. I I don't know how many individual contributors we're up to on the GitHub now, but it's, like, hundreds 552 00:52:11.560 --> 00:52:17.240 It's a bunch. From from day one. Yeah. So... No. Absolutely. I think, guys, that's a that's a good place to 553 00:52:17.535 --> 00:52:29.135 to wrap it this week, I think. Matthew, stay on for a second. The the software we use for these calls, once we end, you... There'll be, like, a upload percentage, and you just have to hit a 100 before you hop off if that's okay. 554 00:52:30.030 --> 00:52:30.750 Okay. 555 00:52:30.830 --> 00:52:31.630 Otherwise, 556 00:52:31.630 --> 00:52:40.670 everyone joining us live, thank you so much. As always, if you're watching the recording, thank you as well. And we will see you same time, same place, next Thursday. 557 00:52:40.990 --> 00:52:41.950 K. Thanks, guys.