1 00:00:00,020 --> 00:00:03,060 Josh: You probably have six or seven apps on your phone that each do one thing. 2 00:00:03,200 --> 00:00:05,840 Josh: One controls the lights, a thermostat, your security cameras. 3 00:00:06,020 --> 00:00:08,980 Josh: Each has its own login, its own interface, and its own notifications. 4 00:00:09,180 --> 00:00:13,200 Josh: It's horribly annoying. Now, imagine replacing that all with a single WhatsApp 5 00:00:13,200 --> 00:00:15,520 Josh: thread or a single text message through your phone. 6 00:00:15,840 --> 00:00:19,460 Josh: That is CLAW. And it's not just for home automation. People are running CLAWs 7 00:00:19,460 --> 00:00:21,880 Josh: that monitor code bases and open pull requests when they sleep. 8 00:00:22,200 --> 00:00:24,860 Josh: They even draft these investor reports by pulling live market data. 9 00:00:25,040 --> 00:00:28,320 Josh: It can do anything that you can do, but digitally on your computer. 10 00:00:28,620 --> 00:00:31,680 Josh: The word claw describes something different from the AI that most people know. 11 00:00:31,800 --> 00:00:35,140 Josh: A lot of people are used to chatbots, which you type into, and it forgets everything 12 00:00:35,140 --> 00:00:36,320 Josh: about you by the next session. 13 00:00:36,580 --> 00:00:39,380 Josh: And then there's agents that are capable of actually going on the web and doing 14 00:00:39,380 --> 00:00:41,140 Josh: things for you, acting on your behalf. 15 00:00:41,400 --> 00:00:44,580 Josh: But then there's claws, things that run in the background that can orchestrate 16 00:00:44,580 --> 00:00:47,760 Josh: complicated tasks that have persistent memory that never forget. 17 00:00:47,760 --> 00:00:52,820 Josh: And that, that is what every major AI company in the world is building right now. 18 00:00:53,100 --> 00:00:56,940 Josh: Anthropic, OpenAI, Google, Perplexity, everybody is building a claw. 19 00:00:57,160 --> 00:00:59,600 Josh: And this episode, we're going to walk through all the options that you have 20 00:00:59,600 --> 00:01:00,960 Josh: and which one is overall best for you. 21 00:01:01,140 --> 00:01:05,580 Ejaaz: Yeah, the AI stack has changed quite aggressively over the last couple of months. 22 00:01:05,720 --> 00:01:09,720 Ejaaz: If we think of like the LLM as like the brain, then claws are kind of like the 23 00:01:09,720 --> 00:01:12,240 Ejaaz: arms and the limbs that allow you to actually do things. 24 00:01:12,520 --> 00:01:16,640 Ejaaz: Now, everyone's familiar with ChatGBT and clawed. It's kind of like you ask 25 00:01:16,640 --> 00:01:20,260 Ejaaz: it something and it gives you an answer in text, but it doesn't actually do anything for you. 26 00:01:20,480 --> 00:01:23,980 Ejaaz: And then we had the progression to AI agents, which is basically strapping an 27 00:01:23,980 --> 00:01:27,740 Ejaaz: LLM with a bunch of different tools, your email, Slack, whatever it might be. 28 00:01:27,860 --> 00:01:29,660 Ejaaz: But it was still super clunky. 29 00:01:29,800 --> 00:01:32,900 Ejaaz: It could like use the tool, but it didn't know when to use the tool. 30 00:01:33,020 --> 00:01:36,560 Ejaaz: And that's when OpenClaw and Claw started becoming really popular because it 31 00:01:36,560 --> 00:01:42,280 Ejaaz: tied in persistent memory and context about you with these different tools to 32 00:01:42,280 --> 00:01:46,580 Ejaaz: create a personal agent that actually worked. And that's why it went super viral. 33 00:01:46,840 --> 00:01:51,520 Ejaaz: Remember, this was a vibe-coded project, OpenClaw, formerly known as ClawedBot, 34 00:01:51,620 --> 00:01:53,780 Ejaaz: over the weekend by Pete Stuy. 35 00:01:53,900 --> 00:01:58,100 Ejaaz: And it went completely viral, where it has now tens of millions of users building 36 00:01:58,100 --> 00:02:00,880 Ejaaz: their own versions of personal agents every single day. 37 00:02:01,000 --> 00:02:03,300 Ejaaz: So if ChatGPT was the brain, you 38 00:02:03,300 --> 00:02:06,960 Ejaaz: now have OpenClaw or all these different claws, which do stuff for you. 39 00:02:07,120 --> 00:02:11,020 Josh: So now that we're familiar with what a claw is, there are kind of two tiers of these claws. 40 00:02:11,160 --> 00:02:15,240 Josh: One being plug and play. these are companies like anthropic open ai meta through 41 00:02:15,240 --> 00:02:18,400 Josh: their madness thing and perplexity these are the ones where you probably have 42 00:02:18,400 --> 00:02:20,880 Josh: an account it's twenty dollars a month they give you access to all the tools 43 00:02:20,880 --> 00:02:25,160 Josh: and then there is the fully custom ones this is what you are probably most familiar with 44 00:02:25,360 --> 00:02:29,300 Josh: being OpenClaw and all of the open source alternatives to the OpenClaw infrastructure. 45 00:02:29,300 --> 00:02:32,640 Josh: That's kind of for the hardcore tinkerers or people who really want to build 46 00:02:32,640 --> 00:02:34,420 Josh: custom stacks for their claws. 47 00:02:34,580 --> 00:02:38,060 Josh: But we should be starting probably with the foundation labs, 48 00:02:38,240 --> 00:02:40,800 Josh: the labs that everyone's familiar with, because chances are you already use 49 00:02:40,800 --> 00:02:42,420 Josh: them and have an account open with them already. 50 00:02:42,740 --> 00:02:48,180 Ejaaz: Yeah. So OpenAI of all the foundation labs is kind of what took OpenClaw being 51 00:02:48,180 --> 00:02:50,920 Ejaaz: a niche project to something more mainstream. 52 00:02:51,160 --> 00:02:55,640 Ejaaz: They technically didn't acquire Pete Stye and the OpenCore project. 53 00:02:55,820 --> 00:02:58,580 Ejaaz: It's technically still open source and they created a foundation, 54 00:02:58,580 --> 00:03:02,420 Ejaaz: but OpenAI basically owns OpenCore at this point. Pete Stye works. 55 00:03:02,580 --> 00:03:07,700 Ejaaz: He's a full-time employee at OpenAI and he's building out their version of OpenCore for OpenAI. 56 00:03:08,060 --> 00:03:12,260 Ejaaz: And so in Sam Altman's announcement post, he says he's joining OpenAI to drive 57 00:03:12,260 --> 00:03:13,540 Ejaaz: the generation of personal agents. 58 00:03:13,680 --> 00:03:18,380 Ejaaz: And what's interesting about Pete Stye's take on this is he still wants to build 59 00:03:18,380 --> 00:03:20,380 Ejaaz: OpenCore in the vision that he has for it. 60 00:03:20,640 --> 00:03:24,280 Ejaaz: But something interesting that Pete Stein mentioned on his podcast with Lex 61 00:03:24,280 --> 00:03:28,340 Ejaaz: Friedman is one of the reasons why Sam Olman was super excited about acquiring 62 00:03:29,120 --> 00:03:34,540 Ejaaz: OpenClaw Project as a whole and bringing it so close to ChatGPT is Sam realized a weakness in ChatGPT. 63 00:03:35,090 --> 00:03:39,850 Ejaaz: No matter how many apps or tools he created, no matter how much of an app store 64 00:03:39,850 --> 00:03:41,810 Ejaaz: he built on top of ChatGPT, 65 00:03:42,350 --> 00:03:46,310 Ejaaz: developers didn't really want to build on OpenAI because it didn't give them 66 00:03:46,310 --> 00:03:50,970 Ejaaz: the customization capability to be able to build the types of apps that they wanted to do. 67 00:03:51,110 --> 00:03:55,690 Ejaaz: And in their fairness, they can't envision exactly what AI apps look like right 68 00:03:55,690 --> 00:03:57,270 Ejaaz: now. This is the Wild West at the moment. 69 00:03:57,530 --> 00:04:01,290 Ejaaz: So he realized that it was probably best to build the number one or bring the 70 00:04:01,290 --> 00:04:04,870 Ejaaz: number one open source ecosystem for building customized personal AI agents 71 00:04:04,870 --> 00:04:07,630 Ejaaz: into OpenAI. And that's what he did with Pete Starr. 72 00:04:07,930 --> 00:04:11,730 Ejaaz: Yeah, so since the acquisition, OpenAI has built out a suite of different features, 73 00:04:11,730 --> 00:04:16,110 Ejaaz: which create their own version of OpenClaw for ChatGPT. 74 00:04:16,410 --> 00:04:20,670 Ejaaz: A few come to mind. So they have computer use, which allows ChatGPT to take 75 00:04:20,670 --> 00:04:24,210 Ejaaz: over your browser or desktop itself and do a bunch of things. 76 00:04:24,470 --> 00:04:29,610 Ejaaz: Now, it's important to know that OpenAI to date has acquired a lot of memory and context about you. 77 00:04:29,710 --> 00:04:32,990 Ejaaz: And now they're matching that with a personal agent, which is exactly what OpenClaw 78 00:04:32,990 --> 00:04:34,010 Ejaaz: is. So that's great to see. 79 00:04:34,150 --> 00:04:38,390 Ejaaz: They also have an AI browser called Atlas, which allows you to do any of the 80 00:04:38,390 --> 00:04:39,970 Ejaaz: browser extension type of stuff. 81 00:04:40,070 --> 00:04:42,730 Ejaaz: So if you want to do research or if you want to browse your email or do anything 82 00:04:42,730 --> 00:04:46,170 Ejaaz: that is in your online digital world, it can now also do that. 83 00:04:46,530 --> 00:04:49,450 Ejaaz: And the final thing is, of course, their coding agent, Codex itself, 84 00:04:49,730 --> 00:04:52,990 Ejaaz: which they integrate with the former two tools that I just mentioned, 85 00:04:53,230 --> 00:04:57,190 Ejaaz: which allow them to do the building and creation of new things that you want. 86 00:04:57,190 --> 00:05:00,810 Ejaaz: So you have this personal agent that sits within ChatGPT right now that can 87 00:05:00,810 --> 00:05:03,210 Ejaaz: do so much more than just be a chatbot LLM. 88 00:05:03,450 --> 00:05:08,370 Josh: Now, perhaps you don't have an OpenAI subscription and you have something like Anthropic on Claude. 89 00:05:08,690 --> 00:05:13,090 Josh: Well, you're in luck. Anthropic has probably the most robust alternative to OpenClaw. 90 00:05:13,170 --> 00:05:16,770 Josh: They didn't acquire OpenClaw, so therefore they were forced to build their own version. 91 00:05:16,770 --> 00:05:21,630 Josh: And they've deployed quite a few features and utilities that really make it 92 00:05:21,630 --> 00:05:25,070 Josh: a compelling offering for the average person who doesn't want to fully build 93 00:05:25,070 --> 00:05:28,390 Josh: a custom Claw stack. And there's this whole laundry list of things, 94 00:05:28,530 --> 00:05:29,530 Josh: starting with Claude Cowork. 95 00:05:30,150 --> 00:05:32,910 Josh: Funny enough, it only launched, what, less than three months ago? 96 00:05:33,090 --> 00:05:34,150 Josh: It hasn't even been around for that long. 97 00:05:34,530 --> 00:05:38,910 Josh: Two months ago. But it has been such a huge game changer because Cowork works 98 00:05:38,910 --> 00:05:40,110 Josh: with your entire computer. 99 00:05:40,310 --> 00:05:44,090 Josh: It allows it to access files on your desktop, execute those files, 100 00:05:44,130 --> 00:05:47,390 Josh: and actually make changes on your home computer so it can take over whatever 101 00:05:47,390 --> 00:05:48,930 Josh: you give it access to in a secure way. 102 00:05:49,210 --> 00:05:53,250 Josh: Just recently, yesterday, they released Routines, which are basically cron jobs 103 00:05:53,250 --> 00:05:56,650 Josh: for those who don't know. They're scheduled tasks that you can do with your computer off. 104 00:05:56,750 --> 00:05:59,690 Josh: You'll find a lot of people have been hyping up purchasing mac minis because 105 00:05:59,690 --> 00:06:02,510 Josh: they allow your computer to sit on your desk and stay always 106 00:06:02,510 --> 00:06:05,610 Josh: on well these new routines offer a cron 107 00:06:05,610 --> 00:06:08,650 Josh: job feature for your claude instance without your 108 00:06:08,650 --> 00:06:11,670 Josh: computer needing to be on all the time that's very cool they recently rolled 109 00:06:11,670 --> 00:06:16,010 Josh: out computer use which lets claude actually see your screen move the mouse around 110 00:06:16,010 --> 00:06:20,090 Josh: it can click type navigate apps and it's all permission gated so it's very secure 111 00:06:20,090 --> 00:06:24,710 Josh: you kind of choose what you want it to have access to and then it will go off 112 00:06:24,710 --> 00:06:27,430 Josh: and do whatever tasks that you ask it to do And then finally, 113 00:06:27,590 --> 00:06:30,030 Josh: there's this little thing you may have heard of called Clawed Code, 114 00:06:30,050 --> 00:06:34,950 Josh: which is probably the most popular coding developer engine that currently exists, 115 00:06:35,150 --> 00:06:36,570 Josh: all baked into one super app. 116 00:06:36,750 --> 00:06:41,190 Josh: So the Clawed Desktop app is a super app that is functioning as a hybrid claw 117 00:06:41,190 --> 00:06:44,850 Josh: that allows you to schedule these tasks, manage agents, actually engage with 118 00:06:44,850 --> 00:06:48,290 Josh: your computer and do so whether it's turned on or off. It's a really powerful piece of software. 119 00:06:48,390 --> 00:06:52,210 Josh: And personally, this is what I find myself using the most. I love the Anthropic 120 00:06:52,210 --> 00:06:55,370 Josh: stack that they've built, how there's connectors and plugins to a lot of the 121 00:06:55,370 --> 00:06:56,810 Josh: tools that I use that are third party. 122 00:06:56,990 --> 00:07:01,150 Josh: And I found it to be just very easy to set up, very robust and very secure in 123 00:07:01,150 --> 00:07:04,270 Josh: the sense that it only accesses the things that I explicitly give it permission to. 124 00:07:04,770 --> 00:07:09,970 Ejaaz: Yeah, it's really funny. Two months ago, Anthropic was the most hated company 125 00:07:09,970 --> 00:07:12,590 Ejaaz: from the OpenClaw community for one simple reason. 126 00:07:12,770 --> 00:07:17,930 Ejaaz: They started just banning people using Anthropic claw tools for OpenClaw. 127 00:07:18,150 --> 00:07:21,690 Ejaaz: And no one quite knew why. People thought that they fumbled by like, 128 00:07:21,790 --> 00:07:24,350 Ejaaz: you know, this was a missed PR opportunity. You could have gotten so many new 129 00:07:24,350 --> 00:07:25,310 Ejaaz: subscriptions and users. 130 00:07:25,630 --> 00:07:28,830 Ejaaz: But Anthropic wouldn't let loose why they were doing this thing. 131 00:07:28,830 --> 00:07:33,050 Ejaaz: Fast forward two months, and they've released, I believe it's 30 plus features 132 00:07:33,050 --> 00:07:39,150 Ejaaz: specifically to rebuild OpenClaw within a sandboxed and closed environment. 133 00:07:39,390 --> 00:07:43,330 Ejaaz: And the reason is pretty fair, which is OpenClaw is kind of the wild west. 134 00:07:43,450 --> 00:07:44,750 Ejaaz: There's a lot of security flaws. 135 00:07:44,950 --> 00:07:48,890 Ejaaz: It's kind of expensive. It's kind of hard to use. We create a packaged, 136 00:07:49,090 --> 00:07:53,170 Ejaaz: curated, cheap version all for 20 bucks a month for you to use. 137 00:07:53,230 --> 00:07:56,030 Ejaaz: And you don't have to worry about, you know, losing all your photos or your 138 00:07:56,030 --> 00:08:00,670 Ejaaz: hard drive information. You could just quietly and confidently use this product and it just works. 139 00:08:00,830 --> 00:08:03,290 Ejaaz: And that's what they shipped over the last two months. 140 00:08:03,470 --> 00:08:07,490 Ejaaz: That's kind of a little bit of a misdirection because really it's over the last 141 00:08:07,490 --> 00:08:11,610 Ejaaz: three weeks, they shipped the bulk of those features, just insane amount of delivery. 142 00:08:11,790 --> 00:08:15,250 Ejaaz: It wouldn't surprise me if they're using some kind of closed open core implementation 143 00:08:15,250 --> 00:08:18,250 Ejaaz: that they built within themselves to actually build and launch these features. 144 00:08:18,890 --> 00:08:22,570 Ejaaz: A few more things that are pretty cool about their version of OpenClaw, 145 00:08:22,750 --> 00:08:24,950 Ejaaz: which OpenClaw itself does not have. 146 00:08:25,090 --> 00:08:30,070 Ejaaz: You can currently shut your laptop that you're watching this video on right 147 00:08:30,070 --> 00:08:34,210 Ejaaz: now and still operate their version of OpenClaw because they have these things 148 00:08:34,210 --> 00:08:38,290 Ejaaz: called dispatch and routines that you mentioned earlier, which allows you to do this. 149 00:08:38,430 --> 00:08:42,610 Ejaaz: So they've really built a comprehensive suite that has rebuilt OpenClaw in a 150 00:08:42,610 --> 00:08:43,730 Ejaaz: much more meaningful way. 151 00:08:43,750 --> 00:08:48,970 Ejaaz: And it's important because Anthropic currently has the context and data on millions 152 00:08:48,970 --> 00:08:52,490 Ejaaz: and millions and millions of new retail users, which is key. 153 00:08:52,690 --> 00:08:55,170 Ejaaz: Like when it comes down to it, if everyone's building an open call, 154 00:08:55,550 --> 00:08:57,930 Ejaaz: the key differentiator ends up becoming context and data. 155 00:08:58,330 --> 00:09:01,750 Ejaaz: And Anthropic is one of two major companies right now, the other one being OpenAI 156 00:09:01,750 --> 00:09:03,190 Ejaaz: that has this advantage. 157 00:09:03,470 --> 00:09:05,370 Josh: So if you've heard all that, and Claude isn't quite for you, 158 00:09:05,470 --> 00:09:08,410 Josh: there's another competitor on the block that goes by the name of Meta. 159 00:09:08,590 --> 00:09:11,850 Josh: And now Meta, they actually haven't built their claw harness themselves. 160 00:09:11,850 --> 00:09:14,830 Josh: Instead, they acquired a company named Manus, if you remember, 161 00:09:14,930 --> 00:09:17,430 Josh: for about $2 billion in December of last year. 162 00:09:17,650 --> 00:09:23,130 Josh: And Manus is this harness that deploys fully autonomous agents to write code, 163 00:09:23,230 --> 00:09:26,910 Josh: deploy apps, browse websites, do all the things that you would expect in a claw, 164 00:09:27,050 --> 00:09:30,590 Josh: except doing so now with the meta operating system. 165 00:09:30,710 --> 00:09:33,870 Josh: So if you'll remember, we recorded an episode last week on Meta's new local models. 166 00:09:34,010 --> 00:09:37,950 Josh: They are looking to converge these two models and the agent harness into one 167 00:09:37,950 --> 00:09:42,270 Josh: instance. that seems like it's on the right track to becoming a pretty serious 168 00:09:42,270 --> 00:09:43,370 Josh: open claw competitor as well. 169 00:09:43,530 --> 00:09:48,470 Ejaaz: But not for the reason that you might think. So Manus and Meta have differentiated 170 00:09:48,470 --> 00:09:50,730 Ejaaz: their open claw offering in a very specific way. 171 00:09:51,050 --> 00:09:54,410 Ejaaz: They haven't offered it necessarily. Their hit product hasn't necessarily been 172 00:09:54,410 --> 00:09:57,550 Ejaaz: for retail users. It's been for advertisers. 173 00:09:57,870 --> 00:10:01,070 Ejaaz: Now, that's where Meta makes majority of their money through Facebook, 174 00:10:01,210 --> 00:10:02,490 Ejaaz: Instagram, and the likes. 175 00:10:02,650 --> 00:10:05,970 Ejaaz: They basically have a crazy advertising network. In fact, I read an article 176 00:10:05,970 --> 00:10:09,770 Ejaaz: earlier this week, I think two days ago, where their advertising rev is now 177 00:10:09,770 --> 00:10:13,530 Ejaaz: becoming extremely competitive with Google, which is like the monopoly on the 178 00:10:13,530 --> 00:10:15,230 Ejaaz: sector for God knows how long. 179 00:10:15,840 --> 00:10:21,260 Ejaaz: Manus specifically is being used to help advertisers not only create their advertising 180 00:10:21,260 --> 00:10:23,860 Ejaaz: campaign, but select product imagery, 181 00:10:24,320 --> 00:10:29,420 Ejaaz: control the tools that they use to kind of manufacture and articulate specific 182 00:10:29,420 --> 00:10:30,980 Ejaaz: advertisements to a target audience. 183 00:10:31,180 --> 00:10:34,420 Ejaaz: It does all the research for them in terms of how much time they should expose 184 00:10:34,420 --> 00:10:38,380 Ejaaz: an ad to a particular user. All these finer details is now managed by their 185 00:10:38,380 --> 00:10:40,100 Ejaaz: version of OpenClaw Manus. 186 00:10:40,280 --> 00:10:43,460 Ejaaz: And I think that's such an interesting niche because this is the sort of direction 187 00:10:43,460 --> 00:10:46,340 Ejaaz: that Meta has started to take with their LLMs as well. 188 00:10:46,540 --> 00:10:49,180 Ejaaz: Last week, they launched their own foundational model, the first that they've 189 00:10:49,180 --> 00:10:51,080 Ejaaz: launched in over a year called MuseSpark. 190 00:10:51,200 --> 00:10:54,960 Ejaaz: And it's not amazing as a general LLM, but it's great for specific tasks. 191 00:10:55,060 --> 00:10:57,920 Ejaaz: And they're doing the same with OpenCore. So if you're an advertiser watching 192 00:10:57,920 --> 00:11:00,640 Ejaaz: this, or if you're someone that wants to market their product on any of Meta's 193 00:11:00,640 --> 00:11:03,600 Ejaaz: apps, definitely go check out this product because it's extremely useful for 194 00:11:03,600 --> 00:11:05,200 Ejaaz: you right now and no one else is doing it. 195 00:11:05,340 --> 00:11:09,840 Josh: And the final claw in a box service that we're going to be talking about today is Perplexity, 196 00:11:09,980 --> 00:11:13,340 Josh: who just recently released perplexity computer and perplexity 197 00:11:13,340 --> 00:11:16,020 Josh: computer is interesting and slightly different from the rest of 198 00:11:16,020 --> 00:11:19,060 Josh: these options because perplexity is model agnostic it 199 00:11:19,060 --> 00:11:21,840 Josh: actually uses i believe 19 separate models that you can kind 200 00:11:21,840 --> 00:11:24,740 Josh: of choose from and it'll pick depending on what the task 201 00:11:24,740 --> 00:11:27,540 Josh: that you're using is and that model agnostic allows it 202 00:11:27,540 --> 00:11:30,720 Josh: to do things that are kind of unique to the perplexity platform it 203 00:11:30,720 --> 00:11:33,440 Josh: does everything that you expect it to it can do the research it can 204 00:11:33,440 --> 00:11:37,520 Josh: design code deploy projects whatever kind of miscellaneous tasks you want to 205 00:11:37,520 --> 00:11:41,760 Josh: do but it also has the benefit of being able to do so i mean you could do it 206 00:11:41,760 --> 00:11:43,960 Josh: fully in the cloud on the servers but they're also going to be rolling out a 207 00:11:43,960 --> 00:11:46,760 Josh: version that runs locally on your machine similar to what we were talking about 208 00:11:46,760 --> 00:11:50,720 Josh: with claude and open ai so perplexity is also in there it seems like the most 209 00:11:50,720 --> 00:11:52,340 Josh: accessible one i think a lot of people. 210 00:11:52,990 --> 00:11:56,070 Josh: Use perplexity kind of as a google extension and 211 00:11:56,070 --> 00:11:58,990 Josh: the way they've integrated computer is very easy and simple to 212 00:11:58,990 --> 00:12:01,870 Josh: understand in fact one of the new things that i found was cool 213 00:12:01,870 --> 00:12:04,650 Josh: is they deployed a tax expert where it 214 00:12:04,650 --> 00:12:07,430 Josh: will actually go through your filings and your w-2s and 215 00:12:07,430 --> 00:12:10,290 Josh: it'll give you the best tax advice that it can 216 00:12:10,290 --> 00:12:12,970 Josh: being a non-cpa but what i 217 00:12:12,970 --> 00:12:15,770 Josh: find interesting about this is just how accessible it is 218 00:12:15,770 --> 00:12:19,090 Josh: i find that a lot of the other things that we've talked about they're 219 00:12:19,090 --> 00:12:22,010 Josh: fairly open-ended it's on you to figure out your use cases for 220 00:12:22,010 --> 00:12:25,150 Josh: what you want to use this claw architecture for with perplexity 221 00:12:25,150 --> 00:12:27,950 Josh: they really do a good job of kind of honing in on specific use 222 00:12:27,950 --> 00:12:32,310 Josh: cases for you like the taxes like recently they interray with your bank accounts 223 00:12:32,310 --> 00:12:35,450 Josh: where it could actually see your specific finances and and give you these kind 224 00:12:35,450 --> 00:12:39,350 Josh: of unique but specific use cases and i think for a lot of people that may be 225 00:12:39,350 --> 00:12:42,650 Josh: interesting so i would say perplexity probably is the one that holds your hand 226 00:12:42,650 --> 00:12:46,210 Josh: the most and if you want something simple model agnostic that would be the one to choose. 227 00:12:46,910 --> 00:12:50,690 Ejaaz: It's hilarious to me because every time I think perplexity doesn't have a moat 228 00:12:50,690 --> 00:12:53,590 Ejaaz: and they're going to die, they somehow prove me wrong. 229 00:12:54,250 --> 00:12:58,110 Ejaaz: Like you mentioned, it's model agnostic. They sit on all of these other LLM 230 00:12:58,110 --> 00:13:01,910 Ejaaz: providers. Their moat isn't really the fact that they can access all these different models. 231 00:13:02,630 --> 00:13:05,690 Ejaaz: It's because they've created something called an agent harness, 232 00:13:05,910 --> 00:13:09,810 Ejaaz: which is, I guess, something that sits on top of all the agents that you would create. 233 00:13:09,930 --> 00:13:13,570 Ejaaz: It helps provide the right context at the right time and using the right tools 234 00:13:13,570 --> 00:13:16,790 Ejaaz: at the right time, which sounds like a claw, but they just do it better. 235 00:13:16,910 --> 00:13:20,190 Ejaaz: Across a variety of different models where Anthropik is just focused on Claude 236 00:13:20,190 --> 00:13:22,670 Ejaaz: and OpenAI is just focused on ChatGPT. 237 00:13:23,110 --> 00:13:28,190 Ejaaz: And the advantage that they have is not only is it cheaper, but it's more contextually 238 00:13:28,190 --> 00:13:30,170 Ejaaz: oriented around a specific task that you're doing. 239 00:13:30,270 --> 00:13:33,970 Ejaaz: So you mentioned two, you mentioned taxes and you mentioned the financials, 240 00:13:34,210 --> 00:13:36,370 Ejaaz: specifically like you can act like a CFO right now. 241 00:13:36,610 --> 00:13:40,970 Ejaaz: That's a good example of like how it can go deep into a specific vertical for 242 00:13:40,970 --> 00:13:44,090 Ejaaz: something that you don't want to do, whether it's taxes or sorting out your 243 00:13:44,090 --> 00:13:44,910 Ejaaz: accounting or finances. 244 00:13:44,950 --> 00:13:47,270 Ejaaz: And it can like figure that out in a couple of seconds. 245 00:13:47,790 --> 00:13:52,870 Ejaaz: And all of it for accessible, I think it's like for like a $20 a month package. So very cool. 246 00:13:53,170 --> 00:13:55,750 Josh: Yeah, that's the price of all these kind of all-in-one boxes. 247 00:13:55,750 --> 00:13:59,050 Josh: Is there about $20 a month to use the software that gives you access to all 248 00:13:59,050 --> 00:14:00,830 Josh: of these claw functionality tools? 249 00:14:01,070 --> 00:14:05,230 Josh: Now, that is the conclusion of the closed source all in a box, 250 00:14:05,530 --> 00:14:07,430 Josh: press one button and have everything done for you. 251 00:14:07,810 --> 00:14:12,310 Josh: Next is the open source movement. This is where OpenClaw lives and breathes. 252 00:14:12,390 --> 00:14:17,370 Josh: This is the most popular form of claw currently, just because OpenClaw went so viral so quickly. 253 00:14:17,830 --> 00:14:21,730 Josh: And OpenClaw is, like everyone knows, the open source version that started it all. 254 00:14:21,850 --> 00:14:26,570 Josh: It is the orchestration platform for agents that gets things done and is fully customizable. 255 00:14:26,750 --> 00:14:30,950 Josh: So if you find yourself as someone who is a tinkerer, someone who likes to experiment, 256 00:14:31,150 --> 00:14:35,550 Josh: someone who has resilience to fix a lot of bug issues that you may run into 257 00:14:35,550 --> 00:14:39,010 Josh: this is the platform for you because of how customizable it is due to the open 258 00:14:39,010 --> 00:14:43,210 Josh: source nature you have full platform control of everything that it does 259 00:14:43,650 --> 00:14:48,190 Josh: from the md files which suggest how it should act and behave to the actual code 260 00:14:48,190 --> 00:14:51,850 Josh: base that that you that it uses every day to actually make these decisions and 261 00:14:51,850 --> 00:14:55,470 Josh: enact on your behalf if you're an open claw user you're actively using open 262 00:14:55,470 --> 00:14:58,250 Josh: claw or you've at least already tried it it's more for the people who want the 263 00:14:58,250 --> 00:15:01,470 Josh: full customization who want to do anything that they can imagine with it and 264 00:15:01,470 --> 00:15:03,590 Josh: And there's also some interesting examples, Ides, you were telling me, 265 00:15:03,970 --> 00:15:07,310 Josh: of things that people have done that you wouldn't really necessarily think is 266 00:15:07,310 --> 00:15:09,630 Josh: something an OpenClaw instance would solve. 267 00:15:09,850 --> 00:15:14,090 Ejaaz: Okay. What's a problem that you might have five years from now? 268 00:15:14,290 --> 00:15:18,110 Ejaaz: Let me explain. Let me paint this picture for you. You own a house. 269 00:15:18,760 --> 00:15:23,460 Ejaaz: But it's missing one quintessential component. You don't have a pool in your backyard, 270 00:15:23,920 --> 00:15:24,040 Josh: Josh. 271 00:15:24,540 --> 00:15:27,320 Ejaaz: Well, what if I could help you with that? 272 00:15:27,560 --> 00:15:31,460 Ejaaz: What if not me could help you with that, but what if my OpenClaw or my Claw 273 00:15:31,460 --> 00:15:32,680 Ejaaz: agent could help you with that? 274 00:15:32,820 --> 00:15:33,420 Josh: Sounds lovely. 275 00:15:33,880 --> 00:15:40,540 Ejaaz: Someone created an OpenClaw agent that finds 500,000 to $1.2 million homes without a pool, 276 00:15:41,120 --> 00:15:46,520 Ejaaz: gets the images from Zillow or any other public website, renders a pool or an 277 00:15:46,520 --> 00:15:50,800 Ejaaz: image of a pool into their backyard, and then sends it to them with a cold DM 278 00:15:50,800 --> 00:15:54,800 Ejaaz: or cold email, basically saying, hey, doesn't this pool look really nice on you? 279 00:15:55,020 --> 00:15:59,120 Ejaaz: You should like consider leasing us or using us to create and set up your pool for them. 280 00:15:59,420 --> 00:16:02,920 Ejaaz: And the inbound has just been insane for it. So it scans satellite imagery for 281 00:16:02,920 --> 00:16:07,060 Ejaaz: mid-market homes, filters lot by size, pulls homeowner direct from public records, 282 00:16:07,200 --> 00:16:09,540 Ejaaz: and then like gets in contact with them with the new rendered image. 283 00:16:09,780 --> 00:16:12,860 Ejaaz: And the conversion rate, supposedly, from the claim, this is Twitter, 284 00:16:13,000 --> 00:16:16,920 Ejaaz: this is X after all, there are some crazy claims, has apparently been pretty 285 00:16:16,920 --> 00:16:19,300 Ejaaz: damn good. And if you look at the video over here on the right, 286 00:16:19,500 --> 00:16:21,000 Ejaaz: it looks pretty photorealistic. 287 00:16:21,260 --> 00:16:24,260 Ejaaz: It has like a video version of it as well. So you can see it kind of like a 288 00:16:24,260 --> 00:16:27,340 Ejaaz: pool floaty going around. Just a cool use of open core. 289 00:16:27,950 --> 00:16:33,370 Ejaaz: And then this second example, which I found like super interesting was this 290 00:16:33,370 --> 00:16:39,310 Ejaaz: mother uses or currently runs 11 open claw agents to help her raise her kids. 291 00:16:39,590 --> 00:16:44,210 Ejaaz: Now, that sounds insanely dystopian, but the framework is ironically, 292 00:16:44,430 --> 00:16:48,010 Ejaaz: in my opinion, one of the most useful examples of using open claw agents ever. 293 00:16:48,210 --> 00:16:51,870 Ejaaz: So one example is she has a job to maintain. 294 00:16:52,130 --> 00:16:56,350 Ejaaz: So one version of using all these open claw agents is she only uses voice space. 295 00:16:56,470 --> 00:16:59,870 Ejaaz: She doesn't type anything. Now, Josh, I know you're a massive fan of using voice 296 00:16:59,870 --> 00:17:01,870 Ejaaz: notes or speaking to your AI alarms. 297 00:17:02,090 --> 00:17:06,230 Ejaaz: So am I. It is just a much more efficient way to use it. So she doesn't code anymore. 298 00:17:06,350 --> 00:17:10,270 Ejaaz: She just speaks to agents about the specific type of thing that she wants to 299 00:17:10,270 --> 00:17:12,850 Ejaaz: build in her particular job. And it goes away and it does that. 300 00:17:13,030 --> 00:17:15,470 Ejaaz: But then she also needs to educate her kids. 301 00:17:15,750 --> 00:17:19,130 Ejaaz: Now, she homeschools her kids. And what she does is she takes her phone. 302 00:17:19,450 --> 00:17:22,790 Ejaaz: She takes a picture of a syllabus that she wants to teach her kid for the day. 303 00:17:22,790 --> 00:17:27,170 Ejaaz: And it generates a personalized lesson plan for each of her different kids tailored 304 00:17:27,170 --> 00:17:30,090 Ejaaz: to their specific preferences and she walks them through it. 305 00:17:30,290 --> 00:17:34,490 Ejaaz: Now, remember, it's homeschooled. So you don't necessarily have all the apparatus in your home. 306 00:17:34,790 --> 00:17:40,790 Ejaaz: She provides that context to all her open core agents and it knows which tool to use. 307 00:17:40,890 --> 00:17:43,990 Ejaaz: So like if there's a toy that she can use to demonstrate the specific lesson, 308 00:17:44,610 --> 00:17:47,990 Ejaaz: prompts her to go grab that toy and do all of this. So you might be wondering, 309 00:17:48,190 --> 00:17:51,370 Ejaaz: well, this is completely insane. Why would you use AI to teach your kids, 310 00:17:51,430 --> 00:17:52,010 Ejaaz: blah, blah, blah, blah, blah. 311 00:17:52,670 --> 00:17:58,690 Ejaaz: The net net effect is it's freed her up to be more of a parent and spend time with her kids. 312 00:17:58,910 --> 00:18:02,230 Ejaaz: So all of that to tie a loop around and said she has more time on her hands 313 00:18:02,230 --> 00:18:04,930 Ejaaz: to spend time with the kids. So I thought that was a super cool use. 314 00:18:05,090 --> 00:18:11,370 Ejaaz: Now, going back to your point around open source, open claw agents in particular. 315 00:18:11,590 --> 00:18:16,910 Ejaaz: The customization is like a massive win for people, but it's also how cheap 316 00:18:16,910 --> 00:18:20,530 Ejaaz: it is for you to be able to run it on a local device. Maybe it's on your phone. 317 00:18:20,790 --> 00:18:24,450 Ejaaz: And also if you want to run like completely crazy projects like raising your 318 00:18:24,450 --> 00:18:28,390 Ejaaz: kids, using 12 open claw agents, you probably don't want to be spending a bunch 319 00:18:28,390 --> 00:18:29,390 Ejaaz: of different money on it. 320 00:18:29,490 --> 00:18:33,550 Ejaaz: And so she specifically uses PicoClaw, which I'm referencing right here. 321 00:18:33,650 --> 00:18:36,870 Ejaaz: And I think in some instances, IronClaw, when she's using personal data and 322 00:18:36,870 --> 00:18:38,530 Ejaaz: she wants to use like a Rust-based architecture. 323 00:18:38,790 --> 00:18:40,790 Ejaaz: All of that to say is all of these are free. 324 00:18:41,070 --> 00:18:45,050 Ejaaz: You can download them, access it yourself, and you can customize it in whichever way you want. 325 00:18:45,150 --> 00:18:47,250 Ejaaz: So if you're listening to this and you come up with a crazy idea, 326 00:18:47,430 --> 00:18:50,010 Ejaaz: please do it and let us know what you built because we're trying to track all 327 00:18:50,010 --> 00:18:51,310 Ejaaz: of these use cases whenever we can. 328 00:18:51,840 --> 00:18:56,180 Josh: Yeah. And a testament to the claw architecture being so resilient and robust. 329 00:18:56,380 --> 00:19:00,240 Josh: The real limitation, as it's always been, is how you can most creatively implement 330 00:19:00,240 --> 00:19:02,400 Josh: these tools to be most effective for you. 331 00:19:02,560 --> 00:19:06,840 Josh: And the ceiling is still like largely untapped. We don't know where this clocks 332 00:19:06,840 --> 00:19:10,560 Josh: out because people are doing really cool and interesting things with it every single day. 333 00:19:10,720 --> 00:19:13,720 Josh: There's so many great examples of it and great enhancements too. 334 00:19:13,800 --> 00:19:17,640 Josh: Now what sits on top of this layer even are the markdown files. 335 00:19:17,860 --> 00:19:21,580 Josh: So if anyone who's familiar with OpenClaw, there is like a soul.md 336 00:19:21,580 --> 00:19:24,400 Josh: file and there there are a series of these markdown files 337 00:19:24,400 --> 00:19:27,060 Josh: that kind of give the model intention they tell it 338 00:19:27,060 --> 00:19:30,120 Josh: what to do they tell it how to act and there's an entire kind 339 00:19:30,120 --> 00:19:34,420 Josh: of pseudo industry that's been built on top of this orchestrating of the agents 340 00:19:34,420 --> 00:19:38,980 Josh: of telling them how to act by embedding interesting markdown files with unique 341 00:19:38,980 --> 00:19:43,660 Josh: sets of directions now gary tan founder of y combinator he appears to have built 342 00:19:43,660 --> 00:19:48,000 Josh: one that has kind of become the most popular one perhaps which is G-Stack. 343 00:19:48,200 --> 00:19:52,060 Josh: He says, can you explain what G-Stack is and why it's this kind of valuable 344 00:19:52,060 --> 00:19:54,500 Josh: second layer to be built on top of your claws? 345 00:19:54,700 --> 00:19:59,400 Ejaaz: Yeah, I've seen conflicting opinions on both sides about what Karytown has built 346 00:19:59,400 --> 00:20:00,820 Ejaaz: here because he hyped it up. 347 00:20:00,940 --> 00:20:05,020 Ejaaz: Basically he said, I have made OpenClaw into this insane personal assistant 348 00:20:05,020 --> 00:20:09,420 Ejaaz: that actually does productive work for you and mine's the best implementation of that. 349 00:20:09,720 --> 00:20:13,740 Ejaaz: A bunch of engineers then evaluated what he had actually built and realized 350 00:20:13,740 --> 00:20:17,320 Ejaaz: that he had just wrote a very extensively And in his fairness, 351 00:20:17,580 --> 00:20:23,520 Ejaaz: very well articulated, meticulously detailed document, which describes what he wants his agent to do. 352 00:20:24,340 --> 00:20:28,880 Ejaaz: Gary Tan, if any of you have watched any of his interviews, is an extremely verbose kind of guy. 353 00:20:29,020 --> 00:20:31,480 Ejaaz: So it makes sense that, you know, he's worked with a bunch of these companies. 354 00:20:31,580 --> 00:20:32,800 Ejaaz: He's had a bit of experience. 355 00:20:32,980 --> 00:20:36,460 Ejaaz: He kind of knows what to ask for. And so the point that he proves with G-Stack 356 00:20:36,460 --> 00:20:40,800 Ejaaz: is, number one, here's an easy and simple way that you can level up your open 357 00:20:40,800 --> 00:20:42,340 Ejaaz: core agent in a matter of minutes. 358 00:20:42,540 --> 00:20:45,260 Ejaaz: Use his framework and architecture and just fill in the blanks, 359 00:20:45,380 --> 00:20:47,160 Ejaaz: like add in the details for your specific project. 360 00:20:47,300 --> 00:20:51,260 Ejaaz: And you now have an amazing blueprint to build an amazing open core agent. 361 00:20:51,420 --> 00:20:55,420 Ejaaz: But number two, how easy is it that it just requires two documents that you 362 00:20:55,420 --> 00:21:00,200 Ejaaz: can spin up on your MacBook or on your laptop right now that can change your agent for the better. 363 00:21:00,380 --> 00:21:04,220 Ejaaz: So the criticism for a bunch of industry folk was like, Gary, 364 00:21:04,340 --> 00:21:06,860 Ejaaz: this is amazing, but this isn't anything necessarily novel. 365 00:21:07,020 --> 00:21:09,500 Ejaaz: It's a skill issue. You need to be able to articulate. 366 00:21:09,760 --> 00:21:14,040 Ejaaz: And that's basically something that is very apparent in the AI age. 367 00:21:14,200 --> 00:21:18,860 Ejaaz: It's moved from being able to know how to code to be able to know how to communicate and explain. 368 00:21:19,040 --> 00:21:22,440 Ejaaz: And that's the trend that we see with OpenClaw and a bunch of these other AILLMs. 369 00:21:22,680 --> 00:21:25,280 Josh: And thankfully, all of this explaining happens in plain English. 370 00:21:25,420 --> 00:21:28,680 Josh: And I think that's the most exciting and empowering part is it doesn't require 371 00:21:28,680 --> 00:21:31,820 Josh: you to be technical or to be a developer in order to improve these things. 372 00:21:31,980 --> 00:21:35,460 Josh: You just need to speak to it and kind of coerce it into doing the things that 373 00:21:35,460 --> 00:21:39,880 Josh: you want to do, whether it be through Anthropic and Claude or through the OpenClaw 374 00:21:39,880 --> 00:21:41,260 Josh: instance that is fully open source. 375 00:21:41,260 --> 00:21:44,360 Josh: There is so much customization to be had and 376 00:21:44,360 --> 00:21:47,120 Josh: i think it's kind of on you to decide where you fall 377 00:21:47,120 --> 00:21:50,160 Josh: on the spectrum of plug and play where you just kind of want a 378 00:21:50,160 --> 00:21:53,860 Josh: claw on rails that something like anthropic can provide to you or if you're 379 00:21:53,860 --> 00:21:57,320 Josh: the tinkering type who really wants to build custom stacks like that mother 380 00:21:57,320 --> 00:22:00,940 Josh: who is using it to help her parents or a company that is built being built around 381 00:22:00,940 --> 00:22:06,000 Josh: creating pools there is no limit to the use cases you can find it's just a matter 382 00:22:06,000 --> 00:22:06,960 Josh: of your preferences and how you 383 00:22:06,960 --> 00:22:10,140 Josh: want to engage with this but it very much seems like this This is the new, 384 00:22:10,800 --> 00:22:14,440 Josh: architecture that's going to stick around. This is kind of the foundation of 385 00:22:14,440 --> 00:22:16,760 Josh: what an AI first operating system could look like. 386 00:22:16,940 --> 00:22:21,300 Josh: And it's very clear that all these companies are converging on the same conclusion that this is true. 387 00:22:21,480 --> 00:22:24,980 Josh: And this is a very important part of the stack to own. And I think the one thing 388 00:22:24,980 --> 00:22:28,260 Josh: that is for sure is we are going to see this continued improvement and progress 389 00:22:28,260 --> 00:22:30,240 Josh: on this at a very rapid rate. 390 00:22:30,420 --> 00:22:34,460 Josh: I mean, Anthropic, they've been on fire, man. They've been releasing new features 391 00:22:34,460 --> 00:22:36,140 Josh: for this every single day at this point. 392 00:22:36,260 --> 00:22:39,960 Josh: So I encourage anyone with an account anywhere, really, just to try it out and 393 00:22:39,960 --> 00:22:43,920 Josh: start flexing that muscle to see where it can improve these whatever parts of your life. 394 00:22:44,080 --> 00:22:48,640 Ejaaz: Yeah, I wonder where this is going over the next couple of years. 395 00:22:48,780 --> 00:22:51,800 Ejaaz: You have all these companies spending hundreds of millions of dollars to either 396 00:22:51,800 --> 00:22:53,720 Ejaaz: acquire the thing or build the thing. 397 00:22:53,900 --> 00:22:57,240 Ejaaz: They're all kind of building similar products. I think what it's ultimately 398 00:22:57,240 --> 00:23:01,860 Ejaaz: going to come down to is the data and context that you can use to train these things. 399 00:23:02,160 --> 00:23:04,400 Ejaaz: After all, that's what like made it a differentiated product. 400 00:23:04,440 --> 00:23:09,040 Ejaaz: And that's what makes it an exceptionally good AI operating system. 401 00:23:09,340 --> 00:23:13,480 Ejaaz: Two companies that I've noticed haven't really got a version of this, 402 00:23:13,640 --> 00:23:16,620 Ejaaz: but probably should, is Google. Well, Google actually does. 403 00:23:16,800 --> 00:23:20,180 Ejaaz: It's called Mariner, right? But no one's ever heard of that, right? 404 00:23:20,520 --> 00:23:23,720 Ejaaz: No, but no one seems to use that. And they seem to have kind of like left that 405 00:23:23,720 --> 00:23:26,040 Ejaaz: by the wayside to focus on their main LLM. 406 00:23:26,280 --> 00:23:30,820 Ejaaz: And the other is Apple, who I'm actually more optimistic on because I know that 407 00:23:30,820 --> 00:23:32,560 Ejaaz: they've signed the deal with Google Gemini, 408 00:23:32,660 --> 00:23:37,580 Ejaaz: and they're going to be launching AI Siri hopefully for the fifth time on a 409 00:23:37,580 --> 00:23:42,220 Ejaaz: failed attempt later this year which should end up becoming a really useful version of 410 00:23:42,720 --> 00:23:46,000 Ejaaz: Their open claw agent. So I'm excited for these companies to launch their things. 411 00:23:46,160 --> 00:23:48,920 Ejaaz: But I believe that is all we have for this episode. 412 00:23:49,160 --> 00:23:52,580 Ejaaz: As Josh mentioned earlier, if you are listening to this, and if you are inspired 413 00:23:52,580 --> 00:23:56,140 Ejaaz: by either the closed sandbox version from Anthropic, 414 00:23:56,420 --> 00:23:59,720 Ejaaz: or the open source version from a PicoClaw or an IronClaw, please, 415 00:23:59,880 --> 00:24:04,020 Ejaaz: we encourage you to go out and use these things because there is a draught of 416 00:24:04,020 --> 00:24:07,220 Ejaaz: viable use cases for a few of these different AI things. Everyone has ideas, 417 00:24:07,320 --> 00:24:08,500 Ejaaz: but no one's actually building the thing. 418 00:24:08,640 --> 00:24:12,540 Ejaaz: It's never been easier to vibe code a custom AI agent. And we want to hear what 419 00:24:12,540 --> 00:24:14,260 Ejaaz: you guys are interested in and are building. 420 00:24:14,760 --> 00:24:17,560 Ejaaz: If you are a listener on this show and you aren't subscribed, 421 00:24:17,760 --> 00:24:21,920 Ejaaz: please do so. Or if you're listening to this on Spotify or Apple Podcasts, please give us a rating. 422 00:24:22,240 --> 00:24:24,660 Ejaaz: Leave us a comment. We always love hearing feedback from you guys. 423 00:24:24,800 --> 00:24:27,780 Josh: And we'll be back tomorrow for the weekly roundup and possibly a surprise. 424 00:24:28,020 --> 00:24:30,700 Josh: There's a rumor in the mill that Opus 4.7 is coming. 425 00:24:31,000 --> 00:24:34,300 Josh: So we may have some big news dropping later this week. We'll see. 426 00:24:34,500 --> 00:24:36,440 Josh: But we will be here to monitor the situation as always. 427 00:24:36,820 --> 00:24:40,000 Josh: And as always, Also, thank you so much for listening, for watching wherever 428 00:24:40,000 --> 00:24:41,860 Josh: you are, and we'll see you guys in the next one.