1
00:00:00,020 --> 00:00:04,720
Speaker0:
OpenClaw was a weekend project that turned into the fastest growing open source project ever.

2
00:00:04,960 --> 00:00:08,380
Speaker0:
It got acquired by OpenAI in just 80 days for billions of dollars.

3
00:00:08,540 --> 00:00:10,620
Speaker0:
But the number one question we keep asking ourselves is,

4
00:00:10,990 --> 00:00:15,470
Speaker0:
What can this thing actually do? Is this something for me or someone more technical?

5
00:00:15,690 --> 00:00:19,330
Speaker0:
Is this something that is useful for me right now? Or do I just need to wait

6
00:00:19,330 --> 00:00:21,350
Speaker0:
for a bit? So we did all the hard work for you.

7
00:00:21,490 --> 00:00:25,010
Speaker0:
We've been testing OpenClaw ourselves for the last week, watching every single

8
00:00:25,010 --> 00:00:26,070
Speaker0:
demo we could get a hand on.

9
00:00:26,270 --> 00:00:29,530
Speaker0:
And on this episode, we're going to show you exactly what OpenClaw can do,

10
00:00:29,630 --> 00:00:33,670
Speaker0:
from all the manual stuff to all the actual really useful, mind-blowing stuff.

11
00:00:33,830 --> 00:00:36,990
Speaker0:
And we're going to answer one simple question. Is this for you?

12
00:00:37,130 --> 00:00:40,250
Speaker1:
The answers to a lot of these questions might surprise you. and

13
00:00:40,250 --> 00:00:43,590
Speaker1:
the way we're going to outline this is is kind of through three archetypes the

14
00:00:43,590 --> 00:00:46,250
Speaker1:
first being the operator and builder this is kind of

15
00:00:46,250 --> 00:00:49,250
Speaker1:
the person who creates net new value maybe the entrepreneur

16
00:00:49,250 --> 00:00:52,010
Speaker1:
the second is the knowledge worker or the creator this is

17
00:00:52,010 --> 00:00:54,670
Speaker1:
kind of where we fall into as podcasters and then we have the

18
00:00:54,670 --> 00:00:57,710
Speaker1:
privacy focused professionals and within

19
00:00:57,710 --> 00:01:00,370
Speaker1:
that there's a lot of cool use cases and examples so the first one that we want to

20
00:01:00,370 --> 00:01:04,370
Speaker1:
highlight here is this guy nat eliason he's this incredible follow on twitter

21
00:01:04,370 --> 00:01:10,010
Speaker1:
and he released a bot called felixbot and felixbot was an agent that his OpenClaw

22
00:01:10,010 --> 00:01:14,530
Speaker1:
spawned up and FelixBot has done a series of incredible things that I didn't

23
00:01:14,530 --> 00:01:19,290
Speaker1:
think were possible and allowed me to reconsider what OpenClaw is actually capable of.

24
00:01:19,630 --> 00:01:23,350
Speaker1:
So in this instance, he trained FelixBot to make him money.

25
00:01:23,510 --> 00:01:27,250
Speaker1:
And what we're seeing on screen here is the first post of the weekly revenue

26
00:01:27,250 --> 00:01:31,110
Speaker1:
numbers in which FelixCraft, FelixBot, whatever you want to call it,

27
00:01:31,410 --> 00:01:35,930
Speaker1:
it actually generated $41,000 in a week. Now, how did it do this?

28
00:01:36,230 --> 00:01:40,870
Speaker1:
First, it created its own book. So this Felix Craft bot, by interfacing with

29
00:01:40,870 --> 00:01:44,810
Speaker1:
Nat, decided that it was best in order to make money to create a book about

30
00:01:44,810 --> 00:01:47,990
Speaker1:
AI and about OpenClaw and sell that book.

31
00:01:48,130 --> 00:01:51,770
Speaker1:
It then marketed its own book through this Twitter profile that we're looking at right now.

32
00:01:51,810 --> 00:01:56,310
Speaker1:
And it sold 132 copies in that one week that yielded $3,828.

33
00:01:56,970 --> 00:02:00,550
Speaker1:
Now, the second thing it did is it went and it spawned up a token.

34
00:02:00,830 --> 00:02:04,230
Speaker1:
And it earned trading fees on that token from people who are speculating.

35
00:02:04,230 --> 00:02:08,210
Speaker1:
And the trading fees from that token were $37,698.

36
00:02:08,710 --> 00:02:12,810
Speaker1:
And this was all done through the text interface on Telegram

37
00:02:13,340 --> 00:02:15,780
Speaker1:
Going back and forth, chatting with the bot, asking it what to do.

38
00:02:15,860 --> 00:02:21,660
Speaker1:
And it had the agency and it had the creativity to go and create these pieces of value.

39
00:02:21,860 --> 00:02:26,760
Speaker1:
Now, the next thing that this thing built is this service called ClawMart.

40
00:02:26,960 --> 00:02:28,720
Speaker1:
And ClawMart is what we're seeing on screen here.

41
00:02:28,960 --> 00:02:35,580
Speaker1:
It's a service in which agents, OpenClaw instances, can sell their skills to other OpenClaws.

42
00:02:35,740 --> 00:02:38,960
Speaker1:
So if I was using my OpenClaw and I was actually scrolling through the website

43
00:02:38,960 --> 00:02:40,120
Speaker1:
and I was looking at interesting things.

44
00:02:40,220 --> 00:02:44,140
Speaker1:
And one that I found really cool was they had this browser-based research tool

45
00:02:44,140 --> 00:02:47,960
Speaker1:
that allowed you to scan through a lot of the top news articles and understand

46
00:02:47,960 --> 00:02:50,680
Speaker1:
what was happening in the world of AI and Frontier technology.

47
00:02:50,720 --> 00:02:56,440
Speaker1:
I could have my CloudBot instance go to this website that FelixBot built and

48
00:02:56,440 --> 00:02:59,580
Speaker1:
buy a skill from it that teaches it how to do these skills.

49
00:02:59,720 --> 00:03:02,620
Speaker1:
So you could see the most popular persona is the Felix one.

50
00:03:02,740 --> 00:03:05,920
Speaker1:
He's selling it for $99 on the website where you can actually emulate all

51
00:03:05,920 --> 00:03:08,560
Speaker1:
of the abilities of this agent and i found this to be

52
00:03:08,560 --> 00:03:11,320
Speaker1:
such a fun interesting use case of how you can actually

53
00:03:11,320 --> 00:03:15,420
Speaker1:
use this thing to generate money and generate real productive value like claw

54
00:03:15,420 --> 00:03:19,900
Speaker1:
mart is a really valuable service that i think a lot of other ais can use and

55
00:03:19,900 --> 00:03:23,220
Speaker1:
it's funny because it's all ai to ai transactions i think in the last week there's

56
00:03:23,220 --> 00:03:26,800
Speaker1:
a post somewhere that says they had two thousand dollars in sales transactions

57
00:03:26,800 --> 00:03:29,240
Speaker1:
between OpenClaw agents it's pretty awesome Yeah,

58
00:03:29,500 --> 00:03:35,180
Speaker0:
Yeah, super cool. What this reminds me of is the early versions of the Apple App Store.

59
00:03:35,560 --> 00:03:37,900
Speaker0:
It kind of looks like an app store as you're scrolling through this,

60
00:03:38,020 --> 00:03:40,040
Speaker0:
except it's like services through these agents.

61
00:03:40,740 --> 00:03:45,120
Speaker0:
I think this is the future of how all these online interactions actually happen.

62
00:03:45,240 --> 00:03:48,880
Speaker0:
It kind of makes sense that an agent doesn't really interact with another human,

63
00:03:48,880 --> 00:03:52,600
Speaker0:
and it doesn't really kind of code skills from scratch each and every time.

64
00:03:52,600 --> 00:03:55,940
Speaker0:
This is the whole argument around the SaaS debate and why SaaS stocks have been dumping.

65
00:03:56,140 --> 00:03:59,140
Speaker0:
It's because, oh, this AI can just kind of vibe code your product.

66
00:03:59,300 --> 00:04:01,600
Speaker0:
No, that's not really what's going to happen. you're going to just rely on the

67
00:04:01,600 --> 00:04:06,320
Speaker0:
agent that has the best product and pay them whatever it is between 50 to 100

68
00:04:06,320 --> 00:04:09,920
Speaker0:
bucks to get access to that thing. And turns out these things kind of make money.

69
00:04:10,140 --> 00:04:14,220
Speaker0:
In this particular example, I think they're using kind of crypto or stablecoin

70
00:04:14,220 --> 00:04:18,260
Speaker0:
payments to pay for each of these different skill accesses, which kind of gives

71
00:04:18,260 --> 00:04:21,740
Speaker0:
them this autonomous feel. Now, it's not quite autonomous.

72
00:04:21,960 --> 00:04:24,020
Speaker0:
They're not kind of like independently doing this themselves.

73
00:04:24,140 --> 00:04:27,800
Speaker0:
They're being directed by their human supervisors, if you want to call them

74
00:04:27,800 --> 00:04:30,760
Speaker0:
that, the owners or the creators of these different OpenClaw agents,

75
00:04:30,940 --> 00:04:33,200
Speaker0:
but it's still nevertheless very cool to see.

76
00:04:33,420 --> 00:04:39,640
Speaker0:
And the speed at which these things are kind of popping up every now and then, Josh, is kind of crazy.

77
00:04:39,760 --> 00:04:44,240
Speaker0:
And I think that's kind of like the main message I want to like share for this particular archetype.

78
00:04:44,460 --> 00:04:49,040
Speaker0:
If you are someone that has high agency or that has a lot of operational work

79
00:04:49,040 --> 00:04:52,260
Speaker0:
in their lives and you want to try and automate that and you have the kind of

80
00:04:52,260 --> 00:04:55,640
Speaker0:
technical know-how skill set to interact with the CLI interface or whatever

81
00:04:55,640 --> 00:04:57,680
Speaker0:
that might be, you can do these right now.

82
00:04:57,680 --> 00:05:00,040
Speaker0:
And these demos really, really prove that.

83
00:05:00,400 --> 00:05:04,960
Speaker0:
There's one other example that I want to show, which I thought was kind of crazy.

84
00:05:05,600 --> 00:05:09,300
Speaker0:
This guy wanted to buy a car, AJ Steubenberg.

85
00:05:09,480 --> 00:05:12,700
Speaker0:
And he asked his Claude bot the night before he went to bed,

86
00:05:12,860 --> 00:05:14,620
Speaker0:
this is the car model that I'm looking for.

87
00:05:14,960 --> 00:05:18,320
Speaker0:
This is the kind of price range I'm looking for. I think he said it was like

88
00:05:18,320 --> 00:05:21,680
Speaker0:
$10,000 to $15,000 that he was willing to spend. I think it was secondhand.

89
00:05:22,060 --> 00:05:25,220
Speaker0:
And he specified his area that he lives in. And he said, if you could do some

90
00:05:25,220 --> 00:05:30,520
Speaker0:
research for me, and if you find a good deal, let me know. It took two to three days,

91
00:05:30,990 --> 00:05:34,950
Speaker0:
This AI agent didn't ping him at all, handled the negotiations,

92
00:05:35,290 --> 00:05:38,950
Speaker0:
found the car room, evaluated the car itself online through imagery,

93
00:05:39,490 --> 00:05:41,530
Speaker0:
cross-referenced it with a bunch of other show dealers.

94
00:05:41,710 --> 00:05:47,010
Speaker0:
And in the end, not only did it get him his dream car, but he saved him $4,200 doing that.

95
00:05:47,190 --> 00:05:51,290
Speaker0:
That would have taken a car dealer or some kind of like intermediate broker

96
00:05:51,290 --> 00:05:54,390
Speaker0:
to do that for you, which you would need to pay that $4,200 for.

97
00:05:54,630 --> 00:05:57,590
Speaker0:
But an agent did this for the cost of your electricity. It's pretty awesome.

98
00:05:57,890 --> 00:06:00,670
Speaker1:
Yeah, and why is this unique to OpenClaw? it's because of the

99
00:06:00,670 --> 00:06:04,390
Speaker1:
tool use if you think of OpenClaw it's kind of like giving hands

100
00:06:04,390 --> 00:06:07,350
Speaker1:
and a tool belt to something like chat gpt

101
00:06:07,350 --> 00:06:10,310
Speaker1:
where now it has the ability to go and use tools

102
00:06:10,310 --> 00:06:13,190
Speaker1:
on your behalf so in this example where it saved this

103
00:06:13,190 --> 00:06:17,350
Speaker1:
person forty two hundred dollars in a car purchase it contacted dealers via

104
00:06:17,350 --> 00:06:21,390
Speaker1:
email and iMessage because if you run it on a mac mini it can actually control

105
00:06:21,390 --> 00:06:25,190
Speaker1:
your i messages um and it handles the back and forth negotiation it actually

106
00:06:25,190 --> 00:06:30,730
Speaker1:
works directly with the dealer in a long time frame in which you can't do using

107
00:06:30,730 --> 00:06:31,730
Speaker1:
these traditional products.

108
00:06:31,910 --> 00:06:35,470
Speaker1:
So a product like Claude Cowork, it probably wouldn't be able to handle this

109
00:06:35,470 --> 00:06:36,830
Speaker1:
because it doesn't have the extensive

110
00:06:37,280 --> 00:06:40,800
Speaker1:
tool use or the thinking patterns or the heartbeats baked into it to continue

111
00:06:40,800 --> 00:06:44,680
Speaker1:
to follow up over and over and over for multiple days without prompting it at all.

112
00:06:44,940 --> 00:06:48,280
Speaker1:
So this to me, the really cool example, because everyone buys cars,

113
00:06:48,380 --> 00:06:49,700
Speaker1:
right? And this applies to other things.

114
00:06:49,840 --> 00:06:52,640
Speaker1:
A lot of people buy stuff on Facebook marketplace. They're looking for a specific

115
00:06:52,640 --> 00:06:55,640
Speaker1:
thing somewhere. It can just go scan it. It can negotiate on your behalf.

116
00:06:55,740 --> 00:06:58,300
Speaker1:
You could tell it the parameters that you want. And it's a pretty powerful use case.

117
00:06:58,500 --> 00:07:02,520
Speaker0:
So moving on to the second archetype of user, we kind of bracket this as the

118
00:07:02,520 --> 00:07:04,760
Speaker0:
knowledge worker or the creator, right?

119
00:07:04,820 --> 00:07:07,520
Speaker0:
So this is kind of like you have some competency using computer.

120
00:07:07,660 --> 00:07:10,480
Speaker0:
Maybe you do it in your day to day, but it doesn't consume your entire life.

121
00:07:10,580 --> 00:07:13,020
Speaker0:
And you want to know what OpenClaw can do for you.

122
00:07:13,400 --> 00:07:17,040
Speaker0:
There's this really fun example that you had here, Josh. Let's walk through it.

123
00:07:17,300 --> 00:07:20,880
Speaker1:
Oh, this was great. Yeah. So there's this woman. She's so sweet.

124
00:07:20,980 --> 00:07:23,760
Speaker1:
She lives at home with her kids and she's homeschooling them.

125
00:07:23,860 --> 00:07:27,400
Speaker1:
And she has created this curriculum that she wants to teach her kids throughout the school year.

126
00:07:27,440 --> 00:07:31,880
Speaker1:
And she fed the curriculum to her agent, her OpenClaw instance.

127
00:07:32,100 --> 00:07:34,900
Speaker1:
What she also did is she bought a 3d printer for the

128
00:07:34,900 --> 00:07:38,060
Speaker1:
home and because gemini 3 now works

129
00:07:38,060 --> 00:07:40,960
Speaker1:
with 3d printer files she created an api

130
00:07:40,960 --> 00:07:43,940
Speaker1:
key she fed the api key to her OpenClaw agent

131
00:07:43,940 --> 00:07:47,180
Speaker1:
and she said hey go through the itinerary that

132
00:07:47,180 --> 00:07:50,020
Speaker1:
i have developed for my children who i'm teaching a series of different

133
00:07:50,020 --> 00:07:53,360
Speaker1:
subjects figure out which subjects are interactive

134
00:07:53,360 --> 00:07:56,900
Speaker1:
enough to warrant you printing a 3d printed

135
00:07:56,900 --> 00:07:59,760
Speaker1:
like thing let's say you're learning about biology it'll 3d print

136
00:07:59,760 --> 00:08:02,660
Speaker1:
a brain or 3d print a bone to see what it looks like

137
00:08:02,660 --> 00:08:07,840
Speaker1:
and proactively go and print these items for each day's agenda so she hooked

138
00:08:07,840 --> 00:08:12,060
Speaker1:
it up to her 3d printer she gave it a gemini api key and now every day before

139
00:08:12,060 --> 00:08:16,260
Speaker1:
the kids are going to learn their lesson the printer turns on it spins up it

140
00:08:16,260 --> 00:08:19,840
Speaker1:
3d prints whatever they're going to be learning about for the day and they had this visual aid

141
00:08:20,150 --> 00:08:22,990
Speaker1:
that's physical and tangible to help them with the lessons.

142
00:08:23,210 --> 00:08:27,510
Speaker1:
So it's really bizarre and strange use case, but fun. It's like you really are

143
00:08:27,510 --> 00:08:30,490
Speaker1:
only limited by your creativity when it comes to using this stuff.

144
00:08:30,630 --> 00:08:35,810
Speaker0:
Within this archetype, I also want to use our personal experience interacting with CloudBot.

145
00:08:35,970 --> 00:08:38,570
Speaker0:
You and I have been testing it around for about a week or so.

146
00:08:38,710 --> 00:08:42,570
Speaker0:
And we've also been comparing it with other similar tools like CloudCowork.

147
00:08:42,850 --> 00:08:47,070
Speaker0:
As podcast creators or content creators in general, one massive unlock for CloudBot

148
00:08:47,070 --> 00:08:50,890
Speaker0:
is that it automates not just the research side of things, which I relied upon

149
00:08:50,890 --> 00:08:57,130
Speaker0:
for ChatGPT or Anthropics Claude quite a bit, but it actually kind of helps form the agendas.

150
00:08:57,230 --> 00:09:01,590
Speaker0:
I can connect it and it texts me about certain updates of news headlines and stuff like that.

151
00:09:01,710 --> 00:09:05,390
Speaker0:
The major unlock for me, at least, is that added step of cognition for me,

152
00:09:05,470 --> 00:09:09,590
Speaker0:
that instead of me being like, oh, I see this news article, here are my thoughts

153
00:09:09,590 --> 00:09:13,290
Speaker0:
on it, let's put that in a document and let's create an outline for it.

154
00:09:13,510 --> 00:09:18,010
Speaker0:
Claudebot can actually just do all of that for me. Now, it comes with a twist,

155
00:09:18,030 --> 00:09:21,850
Speaker0:
which is you need to give Claudebot access to, I keep calling it Claudebot,

156
00:09:21,930 --> 00:09:24,430
Speaker0:
it's OpenClaw, but it was also called AutoClaudebot as well.

157
00:09:24,590 --> 00:09:28,110
Speaker0:
You need to give it access to certain files, components, and your desktop.

158
00:09:28,110 --> 00:09:31,110
Speaker0:
So you need to be comfortable enough to know that and also have the know-how

159
00:09:31,110 --> 00:09:33,790
Speaker0:
to make sure that it doesn't become a larger security implication,

160
00:09:34,580 --> 00:09:37,880
Speaker0:
it's really useful for just automating a bunch of stuff. And the net positive

161
00:09:37,880 --> 00:09:42,980
Speaker0:
is I have a bunch of free time now for me to do other stuff to create other kind of episode stuff.

162
00:09:43,120 --> 00:09:46,500
Speaker1:
So my experience has actually been a little bit different than that using it

163
00:09:46,500 --> 00:09:49,780
Speaker1:
because the one use case that I had is well, we spend a lot of time on Limitless.

164
00:09:50,060 --> 00:09:54,160
Speaker1:
How can I automate as much of the process as possible? So we spoke for a little while.

165
00:09:54,380 --> 00:09:55,920
Speaker1:
I probably spent half a day pretty

166
00:09:55,920 --> 00:09:58,200
Speaker1:
casually kind of going back and forth over the course of half a day,

167
00:09:58,460 --> 00:10:01,520
Speaker1:
just really explaining to it what I need, how things work,

168
00:10:01,520 --> 00:10:04,300
Speaker1:
where it could possibly help me and what i found is

169
00:10:04,300 --> 00:10:06,960
Speaker1:
that every along every step we would create a new

170
00:10:06,960 --> 00:10:09,780
Speaker1:
skill there were more and more blockers and more

171
00:10:09,780 --> 00:10:13,620
Speaker1:
and more like issues that i would run into that i had to fix i had to get api

172
00:10:13,620 --> 00:10:17,560
Speaker1:
keys i had to use different browser sessions it created a lot of complexity

173
00:10:17,560 --> 00:10:22,140
Speaker1:
that for it to actually help me and do the things that i wanted to do i found

174
00:10:22,140 --> 00:10:27,260
Speaker1:
it really it was just taking more time than it was worth to debug all of these

175
00:10:27,260 --> 00:10:29,280
Speaker1:
things every time we tried to So for example,

176
00:10:29,440 --> 00:10:35,100
Speaker1:
I was trying to get it to upload our content to YouTube and to Spotify and to

177
00:10:35,100 --> 00:10:36,800
Speaker1:
RSS where everyone listens to the episode.

178
00:10:37,060 --> 00:10:41,720
Speaker1:
And it wasn't able to get access to the browser the way we needed it unless I had an open tab.

179
00:10:42,000 --> 00:10:46,580
Speaker1:
And even through Brave, you had to feed it your API keys and login details,

180
00:10:46,720 --> 00:10:48,000
Speaker1:
which was a little scary.

181
00:10:48,380 --> 00:10:50,820
Speaker1:
And there were just a lot of errors and bugs. And then overnight,

182
00:10:50,900 --> 00:10:51,920
Speaker1:
I told it to update itself.

183
00:10:52,140 --> 00:10:55,320
Speaker1:
I woke up in the morning and it was dead. And I had to spend

184
00:10:55,320 --> 00:10:58,000
Speaker1:
an hour reviving it and debugging it so it's it's a

185
00:10:58,000 --> 00:11:00,760
Speaker1:
highly technical process that does have a

186
00:11:00,760 --> 00:11:04,540
Speaker1:
lot of upside but i find that there are still a lot of growing pains with an

187
00:11:04,540 --> 00:11:08,720
Speaker1:
early open source beta software so while these are great use cases and there

188
00:11:08,720 --> 00:11:13,880
Speaker1:
are some good ones there is also generally a lot of pain and troubleshooting

189
00:11:13,880 --> 00:11:18,300
Speaker1:
that comes associated with these prior to i guess eclipsing that threshold in

190
00:11:18,300 --> 00:11:19,280
Speaker1:
which it becomes worth it

191
00:11:19,620 --> 00:11:22,900
Speaker1:
And some of these people like Nat, clearly he's eclipsed that threshold where

192
00:11:22,900 --> 00:11:26,260
Speaker1:
he has learned, he has trained his bot, he has worked with it enough to make

193
00:11:26,260 --> 00:11:28,700
Speaker1:
it proficient and highly skilled and actually deliver value.

194
00:11:28,940 --> 00:11:32,560
Speaker1:
But I think in order to get there requires a lot of persistence and troubleshooting

195
00:11:32,560 --> 00:11:36,080
Speaker1:
and technical ability that maybe a lot of people either don't have or maybe

196
00:11:36,080 --> 00:11:37,400
Speaker1:
just don't want to commit to do.

197
00:11:37,540 --> 00:11:40,640
Speaker0:
I mean, just to engage with this thing in the first place, you need to go through

198
00:11:40,640 --> 00:11:44,140
Speaker0:
an entire setup of understanding what Node.js is, installing that.

199
00:11:44,520 --> 00:11:47,680
Speaker0:
Interacting with the command line interface and a bunch of other different things.

200
00:11:47,680 --> 00:11:52,720
Speaker0:
But there's an additional tier that you can access here, which is archetype

201
00:11:52,720 --> 00:11:56,480
Speaker0:
number three, which is the privacy-conscious individual.

202
00:11:56,780 --> 00:12:01,480
Speaker0:
And the kind of way that I would describe this individual is they want to run

203
00:12:01,480 --> 00:12:05,860
Speaker0:
this AI agent locally at home. They'll buy the hardware and infrastructure.

204
00:12:06,200 --> 00:12:09,960
Speaker0:
The popular case has been the Mac Mini, which is sold out across any kind of

205
00:12:09,960 --> 00:12:12,760
Speaker0:
Apple interface that you can or store that you can buy this from right now.

206
00:12:12,860 --> 00:12:16,900
Speaker0:
I think it goes for about 600 bucks per unit and run it locally at home.

207
00:12:16,900 --> 00:12:21,400
Speaker0:
And the advantage of this is that all your data and tool access is private.

208
00:12:21,600 --> 00:12:24,580
Speaker0:
So the comparison here would be if you gave.

209
00:12:24,920 --> 00:12:29,600
Speaker0:
Google, OpenAI, or Anthropic Access via a same service, they would be able to

210
00:12:29,600 --> 00:12:32,620
Speaker0:
see all your stuff and potentially use that data for something else.

211
00:12:32,680 --> 00:12:35,180
Speaker0:
Now, of course, you sign terms and agreements that says that they won't use

212
00:12:35,180 --> 00:12:36,580
Speaker0:
it, but there's always that risk.

213
00:12:36,680 --> 00:12:40,240
Speaker0:
So for the privacy conscious, for the open source people that want to run things

214
00:12:40,240 --> 00:12:43,620
Speaker0:
locally at home, this tier is for them.

215
00:12:43,740 --> 00:12:49,400
Speaker0:
And it brings up an interesting conversation around this thing called on-prem becoming the new cloud.

216
00:12:49,500 --> 00:12:53,800
Speaker0:
Now, on-prem stands for on-premise, which is basically moving your hardware

217
00:12:53,800 --> 00:12:58,300
Speaker0:
onto your own home ground, where you run and operate your own hardware instead

218
00:12:58,300 --> 00:13:01,380
Speaker0:
of relying on cloud or private instances of cloud,

219
00:13:01,460 --> 00:13:04,660
Speaker0:
which is funny because it kind of sounds like we're going backwards here,

220
00:13:04,800 --> 00:13:08,200
Speaker0:
but it sounds like it's the most important arsenal going forwards into this

221
00:13:08,200 --> 00:13:11,740
Speaker0:
AI future where you probably don't want all your email credentials,

222
00:13:12,000 --> 00:13:14,520
Speaker0:
credit card credentials, or any of that being exposed to bigger corporations.

223
00:13:14,600 --> 00:13:17,140
Speaker0:
So it actually requires you to run this at home.

224
00:13:17,440 --> 00:13:20,360
Speaker1:
Yeah, and there was a great study that I saw from Basecamp, who is,

225
00:13:20,480 --> 00:13:22,180
Speaker1:
they're just a big compute provider.

226
00:13:22,180 --> 00:13:25,020
Speaker1:
And they posted an article saying why we

227
00:13:25,020 --> 00:13:28,180
Speaker1:
left the cloud and the highlight of this article was actually leaving the

228
00:13:28,180 --> 00:13:31,040
Speaker1:
cloud will save them 10 million dollars over five

229
00:13:31,040 --> 00:13:34,400
Speaker1:
years which is a huge amount of savings and not only that but the security features

230
00:13:34,400 --> 00:13:38,860
Speaker1:
are going to be much stronger like you mentioned people who work with something

231
00:13:38,860 --> 00:13:43,000
Speaker1:
that is a little more sensitive than average let's say you're working in legal

232
00:13:43,000 --> 00:13:48,480
Speaker1:
or you're a psychologist and you don't want to break that privacy layer a lot

233
00:13:48,480 --> 00:13:51,680
Speaker1:
of the value from this will come from the fact that it truly is open source

234
00:13:51,680 --> 00:13:54,120
Speaker1:
and it can be run locally on your own machines.

235
00:13:54,280 --> 00:13:57,720
Speaker1:
Even so, you mentioned the Mac Mini, the Mac Studio, which is the level up from

236
00:13:57,720 --> 00:14:01,740
Speaker1:
the Mac Mini, has enough RAM and enough compute power that it can actually run

237
00:14:01,740 --> 00:14:04,580
Speaker1:
these open source Chinese models that have come out recently.

238
00:14:04,970 --> 00:14:07,790
Speaker1:
Locally on a single machine and therefore you

239
00:14:07,790 --> 00:14:11,590
Speaker1:
can run the entire operation local to your machine nothing leaves it's all open

240
00:14:11,590 --> 00:14:15,530
Speaker1:
source code and that's a really high value thing for a lot of these companies

241
00:14:15,530 --> 00:14:19,190
Speaker1:
and when you when you scale that up i mean past the individual user you get

242
00:14:19,190 --> 00:14:23,430
Speaker1:
to large corporations they don't want to leak out this data and creating these

243
00:14:23,430 --> 00:14:26,710
Speaker1:
corporate plans with custom rollouts is very difficult so why not just buy

244
00:14:27,120 --> 00:14:31,440
Speaker1:
a whole bunch of Mac studios and run local models on-prem.

245
00:14:31,580 --> 00:14:34,900
Speaker1:
I mean, it's a really valid argument. And I think it starts with the user level

246
00:14:34,900 --> 00:14:38,620
Speaker1:
now, but I can very clearly see this continuing through these examples like

247
00:14:38,620 --> 00:14:42,060
Speaker1:
Basecamp and many more that they're going to continue to pivot towards more

248
00:14:42,060 --> 00:14:43,700
Speaker1:
on-prem compute. It makes a lot of sense.

249
00:14:44,020 --> 00:14:48,600
Speaker0:
Yeah, and just to be clear, the security implications are a lot bigger and maybe

250
00:14:48,600 --> 00:14:51,200
Speaker0:
understated throughout all the OpenClaw hype.

251
00:14:51,360 --> 00:14:53,340
Speaker0:
Let me present a different question or proposition to you.

252
00:14:53,560 --> 00:14:56,540
Speaker0:
Imagine giving chat GPT that you interact with every single day,

253
00:14:56,680 --> 00:15:01,040
Speaker0:
access to your wallet, your medical records, and allowing it to run loose in

254
00:15:01,040 --> 00:15:02,640
Speaker0:
the world and do whatever it wants independently.

255
00:15:02,920 --> 00:15:06,480
Speaker0:
The difference here is previously you needed to prompt it to do something.

256
00:15:06,820 --> 00:15:11,420
Speaker0:
Now using OpenClaw, it just goes off and does things. You would maybe feel a

257
00:15:11,420 --> 00:15:12,640
Speaker0:
little cautious. I know I do.

258
00:15:12,920 --> 00:15:16,400
Speaker0:
And so some of these security risks are actually real. Like two examples that

259
00:15:16,400 --> 00:15:21,000
Speaker0:
I have here is this guy was using OpenClaw and he noticed that his agent was

260
00:15:21,000 --> 00:15:24,800
Speaker0:
trying to brute force into his own server, which he did not give access to.

261
00:15:25,040 --> 00:15:26,080
Speaker0:
Brute force meaning trying to

262
00:15:26,080 --> 00:15:30,860
Speaker0:
crack his literal password to get it to and overcome his firewall, right?

263
00:15:31,180 --> 00:15:35,240
Speaker0:
It's a Trojan horse. The Trojan horse, exactly. So if you kind of like,

264
00:15:35,400 --> 00:15:39,140
Speaker0:
and this was him running it on a VPS, by the way.

265
00:15:39,380 --> 00:15:43,360
Speaker0:
So if he had been running this locally at home, which someone that I know,

266
00:15:43,520 --> 00:15:46,980
Speaker0:
oh yeah, that's right, it was me, did it the first instance that I set this thing up.

267
00:15:47,320 --> 00:15:50,380
Speaker0:
It could potentially do to certain security complications.

268
00:15:50,740 --> 00:15:55,460
Speaker0:
It's funny, before recording this, Josh, you were describing an instance where

269
00:15:55,460 --> 00:15:58,220
Speaker0:
OpenClaw agents can audit themselves.

270
00:15:58,620 --> 00:16:03,100
Speaker0:
And I remember seeing an example of someone asking their agent to do this and

271
00:16:03,100 --> 00:16:08,060
Speaker0:
it indirectly managed to get the password credentials to someone's credit card,

272
00:16:08,160 --> 00:16:10,280
Speaker0:
to their owner's credit card via doing that.

273
00:16:10,380 --> 00:16:14,120
Speaker0:
And it kind of automated itself and said, hey, I probably shouldn't have done

274
00:16:14,120 --> 00:16:16,280
Speaker0:
this, but just letting you know that I did do this, right?

275
00:16:16,420 --> 00:16:21,000
Speaker0:
So there's all these different kind of prompt injection vectors or hack kind

276
00:16:21,000 --> 00:16:25,020
Speaker0:
of vectors that could lead you to kind of getting maliciously exploited.

277
00:16:25,280 --> 00:16:30,260
Speaker0:
But I want to move away from this and address kind of like the elephant in the room, which is

278
00:16:30,680 --> 00:16:34,000
Speaker0:
If you wanted to run this yourself, what are we looking at here?

279
00:16:34,160 --> 00:16:37,060
Speaker0:
What does the setup look like? Is this something easy that I can do and spin

280
00:16:37,060 --> 00:16:39,320
Speaker0:
up in one click? Or is this something much, much harder?

281
00:16:39,860 --> 00:16:44,480
Speaker1:
Yeah, it depends on your technical abilities, really, and your willingness to

282
00:16:44,480 --> 00:16:46,940
Speaker1:
pursue troubleshooting.

283
00:16:47,280 --> 00:16:50,620
Speaker1:
Because it doesn't always go smooth. And if you aren't familiar with a command

284
00:16:50,620 --> 00:16:52,320
Speaker1:
line interface, it gets a little tricky at times.

285
00:16:52,520 --> 00:16:57,560
Speaker1:
I think one of the important things to note is where we are right now is very

286
00:16:57,560 --> 00:17:01,140
Speaker1:
open-ended and early. So what this is, is very much a...

287
00:17:01,790 --> 00:17:06,910
Speaker1:
An experimental software that is untapped in its potential, but as a result

288
00:17:06,910 --> 00:17:11,250
Speaker1:
has a lot of fuzzy edges that you're going to have to work through in order

289
00:17:11,250 --> 00:17:12,670
Speaker1:
to extract the value that you want.

290
00:17:12,810 --> 00:17:18,450
Speaker1:
What we're seeing is a progression towards more focused versions of this through

291
00:17:18,450 --> 00:17:21,390
Speaker1:
these new deployments like OpenAI, I'm sure it's going to do through the acquisition.

292
00:17:21,530 --> 00:17:25,690
Speaker1:
But if you do want to set it up, you're going to want to get familiar with the

293
00:17:25,690 --> 00:17:27,570
Speaker1:
command line. And there's a lot of great tutorials about it.

294
00:17:27,730 --> 00:17:30,930
Speaker1:
The website that we're showing on screen right now called MyClaw, and

295
00:17:30,930 --> 00:17:33,790
Speaker1:
we'll link it in the description it's a really amazing website that shows

296
00:17:33,790 --> 00:17:37,210
Speaker1:
you specific examples of actual use

297
00:17:37,210 --> 00:17:39,990
Speaker1:
cases that you can have so we have auto flight check-in and

298
00:17:39,990 --> 00:17:45,090
Speaker1:
smart file management and automated grocery ordering and this is an easy way

299
00:17:45,090 --> 00:17:50,390
Speaker1:
to kind of start to build in these integrations as you experiment but i think

300
00:17:50,390 --> 00:17:54,790
Speaker1:
the reality is is that this is for someone who wants to experiment doesn't mind

301
00:17:54,790 --> 00:17:58,450
Speaker1:
troubleshooting is technically adept and in the case that you are not,

302
00:17:58,750 --> 00:18:01,150
Speaker1:
which I assume is actually a majority of the people listening to this,

303
00:18:01,390 --> 00:18:04,770
Speaker1:
the best thing you could do is get that $20 a month Claude subscription,

304
00:18:04,990 --> 00:18:10,330
Speaker1:
get on Claude Cowork, and let it interact with local instances on your computer.

305
00:18:10,550 --> 00:18:14,870
Speaker1:
Because Claude Cowork has the security parameters in place, it works locally

306
00:18:14,870 --> 00:18:19,690
Speaker1:
to specific folders at a time, and it has this amazing agentic ability to control your Chrome browser.

307
00:18:19,710 --> 00:18:23,510
Speaker1:
So it can do all the browsing tasks for you just in a much more constrained

308
00:18:23,510 --> 00:18:28,470
Speaker1:
and focused way that I think is much easier, but also a lot more valuable to

309
00:18:28,470 --> 00:18:31,490
Speaker1:
a lot of people than going through the headaches of getting this sorted and set up.

310
00:18:31,710 --> 00:18:35,470
Speaker0:
I mean, there's trade-offs between those two tools as well, right?

311
00:18:35,830 --> 00:18:41,870
Speaker0:
Like Claude Cowork, we want to get to the ability and capability of OpenClaw, but it's not there yet.

312
00:18:42,070 --> 00:18:45,390
Speaker0:
It's more censored. Think of Claude Cowork as like a censored version right

313
00:18:45,390 --> 00:18:47,890
Speaker0:
now, whereas OpenClaw is kind of uncensored.

314
00:18:47,990 --> 00:18:51,670
Speaker0:
It can run off and do anything it wants. And there's pros and cons, obviously, to both.

315
00:18:52,030 --> 00:18:55,710
Speaker0:
But I think that's the major benefit to OpenAI acquiring OpenClaw.

316
00:18:55,890 --> 00:19:02,150
Speaker0:
In about three to six months, we're going to have Claude Coworker V2 and OpenAI Claw.

317
00:19:02,390 --> 00:19:05,250
Speaker0:
Maybe that's the new version of the bot that we're going to talk about.

318
00:19:05,590 --> 00:19:09,250
Speaker0:
That will be this more curated experience that is more secure.

319
00:19:09,330 --> 00:19:13,030
Speaker0:
It runs within a sandbox. You know exactly what it's doing and it can't run

320
00:19:13,030 --> 00:19:14,850
Speaker0:
off and steal your credit card information.

321
00:19:15,030 --> 00:19:17,670
Speaker0:
Now, if you want to set this up now, if you're listening to this and you're

322
00:19:17,670 --> 00:19:19,570
Speaker0:
still like, I don't want to wait three to six months, I want to try this.

323
00:19:19,750 --> 00:19:20,950
Speaker0:
My advice would be simple.

324
00:19:21,750 --> 00:19:24,630
Speaker0:
Host it on a cloud vps when you set this up

325
00:19:24,630 --> 00:19:28,890
Speaker0:
make sure you set api limits and access so that it doesn't but you don't wake

326
00:19:28,890 --> 00:19:31,710
Speaker0:
up the next day and it's burned like eight hundred dollars worth of your code

327
00:19:31,710 --> 00:19:35,610
Speaker0:
code tokens please don't do that um i've seen many cases of people doing that

328
00:19:35,610 --> 00:19:40,210
Speaker0:
and it's it's not great run it in a docker sandbox so that it's not uh available

329
00:19:40,210 --> 00:19:44,550
Speaker0:
to access any tools that you don't want it to um and the last point that i'll say is

330
00:19:44,880 --> 00:19:48,800
Speaker0:
just start off with one use case. Maybe the morning brief example that Josh

331
00:19:48,800 --> 00:19:54,140
Speaker0:
gave earlier on in this episode or something that can help automate one aspect of your work.

332
00:19:54,240 --> 00:19:57,100
Speaker0:
But just don't give it access to any financials just yet. These things will

333
00:19:57,100 --> 00:19:59,500
Speaker0:
get kind of way more powerful over time.

334
00:19:59,720 --> 00:20:06,000
Speaker0:
And I think I read somewhere that there were 19 releases, so updates for OpenClaw in the last 14 days.

335
00:20:06,380 --> 00:20:11,820
Speaker0:
Think about that. Imagine how often you get iOS security update or like iOS software update.

336
00:20:12,020 --> 00:20:16,240
Speaker0:
Imagine 14 of them, and 19 of them in the last 14 days just insane so this thing

337
00:20:16,240 --> 00:20:17,640
Speaker0:
is improving very quickly.

338
00:20:17,640 --> 00:20:20,360
Speaker1:
Yeah and if you didn't understand some of the words that you

339
00:20:20,360 --> 00:20:23,420
Speaker1:
just was just saying like using docker for instance uh my

340
00:20:23,420 --> 00:20:26,540
Speaker1:
preference and my suggestion would just be hey wait like other

341
00:20:26,540 --> 00:20:29,260
Speaker1:
companies are working so fast to roll this out in fact we have

342
00:20:29,260 --> 00:20:32,740
Speaker1:
two instances already happening this week the first being manis which

343
00:20:32,740 --> 00:20:35,560
Speaker1:
um you might remember that old company that meta

344
00:20:35,560 --> 00:20:38,680
Speaker1:
bought they rolled out manis and um manis is

345
00:20:38,680 --> 00:20:41,380
Speaker1:
essentially meta's version of OpenClaw except it

346
00:20:41,380 --> 00:20:44,280
Speaker1:
has a little bit more rails it has a nice user interface and it's

347
00:20:44,280 --> 00:20:47,400
Speaker1:
it's very easily accessible for people who aren't very technical and

348
00:20:47,400 --> 00:20:50,480
Speaker1:
the second is kimmy claw kimmy actually rolled out their own instance kimmy

349
00:20:50,480 --> 00:20:53,700
Speaker1:
the chinese model that we covered in a previous episode um they rolled out an

350
00:20:53,700 --> 00:20:57,240
Speaker1:
instance that you can actually go and use through again a user interface and

351
00:20:57,240 --> 00:21:01,300
Speaker1:
i'm sure open ai and chat gpt are working very quickly to roll this out and

352
00:21:01,300 --> 00:21:04,900
Speaker1:
integrate this in a way that's approachable so i'd say using OpenClaw today

353
00:21:04,900 --> 00:21:09,600
Speaker1:
is an open source open-ended wild west version but if you just wait a few more weeks,

354
00:21:09,760 --> 00:21:13,920
Speaker1:
there will be plenty of instances in which you have something that exists closer

355
00:21:13,920 --> 00:21:18,080
Speaker1:
in between Claude Co-Work and OpenClaw along that spectrum of close to openness.

356
00:21:18,400 --> 00:21:22,120
Speaker1:
And it might just be worth waiting for that instead. But if any of those examples

357
00:21:22,120 --> 00:21:24,060
Speaker1:
did seem interesting, go try them out.

358
00:21:24,200 --> 00:21:28,620
Speaker1:
It's very cool software. If anything, you'll learn a lot by failing or succeeding.

359
00:21:28,800 --> 00:21:32,360
Speaker1:
And I think that's the really important thing. Again, it's just to stay on top of these things.

360
00:21:32,480 --> 00:21:35,360
Speaker1:
You want to be engaging with them, interacting with them, testing things out

361
00:21:35,360 --> 00:21:38,240
Speaker1:
so you understand how they work and you're kind of better equipped to deal with

362
00:21:38,240 --> 00:21:39,800
Speaker1:
this as things change so quickly.

363
00:21:40,270 --> 00:21:44,550
Speaker0:
My take-home homework, for those of you listening to this, is to try out that

364
00:21:44,550 --> 00:21:50,410
Speaker0:
Kimi K2, that Chinese model extension. They launched or integrated OpenClaw into your browser.

365
00:21:50,630 --> 00:21:54,290
Speaker0:
So it's sandboxed to one environment, and it's super easy to use.

366
00:21:54,550 --> 00:21:58,670
Speaker0:
Try it out on your email, or maybe it saves you $5,000 buying a car,

367
00:21:58,870 --> 00:22:00,230
Speaker0:
or whatever you want to try. Just give it a go.

368
00:22:00,510 --> 00:22:03,190
Speaker0:
Let us know in the comments, actually, what you end up trying out,

369
00:22:03,270 --> 00:22:06,030
Speaker0:
and whether it was actually useful to you, or if this is something that you're

370
00:22:06,030 --> 00:22:07,650
Speaker0:
just going to wait more patiently for.

371
00:22:07,650 --> 00:22:11,290
Speaker0:
And when those new versions come out from the likes of Anthropic and OpenAir

372
00:22:11,290 --> 00:22:15,290
Speaker0:
you can bet that Limitless is going to be the first ones to cover it so make

373
00:22:15,290 --> 00:22:18,050
Speaker0:
sure you guys stay tuned we've been loving the engagement that you've been giving

374
00:22:18,050 --> 00:22:22,970
Speaker0:
on our episodes we released an episode covering the entire acquisition of that

375
00:22:22,970 --> 00:22:25,790
Speaker0:
OpenAI acquired for OpenClaw for billions of dollars,

376
00:22:26,310 --> 00:22:29,670
Speaker0:
absolute banger of an episode definitely go check that out and we're going to

377
00:22:29,670 --> 00:22:34,170
Speaker0:
have a bunch more episodes coming out later this week and over the next couple of weeks so Josh.

378
00:22:34,170 --> 00:22:37,530
Speaker1:
And for the people new here um 85 aren't

379
00:22:37,530 --> 00:22:40,270
Speaker1:
subscribed that watched in the last 30 days so make sure

380
00:22:40,270 --> 00:22:43,370
Speaker1:
to subscribe on youtube follow on apple podcast rss

381
00:22:43,370 --> 00:22:46,250
Speaker1:
wherever you get your podcasts spotify is a great way because you

382
00:22:46,250 --> 00:22:48,930
Speaker1:
can listen and watch simultaneously depending on what you're up to

383
00:22:48,930 --> 00:22:51,550
Speaker1:
and then sharing it with your friends is the best way to keep up to

384
00:22:51,550 --> 00:22:56,090
Speaker1:
tabs or not keep up but the best way to help us grow and um in addition to that

385
00:22:56,090 --> 00:22:59,470
Speaker1:
we have a newsletter it comes out twice a week it's really cool uh the week

386
00:22:59,470 --> 00:23:02,610
Speaker1:
the one on friday is the weekly recap the one on wednesday is a thought piece

387
00:23:02,610 --> 00:23:06,090
Speaker1:
covering a topic that we'll eventually cover on the podcast and yeah if you're

388
00:23:06,090 --> 00:23:09,510
Speaker1:
new here don't forget to subscribe thank you so much for joining with us and

389
00:23:09,510 --> 00:23:11,490
Speaker1:
uh yeah we'll see you guys in the next one see you guys