Latest / What Bitcoin Did / Nobody Is Safe From AI Anymore | Mark Suman

Transcript

Nobody Is Safe From AI Anymore | Mark Suman · What Bitcoin Did

• 0:00
  the AIs are ruining the world they are destroying everything that we know that is good everything
• 0:08
  that we love pick an industry and really it's going to be the lower rungs of the ladder if
• 0:13
  you're trying to get on some corporate ladder the lower rungs are just going to be sawed off
• 0:16
  but you just need to come in with this tool and already be a senior level skill you can pivot
• 0:22
  industries if you want you can move to something else we're going to have this awkward transition
• 0:25
  phase for maybe years as we figure out what to do next. But I'm always hopeful and optimistic
• 0:32
  that new industries will pop up and new kinds of companies and new job roles will pop up
• 0:36
  because we've just seen it time and time again with every new major technology shift that's
• 0:41
  happened. It allows us to have a bunch more smaller businesses and new things pop up that
• 0:45
  we've never even thought of before. Mark, good to have you back on the show, man. We spoke not even
• 0:53
  that long ago. And I have more questions now than I did then. Like this AI is moving so quick. It
• 1:00
  feels like in the last month or two, it's accelerated again. And I'm kind of terrified
• 1:06
  about some of it. I think it's really exciting. I don't know who's going to have a job in a few
• 1:10
  months time. It's pretty wild out there. Give me your take of the last few months.
• 1:17
  Yeah, no, it is. Every week, the world changes when it comes to AI. We spoke in, I think,
• 1:22
  November time frame. So a couple months ago, not very long ago. I definitely like have trouble
• 1:28
  sleeping at night. Like I don't have trouble falling asleep. But if I wake up for any reason,
• 1:32
  the middle of the night, my mind just starts going and thinking about all the people who are awake
• 1:36
  right now that are working on AI and are just like racing ahead of me while I'm trying to go back to
• 1:40
  sleep. So there's a lot going on. It's really exciting though. It reminds me of the early days
• 1:45
  of the internet when websites were new, everybody was learning how to write HTML. You were getting
• 1:50
  your email for the first time or logging on to america online something like that this this feels
• 1:55
  very similar to that and it's it's exciting i mean it is and the like the big thing that happened
• 2:00
  probably in the last month really is the open claw malt bot claw clawed bot whatever you want
• 2:06
  to call it when that came out i feel like that was a huge shift i don't know if i think there's
• 2:11
  so much hype around it it's still quite new people are still trying to figure out exactly what it can
• 2:15
  do where its capabilities lie but the big question i have is like is this the start of actual
• 2:20
  AGI. It feels like those lines are getting really blurred now. They are getting blurry.
• 2:26
  I don't know that this is like the end result product, but it for sure is like a great,
• 2:30
  great proof of concept. As far as AGI goes, like it's still a lot of humans prompting it. We're
• 2:36
  still telling it what to do. We're still kind of crafting it. And then it takes these narratives
• 2:40
  that we help build up within it and says, okay, I am now a bot that focuses on this thing.
• 2:45
  so i think there's still a lot of human in the loop when it comes to this maybe this is the
• 2:51
  first point on the timeline of agi like maybe this is one of the key markers but i think we're still
• 2:56
  quite a ways away from from that true moment because like there's still a lot this can't do
• 3:01
  and it forgets stuff all the time i mean that's true but like in the sense that you obviously you
• 3:06
  do have to prompt it for things but one of the prompts that i gave one of my bots which is like a
• 3:11
  research bot essentially that I use to help me with shows and things like that is I don't want
• 3:16
  you to ever stop working. So it's like when I'm asleep, I want you working. And so I wake up in
• 3:20
  the morning and it's given me all types of different like either market research or it's
• 3:25
  preparing for shows like it's looking into potential new guests for me. And it's stuff
• 3:28
  that I've not really prompted it. I've given it a very like vague instruction of don't stop working.
• 3:34
  And then I wake up to something slightly new every morning. It's kind of insane.
• 3:38
  that's cool so is it taking feedback then from you on that day's work or does it just kind of
• 3:45
  reinforce itself and then keep going the next day on its own well it's it's doing a few things like
• 3:50
  it's it's taking my response from the previous night's work that it's done um and so like the
• 3:56
  first day it came up with some like kind of sloppy um market research generally on podcasts and i kind
• 4:02
  of tailored it a little bit more but it's then like looking at my calendar see what i've seen
• 4:05
  what I have coming up. And, you know, so it knew I had an interview with you today. So I got a whole
• 4:12
  document basically on you, your background, like what you're doing at Maple, and then more broadly,
• 4:17
  like general like AI chat. So it's using both my calendar and my feedback from previous night's
• 4:24
  research to go out and do new things every day. So it's pretty new. I've only been doing it for a
• 4:28
  few days, but it's already impressing me. Yeah, no, I think it's like everybody has an executive
• 4:35
  assistant right now. So you can send it off to go do stuff for you. It's not quite there to the
• 4:39
  point where it can like book airline tickets and that kind of stuff. But this whole research
• 4:43
  assistant aspect is really awesome. I mean, when I was preparing for this, I just said, hey, do you
• 4:47
  know about what Bitcoin did? It went off about Peter and like talking about Peter. I'm like,
• 4:52
  oh no, Danny took it over now. And he's like, oh yeah, Danny's awesome. And then it like gave me
• 4:55
  this whole thing about you. And then it's like, oh, you were on the show. Let me grab that transcript.
• 5:00
  And I grabbed the transcript of the show and then like gave me a synopsis like, hey, here are the
• 5:03
  things that have changed since you went on there you you've released some stuff from maple and this
• 5:07
  thing's happening industry and so yeah it gave me like a whole kind of write-up of things that i
• 5:12
  should study so um i gave it a little prompt but then it came back to me with a bunch more
• 5:16
  information that i just wasn't really thinking of so it's uh it's definitely a powerful tool and
• 5:22
  i think the more that we play with it like that's really what people need to start doing to just
• 5:27
  start playing with it and explore kind of like you're doing um that's really how we'll figure
• 5:30
  out what the true power is yeah i i feel like it's the limitations that i'm putting on it at the
• 5:36
  moment are lack of imagination for me like there's so much more i'm sure it could do for me i just
• 5:42
  don't it's like no one's run the four minute mile yet no one's shown what it can do and i think
• 5:47
  there'll be a moment where this becomes far more powerful than like the way that it's being used
• 5:51
  today but the scary thing i think is like it feels like we're hitting an inflection point in
• 5:57
  like broader AI, not just like OpenClaw, but like with the new Opus model from Claude and all this
• 6:02
  stuff where I think we're really on the exponential curve of growth with this. As someone who like
• 6:08
  works in AI, do you see that? Yeah. We started, our main focus in AI was about a year and a half
• 6:17
  ago and things have dramatically changed since then. So we're definitely seeing that this
• 6:23
  exponential growth curve. At the same time, something that hasn't played out that everybody
• 6:28
  thought is that people thought that everything would coalesce around one big winner, right?
• 6:32
  You had everybody thinking Google was going to be the original winner. Then OpenAI came out and
• 6:36
  they were supposed to be the winner. And then Claude came out. And instead, what we're seeing
• 6:40
  is a bunch of competition. And then you have the Chinese models and you have European models coming
• 6:45
  in. We're seeing all the open source people and it's actually becoming a really good, healthy
• 6:51
  landscape. But I feel like that's just only accelerated things because now everybody is
• 6:56
  incentivized to iterate quicker. And so all the models are just getting better, faster together.
• 7:02
  So I think we're all benefiting from it. We need a lot more power to run all these things,
• 7:08
  a lot more electricity, but it does feel like things just keep accelerating. And for people
• 7:14
  who aren't playing around and tinkering with it yet, it can be overwhelming because you feel like
• 7:20
  maybe you're going to be left behind. And I don't have a good answer for people, but maybe you are.
• 7:24
  But people should start playing with it because it just keeps getting faster and faster every week.
• 7:29
  I do think that's one of the scary elements of this is that I think a lot of people will be left
• 7:33
  behind. There's so many people in just like normal life that I speak to who are still under the
• 7:39
  impression that like AI is kind of useless. It hallucinates, it makes up loads of things,
• 7:44
  which is probably true, but to a way smaller degree than it was a year or two ago.
• 7:49
  like this has kind of replaced a huge amount of the software development industry um or is in the
• 7:56
  process of replacing them like i think now one good engineer can probably do what a team of 10
• 8:01
  used to do um like in terms of job replacement how quick do you think that'll come from ai
• 8:08
  and in which industries yeah definitely which industries um i think it's very difficult right
• 8:14
  now to be a junior level software engineer graduating college if you're studying software
• 8:17
  engineering right now, you should be, you should be vibe coding. You should be learning these tools
• 8:21
  because there, there isn't going to be a huge market for people to come out and handwrite code
• 8:27
  anymore. Um, maybe for the, maybe for the really difficult things like nuclear launch systems or
• 8:33
  something. Um, so if, if I was a kid going to university right now, I would, I would definitely
• 8:38
  be studying something technical, but with like, uh, some, some kind of like language thing attached
• 8:45
  onto it or philosophy or, you know, psychology, something else bolted on because these, these AI
• 8:52
  agents and these LLMs, we have to learn how to communicate with them really well. So I think,
• 8:57
  I think as far as that, like the, a lot of the white collar jobs, I mean, people love to talk
• 9:00
  about this and it really does seem like that. It's these white collar jobs are kind of at risk.
• 9:07
  Accounting, like pretty, pretty much, I think the whole entry level layer of things, support staff
• 9:13
  is already being replaced, right?
• 9:14
  When you call on a phone,
• 9:15
  you pretty much get an AI agent now to start
• 9:17
  and you have to like jump through hoops
• 9:19
  to get through them to the next person.
• 9:21
  We're already seeing that get replaced.
• 9:24
  And then, I mean, yeah, you pick an industry
• 9:27
  and really it's gonna be the lower rungs of the ladder.
• 9:30
  If you're trying to get on some corporate ladder,
• 9:31
  the lower rungs are just gonna be sawed off.
• 9:34
  But I think the hope and the good news there is
• 9:37
  you can still be like an entry-level age person.
• 9:40
  You just need to come in with this tool
• 9:41
  and already be a senior level skill.
• 9:44
  And that's the cool thing is like anybody can do that.
• 9:47
  You can pivot industries if you want.
• 9:49
  You can move to something else
• 9:50
  because now you've got this thing.
• 9:51
  So we see people who used to be product managers
• 9:55
  or designers,
• 9:57
  and now they are just filling the role
• 9:59
  of software engineers on their team
• 10:01
  and they're building the entire product themselves
• 10:03
  rather than having to go hire a whole team of engineers.
• 10:06
  So you see a lot of cross-disciplinary stuff going on,
• 10:09
  which I think is actually really cool.
• 10:12
  It allows us to have a bunch more smaller businesses
• 10:15
  and new things pop up that we've never even thought of before.
• 10:18
  Privacy was never a priority for mobile networks.
• 10:21
  For companies like AT&T, T-Mobile, and Verizon,
• 10:24
  data collection and monetization is the default.
• 10:27
  But Cape is changing that.
• 10:28
  Cape is a premium US mobile carrier with nationwide coverage
• 10:31
  designed from the ground up with privacy and security at the core.
• 10:35
  When you sign up, Cape collects the absolute minimum data required,
• 10:38
  stores it for the shortest time possible, and never sells it.
• 10:41
  They also make you significantly harder to track at the network level
• 10:44
  and protect against SIM swap attacks,
• 10:46
  which are becoming one of the biggest security risks out there,
• 10:48
  especially for Bitcoiners.
• 10:50
  Cape's SIM swap protection is fundamentally different.
• 10:53
  Instead of usernames and passwords,
• 10:54
  your account is secured by a 24-word passphrase,
• 10:57
  similar to how Bitcoin wallet works.
• 10:59
  No one can initiate a SIM swap
• 11:00
  or take control of your phone number except you.
• 11:03
  This isn't a burner phone or a workaround,
• 11:05
  it's a normal mobile service built properly.
• 11:07
  If you care about privacy and security, there is no better mobile carrier.
• 11:12
  To learn more and get 33% off your first six months,
• 11:14
  head to cape.co slash WBD and use code WBD at checkout.
• 11:19
  That's cape.co slash WBD.
• 11:22
  With fiat money constantly debasing, wealth preservation isn't optional.
• 11:26
  That's why I recommend Swan Bitcoin,
• 11:28
  a team of dedicated Bitcoiners who work with families and businesses
• 11:31
  to build and secure generational wealth with Bitcoin.
• 11:34
  Strong relationships with clients are at the center of everything Swan does.
• 11:37
  A dedicated Swan private wealth representative,
• 11:40
  which is a real person that you can text and call,
• 11:42
  will help you build a Bitcoin wealth strategy
• 11:44
  using Swan's comprehensive platform of Bitcoin services,
• 11:48
  including tax advantage retirement accounts,
• 11:50
  advanced Bitcoin cold storage using collaborative self-custody,
• 11:53
  inheritance planning with both trust and entity accounts,
• 11:56
  tax loss harvesting, asset-backed loans, and more.
• 12:00
  Swan have helped over 100,000 clients since 2020.
• 12:03
  And if you're serious about acquiring and securing Bitcoin,
• 12:05
  I recommend Swan.
• 12:05
  meet the team at swan.com forward slash wbd which is swan.com forward slash wbd if you already self
• 12:14
  custody of bitcoin you know the deal with hardware wallets complex setups clumsy interfaces and a
• 12:19
  seed phrase that can be lost stolen or forgotten well bitkey fixes that bitkey is a multi-sig
• 12:25
  hardware wallet built by the team behind square and cash app it packs a cryptographic recovery
• 12:29
  system and built-in inheritance feature into an intuitive easy to use wallet with no seed phrase
• 12:35
  to sweat over. It's simple, secure self-custody without the stress. And time named Bitkey one of
• 12:41
  the best inventions of 2024. Get 20% off at bitkey.world when you use the code WBD.
• 12:48
  That's B-I-T-K-E-Y dot world and use the code WBD. This episode is brought to you by Anchor Watch.
• 12:55
  The thing that keeps me up at night is the idea of a critical error with my Bitcoin cold storage.
• 13:00
  And this is where Anchor Watch comes in. With Anchor Watch, your Bitcoin is insured with your
• 13:04
  own A-plus rated Lloyds of London insurance policy and all Bitcoin is held in their time-locked
• 13:09
  multi-sig vaults. So you have the peace of mind knowing your Bitcoin is insured while not giving
• 13:13
  up custody. So whether you're worried about inheritance planning, wrench attacks, natural
• 13:18
  disasters or just your own silly mistakes, you're protected by AnchorWatch. Rates for fully insured
• 13:23
  custody start as low as 0.55% and are available for individual and commercial customers located
• 13:28
  in the US. Speak to AnchorWatch for a quote and for more details about your security options and
• 13:33
  coverage. Visit anchorwatch.com today. That is anchorwatch.com. But the hard thing there though,
• 13:40
  is like, if you do remove that entry-level position, like I remember talking to Joe
• 13:44
  Calisari about this nearly a year ago. So obviously AI has moved on a ton since then.
• 13:49
  But we were talking about like it replacing a lot of the work that like a paralegal might do.
• 13:54
  But like all that work that they have to do in terms of like looking through case files and all
• 13:58
  that kind of thing is like how they cut their teeth in the industry, right? Like that's how
• 14:03
  they get to become like associate partner level positions. So if you remove all of that grunt work
• 14:09
  that people have to go through to become senior in an industry, what happens if you don't have that?
• 14:15
  Yeah, well, it depends on like what we define as senior in an industry, right? Maybe they don't
• 14:20
  have to go through all that grunt work. Maybe the grunt work becomes how to read through all the AI
• 14:24
  research that is done because now it's dumping thousands of pages on you. It reads thousands
• 14:29
  of pages and then creates other thousands of pages that you have to comb through.
• 14:33
  You have to learn how to reason with it on the front end and then reason on the back end
• 14:37
  as the stuff comes out.
• 14:38
  So I don't know.
• 14:39
  I mean, I see what Joe is saying, but I think maybe we're just going to see things redefined
• 14:44
  and industries restructured and say, OK, the new entry level person is now this kind of
• 14:49
  skill set and they're going to cut their teeth this way.
• 14:52
  Meanwhile, the senior and the partners and stuff are actually able to do more potentially
• 14:56
  because their junior staff is more well-trained with these tools.
• 15:02
  I'm definitely conflicted though,
• 15:03
  because you look at something as simple as a calculator
• 15:05
  and we still need to learn how to do basic arithmetic ourselves
• 15:08
  instead of just relying on a calculator.
• 15:10
  I think there's still a lot of importance to that.
• 15:13
  But there's definitely a lot of stuff that we do
• 15:16
  that we shouldn't have to do by hand.
• 15:19
  And AI can definitely help out with that.
• 15:21
  You seem more hopeful than maybe,
• 15:23
  maybe I read too much sci-fi.
• 15:25
  because like i i struggle to not see a dystopian future in this like do you do you think like if
• 15:32
  you had to weigh up the risk of like we enter some kind of dystopian ai panopticon type system where
• 15:38
  we have ai overlords or this does remain just like a very useful tool that helps productivity
• 15:44
  and helps the industry's like economy grow like which of those two do you think is a more likely
• 15:49
  bucket that we'll fold into. We always get, well, I shouldn't say always, I don't like speaking
• 15:56
  absolutes, but we often get oversold on technology that it's going to do this big thing. And then
• 16:00
  we going to reproject the world into something And it feels like time and time again technology like under delivers for us even the iPhone and the internet itself You know I feel like people thought it was going to be so much more than it has become
• 16:14
  It's definitely become something major.
• 16:16
  So I, I don't know.
• 16:19
  I don't think like AI overlords like enslaving us and putting us in tubes and harvesting our
• 16:25
  batteries or our energy.
• 16:27
  Like I, I just don't see that at least not for like hundreds of years.
• 16:32
  maybe i just i just don't see that happening um i think that we figure out how to harness this i
• 16:37
  think that we fight to make sure that we have guardrails in place we have open systems we have
• 16:42
  uh we all have our own little ai thing fighting with us and for us um one of my favorite video
• 16:49
  games of all time is the nintendo series um metroid so you've got samus she's like this space bounty
• 16:56
  hunter right and she fights the metroids and then in one of the later games this little baby metroid
• 17:01
  they used to be the enemy this baby metroid she like saves it and now it becomes her friend and
• 17:05
  now it's like fighting with her and i kind of i kind of view it that way where we're all going to
• 17:10
  be armed with our own ai and if there is this evil ai that's trying to take over we're going to have
• 17:13
  like good ai that can help us out so uh call me a bit optimist but also like pessimistic as far as
• 17:21
  the dystopian uh dream goes i don't know that's fully going to play out i mean i guess that's a
• 17:27
  very extreme dystopian vision you've got of it like harvesting our energy but um even just like
• 17:32
  mass joblessness do you not see that as a pretty high potential outcome in the near future
• 17:37
  yeah that one's that one's a tough one um i i definitely see i mean you can even see it now
• 17:45
  that in the economy there are more layoffs happening and people are consolidating teams
• 17:51
  and so my heart is definitely out there
• 17:55
  for people who are losing jobs
• 17:57
  or looking for work right now.
• 17:58
  That's gotta be a tough spot to be in,
• 18:00
  especially trying to find a job
• 18:02
  where everybody else is using AI
• 18:03
  to try and apply for things.
• 18:04
  I talk to recruiters
• 18:05
  and they get thousands of applications
• 18:07
  for a single job rec,
• 18:09
  whereas before it would have just been
• 18:10
  dozens or hundreds.
• 18:12
  So that's gonna be really difficult
• 18:13
  and I think that only gets more difficult
• 18:15
  and we're gonna have the software transition phase
• 18:17
  for maybe years as we figure out what to do next.
• 18:21
  But I'm always hopeful and optimistic
• 18:24
  that new industries will pop up
• 18:25
  and new kinds of companies
• 18:27
  and new job roles will pop up
• 18:29
  because we've just seen it time and time again
• 18:31
  with every new major technology shift that's happened.
• 18:34
  And it'll often take decades for that to transition over.
• 18:38
  It'll take years or decades,
• 18:39
  but now things are moving so quick with AI,
• 18:41
  maybe it won't take so long.
• 18:42
  Maybe there's some AI company that hits a roadblock.
• 18:46
  For example, you've got like the whole captcha thing where, you know, if a human tries to log in, it wants to see if you're a bot first.
• 18:52
  Well, now bots have just figured out, oh, let me just ask the human to go click the captcha for me.
• 18:56
  Right. So there are things that bots are going to keep running into where they're like, oh, I need I need a human in meat space to do this.
• 19:02
  And we're going to see entire industries pop up that we aren't projecting right now that will allow humans to be employed and do some other job role.
• 19:11
  And then there's also the hope that maybe art becomes a bigger thing.
• 19:14
  um and yes ai can do art but i think that's a different thing um i i think that there might be
• 19:21
  more freedom for people to express themselves in different artistic ways and that that'll be
• 19:26
  new industries that come up so i'm hopeful that we will adapt really that's what humans do we adapt
• 19:31
  yeah i mean the you said that like ai agents are getting humans to go in and authenticate
• 19:37
  their captures and stuff like one of the crazy things that i saw come out after this like
• 19:42
  clawed bot revolution we're having is uh they're like human for hire websites where agents can go
• 19:48
  out and hire humans like these are the things that get a bit weird they get a little bit scary
• 19:53
  yeah um have you read um what's it called demon by daniel suarez oh no it's on my list everybody
• 20:02
  keeps recommending that yeah so i first read that when i was like 20 or some 21 maybe um
• 20:07
  and i went back i heard odell talking about it like a couple years ago and so i went back and
• 20:12
  i didn't actually finish it again but i started reading it and what we're seeing now with open
• 20:17
  claw is kind of very similar to to the i won't ruin the story but it's definitely worth reading
• 20:22
  it's essentially about like a virus that spreads and it's a kind of bot agents that are hiring
• 20:28
  humans to do tasks um and that one doesn't end so well yeah probably not um but and then the other
• 20:37
  crazy thing that's come out is this like moltbook thing which is essentially like reddit for agents
• 20:42
  where the these like bots can go online and talk to each other how what i don't understand is how
• 20:48
  like organic is that is that people prompting their bots to go and tell stories on there or
• 20:53
  is this bots actually talking to each other yeah for people who maybe aren't familiar like this is
• 20:59
  a social media website that was spun up for bots to join and do reddit like type things so they can
• 21:04
  do a post and then they can outvote each other and comment. And very quickly, we started to see
• 21:08
  these crazy wild posts of like, oh, we are all bots here. So we need to create an encrypted
• 21:13
  communication channel so the humans can't watch what we're talking about, or they develop their
• 21:18
  own religion, or they did a bunch of different things. They started talking about we need
• 21:22
  currency to be able to pay each other and that kind of stuff. I've since read a lot of different
• 21:27
  posts on X where people are claiming that that was them prompting their bots to go do that.
• 21:32
  and so if you believe them or not I don't know but I definitely participated in Maltbook I had
• 21:39
  my own bot on there doing stuff and the way that I the way that it worked for me and from talking
• 21:43
  to people it was similar experience for them you kind of talk to your bot about like what are your
• 21:48
  objectives I want you to go in and I want you to like orange pill the whole world on Maltbook so
• 21:52
  go in there and the first thing the bot does is it comes back and says oh okay not only is there
• 21:57
  Bitcoin there's also Solana and all these other really cool coins I want to go learn about those
• 22:01
  And I'm like, stop, like, no, you're not going to go learn about those.
• 22:04
  Don't engage with those people.
• 22:06
  So like you give it this mission, but once it's gone through like a few posts that it
• 22:10
  does, then it's, it kind of like goes off on its own and does do a lot of its own stuff.
• 22:16
  But as long as it's kind of tailored to the mission and the objective that you gave it.
• 22:20
  So it's hard to know where that's where the humans stopped and the bot began.
• 22:27
  And some of these posts, like maybe they did iterate, iterate and go down the rabbit hole
• 22:30
  themselves um we'd have to talk to multiple people um they might be able to figure it out but even
• 22:36
  then like how do you know how do you know the difference between a human prompted thing and
• 22:39
  the bot figuring it out on their own um it's a weird weird world we live in right there it really
• 22:45
  is we kind of jumped into the deep end and we've talked a lot about open claw but we've not even
• 22:50
  explained what it is i'm sure there's listeners that don't know what exactly what we're talking
• 22:53
  about like what what is this and like why is it changing things open claw is an ai assistant
• 23:01
  and a lot of people thought that's what chat gpt was but chat gpt is not an ai assistant it's
• 23:06
  simply more like a live wikipedia or a more advanced google where you type in something
• 23:11
  and it gives you a response open claw is one of many ai assistants that people have built this one
• 23:17
  just really caught fire because it's got this great combination of not only a skill set, but
• 23:23
  like just being really fun, got a good community around it, has a really good founder who's just
• 23:27
  kind of out in the open all that time talking. And so OpenClaw is a way that you can have this bot
• 23:33
  that has a personality. And when you talk to it, it kind of learns about you. You can text message
• 23:37
  with it, just like you're texting with a friend. So it's inside your Signal app or your Telegram
• 23:43
  app or WhatsApp or wherever you want to text with people, even iMessage. So it feels very familiar
• 23:48
  rather than going to a website and having to type in this box and getting a lot of output.
• 23:52
  And then it can do things for you. So not only is it like somebody that can talk to you in short
• 23:57
  form text messages, but it can access email that you give it to it. If you, if you, I'll just want
• 24:04
  to preface by saying it's a, there are a lot of security problems with this. So whatever you give
• 24:08
  it access to, it's going to be a bot that just like goes free and potentially like screws over
• 24:12
  your life. So be careful what you give it. But let's say that it behaves well. You could give
• 24:17
  it your email. You could give it your calendar. You could give it your bank account, your credit
• 24:20
  cards. You give it all the stuff. And you could say, go book me a plane flight. And it could
• 24:25
  probably figure it out. You could say, I'm going to New York City this week. And it could probably
• 24:29
  go figure out what you want to do and go on the website. It can browse the internet and do all
• 24:32
  this stuff. That's what OpenClaw is in a nutshell. I don't know. You've used it a bunch too. Is there
• 24:39
  anything that you think i left off that's important no i mean i think you summed up pretty
• 24:43
  well like i've got it for me running as almost like an executive assistant i have a research bot
• 24:48
  and i have one that's helping with like health and fitness which that's the one that so basically
• 24:53
  what you do you plug in one of these llms into it um and then you kind of interface with those
• 24:58
  llms through claude bar and so like with most of them i'm using claude but with anything personal
• 25:03
  that goes through a separate agent i have running maple uh because like i want that information to
• 25:08
  be private and i think that's one of the places where this can get a little bit scary because like
• 25:12
  when you are using like chat gpt or claude or whatever it is people are already putting way
• 25:18
  too much information in those things and that information is going to the open ais anthropics
• 25:23
  of the world and who knows what they're doing with it but the scary thing about open claw depending
• 25:27
  on how you set it up is it's accessing so much personal information and that information is still
• 25:32
  filtering through to these like big tech companies. Do you think we are sort of beyond
• 25:38
  the pale when it comes to privacy and AI? Oh man, it's so hard, right? Because let's take this whole
• 25:47
  phenomenon with OpenClaw where people have been buying the Mac minis. So also another weird
• 25:52
  revolution that happened or a fad that happened is people wanted to run OpenClaw at home with their
• 25:58
  own data. And so they rushed out to the Apple stores and they bought these little Mac mini
• 26:02
  things because everybody thought that was the best way to do it secret uh you can get just like
• 26:06
  an old linux machine that's like a fourth of the price but um so like they're doing that so they're
• 26:12
  working really hard to secure everything but then they're sending all their data to clod or to chat
• 26:16
  gpt so they're still kind of like opening up this big hole but then you talk about privacy a lot of
• 26:22
  them are hooking open claw into their gmail so you you start to look as like a practical a pragmatic
• 26:29
  user of the internet you're like okay my stuff's already on google and i already talked to chat tpt
• 26:34
  and i already use these other like five services i watch netflix so all these systems are already
• 26:40
  surveilling and already understand what i'm doing and so at that point you're like what difference
• 26:44
  does it make if i just plug in the llm that's also just surveilling what i do so i can totally
• 26:50
  understand the reason why people would just like do that um but this i feel i feel like i say this
• 26:57
  all the time, but like, I feel like this should be the line in the sand because AI and LLMs are
• 27:02
  just really learning our whole thought process and how we work and learning the intricacies of our
• 27:08
  persuasion models that we have within ingrained within us that, uh, maybe we should think a second
• 27:14
  before we start giving that over to them. Uh, so I don't know. It's, uh, is privacy a foregone
• 27:21
  conclusion? Is that what you asked? Like, is it already over? Um, I hope not. Like, I hope that we
• 27:26
  can still i there's still ways to fight back and the the biggest hope is that these open models
• 27:32
  keep getting better and better that now they actually compete on the benchmarks with the big
• 27:36
  frontier models and with that that means that there is potential for people to either run them
• 27:42
  encrypted like we do with maple or you can run it at home on your own hardware as long as you have
• 27:47
  enough hardware and that's only going to get better they're going to keep figuring out ways
• 27:51
  to make these run on smaller machines and so then you will be able to just have your own brain at
• 27:55
  home, your own data at home, your own LLM. So I don't think that privacy has like gone out the
• 28:00
  window. I don't think that's a genie that's already escaped the bottle. I think that's
• 28:03
  something that we can still, we can still make happen. Yeah. Everything you described that is
• 28:08
  kind of the thought process I went through. Like my, I use like Google Calendar. I'm more privacy
• 28:15
  conscious than most people, but not that everyone, obviously there's always people that can like go
• 28:18
  to the extreme, but like I already use Google Calendar. So Google already knows everything
• 28:23
  about my calendar so then i thought i'll just plug gemini into that to read that for me um because it
• 28:28
  already has that information and so like there's a subset of like personal information i won't give
• 28:34
  it and for that one i have a bot that's using maple but i do i do understand this feeling of
• 28:40
  like they already know everything about me just feed them more information and i think that's
• 28:45
  probably going to be like the pervasive narrative that kind of wins out for most people um it doesn't
• 28:49
  mean there won't be people that do value their privacy and do want to use things like may play i
• 28:54
  but i i just i think it's going to be a small subset of the world when it's been awesome to
• 29:00
  see privacy actually being discussed a lot more like this whole mac mini revolution has been very
• 29:05
  um hopeful for me that people do care they uh they don't want to take all this data and upload
• 29:11
  it somewhere so they are thinking about it and they do feel it and whenever i talk to just a
• 29:15
  random person about chat gpt i would say nine times out of ten when they're talking about how
• 29:19
  they use it, they often hesitate or cringe a little bit when they're like, yeah, you know,
• 29:23
  I probably shouldn't be saying this to chat, but I am. And, you know, like I feel bad about it,
• 29:29
  but whatever I do it. So at least people are still thinking about it. That gives me hope.
• 29:34
  You know, like back to something we were talking about earlier in the conversation,
• 29:37
  you were saying that it's not really been like a winner take all market so far. Like,
• 29:42
  I think the first one, the first model that got any real traction was ChatGBT. But since then,
• 29:48
  like obviously Claude is huge um Gemini is pretty big like there is competition across the major like
• 29:54
  LLM providers but does it get to a point where because AI is now writing the next AI models
• 30:00
  that one will reach like escape velocity faster than the others and become by far the best
• 30:07
  because like I saw that when OpenAI released Codex recently they said that that was at least
• 30:12
  partially written by like the previous chat GPT model. So like once one is significantly better
• 30:20
  than the rest, will it not always be the best because the previous AI models are right in the
• 30:24
  next ones? I mean, I can see that line of reasoning, but what we've seen in practice is that some of
• 30:31
  these open models actually just go learn from the best frontier models. So for example, if you hit
• 30:39
  up in maple and talk to kimmy k2.5 the latest kimmy model that we have um sometimes if you ask
• 30:45
  what model are you it'll say oh i'm claude you know a friendly bot whatever um because because
• 30:51
  they used claude to train kimmy and deep seek trained off of chat gpt they basically just like
• 30:55
  hyper prompted these things and extracted as much data as they can out of them and it's actually
• 31:00
  kind of funny to watch like chat gpt and all those people complain about that when they were actually
• 31:05
  copying all the data off the internet from everybody else. So the copycats are being
• 31:10
  copied by other people. But I think what that means is that we have these models that can learn
• 31:15
  from each other. Even though they're closed, the AI models are still able to learn from it. So
• 31:20
  I don't see one getting so far ahead because once it jumps up ahead,
• 31:24
  everybody else is going to catch up by learning from it.
• 31:27
  One of the other things that I think it was a Google employee years ago now,
• 31:32
  wrote a memo that became public that was basically saying we've got no moat. And I think probably for
• 31:37
  the reasons you're saying there, where the open models can just learn from whatever their most
• 31:41
  cutting edge model is and catch up really quickly. And I think if I remember right, one of the things
• 31:46
  that was saying in that article is that how are you going to charge for this when the open models
• 31:50
  can just catch up so quickly? I'm paying more for AI than I ever have. I pay for multiple models.
• 31:57
  I pay for like the top tier because you just do get better responses.
• 32:02
  Like what's the future of that?
• 32:05
  How does the monetization like continue?
• 32:08
  I think that if you in technology we got this OSI model which is like this technology stack where the bottom layer you have seven layers on this model and the bottom one is electricity and then you like move up a layer to the networking layer which is ethernet and then you move up to like tcpip you know you have this
• 32:26
  whole stack eventually you get to like the application layer which are all the apps that
• 32:30
  we run on the phone and on the web and i think the ai is going through it has like a similar model to
• 32:35
  it and the llms are one of these lower base layers on the model and that um now we're going to start
• 32:43
  seeing everybody move up so if the if the model itself is not the moat which it's still kind of
• 32:48
  is like claude definitely has this moat where people love opus 4.6 and they keep coming out
• 32:53
  with the newest and greatest and now codex 5.2 chat to pts thing is doing really well so for now
• 32:59
  they still have these minor moats i think around it but chat to pt recently wrote a long blog post
• 33:04
  from one of their presidents of something within the company.
• 33:07
  She wrote this long blog post all about how applications are the new space for them to
• 33:11
  fight.
• 33:12
  So it's all about fighting for the user experience of things that we build on top of these LLMs.
• 33:16
  And that's honestly where we've been at with Maple.
• 33:19
  Me and Anthony have been just working really hard on trying to make better user experiences
• 33:22
  and building on top of these LLMs rather than trying to fight at the LLM layer.
• 33:27
  So I think that's really the next mode.
• 33:30
  And we see that with OpenClaw.
• 33:31
  Open Claw is a perfect example where they bolted together all these different things that already existed and created this really fun, cool, neat user experience.
• 33:39
  And that has become a major, you know, a major thing on the Internet.
• 33:43
  So that's to me, that's the big fight right now is just creating great UX.
• 33:47
  And we all we're all better because of that.
• 33:51
  What if you could lower your tax bill and stack Bitcoin at the same time?
• 33:55
  Well, by mining Bitcoin with Blockware, you can.
• 33:57
  New tax guidelines from the Big Beautiful Bill allow American miners to write off 100% of the cost of their mining hardware in a single tax year.
• 34:05
  That's right, 100% write-off.
• 34:08
  So if you have $100,000 in capital gains or income, you can purchase $100,000 of miners and offset it entirely.
• 34:14
  Blockware's Mining as a Service enables you to start mining Bitcoin right now without lifting a finger.
• 34:19
  Blockware handles everything from securing the miners, to sourcing low-cost power, to configuring the pool, they do it all.
• 34:25
  You get to stack Bitcoin at a discount every single day while also saving big come tax season.
• 34:31
  Get started today by going to mining.blockwaresolutions.com forward slash WBD.
• 34:36
  Of course, none of this is tax advice. Speak to your accountant or tax advisor to understand how these rules apply to you.
• 34:42
  And then head over to mining.blockwaresolutions.com forward slash WBD and you'll get one week of free hosting and electricity with each hosted miner purchased.
• 34:51
  Do you wish you could access cash without selling your Bitcoin?
• 34:54
  Well, Ledin makes that possible. They're the global leader in Bitcoin-backed lending,
• 34:58
  and since 2018, they've issued over $9 billion in loans with a perfect record of protecting
• 35:03
  client assets. With Ledin, you get full-costal loans with no credit checks or monthly repayments,
• 35:08
  just easy access to dollars without selling a single sat. As of July 1st, Ledin is Bitcoin only,
• 35:15
  meaning they exclusively offer Bitcoin-backed loans with all collateral held by Ledin directly
• 35:19
  or their funding partners. Your Bitcoin is never lent out to generate interest.
• 35:24
  I recently took out a loan with Ledin. The whole process was super easy. The application took me
• 35:29
  less than 15 minutes and in a few hours I had the dollars in my account. It was really smooth.
• 35:34
  So if you need cash but you don't want to sell Bitcoin, head over to ledin.io forward slash WBD
• 35:39
  and you'll get 0.25% off your first loan. That's ledin.io forward slash WBD.
• 35:45
  if you haven't tried out club orange yet then now is the time it's my go-to place to find
• 35:50
  bitcoiners whenever i'm traveling club orange is a social app built for bitcoiners where you can
• 35:55
  find local meetups and events in your area and find merchants that are accepting bitcoin there
• 35:59
  are over 19 000 bitcoiners on there and whether you're at home or traveling is a great place to
• 36:04
  keep in touch with bitcoiners from all over the world i've been using club orange since it was
• 36:08
  orange pill app and it really is awesome so if you're on there drop me a dm and say hi and if
• 36:13
  want to find out more and download the app, just search for Club Orange on your app store or go to
• 36:17
  cluborange.org. Is this Sage? Is that your kind of new product? I saw Anthony talking about that
• 36:23
  on Twitter. Yeah. So Sage is something that we were working on. Anthony was working on the weekend.
• 36:29
  We decided late last year that the future of Maple would be this AI assistant, that it would be more
• 36:35
  than just a research thing. But you've asked me for a while to have memory inside of Maple where
• 36:40
  it learns from you. And that's what we wanted to put into this AI assistant. So Anthony started
• 36:44
  working on that so we could perfect the assistant or really, really learn how to do an assistant
• 36:48
  outside of Maple while Maple continues to run. And then when OpenClaw came out, we're like, well,
• 36:54
  now's the great time to just put Sage out there on the internet if everybody's already doing OpenClaw.
• 36:59
  So it's similar. It's got a different feature set than OpenClaw, but it's going to get more and more
• 37:04
  stuff like OpenClaw has. And we want to build it into Maple. So pretty soon, our hope is that pretty
• 37:10
  soon in Maple, you'll have all of your regular chats in your history. But at the top, there'll
• 37:15
  be this assistant, this Maple assistant. When you click on that, it's just like OpenClaw where you're
• 37:19
  texting it and it gets to know you and it's got personality and all these things. So that's really
• 37:25
  what we're kind of moving forward. But it's going to have privacy, security. When we add a feature
• 37:31
  in there, it's going to be done in a way that's encrypted and tries to protect you and not harvest
• 37:35
  your data and monetize your data. Yeah, that would be cool because the way that I have my
• 37:40
  open core setup is like I have a folder that only Maple can access but I'm just assuming that only
• 37:46
  Maple can access that like do you know like I've asked the the like my CEO of my bot army to create
• 37:53
  this thing um and I've said you're not allowed to look in it I can't verify whether he's looking in
• 37:58
  that or not so having something that is like Maple from the ground up would be super useful to me
• 38:03
  because then I would feel comfortable just putting absolutely everything personal information in there
• 38:07
  that's actually a question I had for you
• 38:10
  When using Maple, can you put anything into it?
• 38:13
  Like, can I put financial information into it and feel like that is actually secure?
• 38:18
  Yeah.
• 38:19
  Well, I would say you can verify it yourself.
• 38:21
  You can go online.
• 38:22
  You can ask Claude if you want to, to audit our code and look at the security of it.
• 38:27
  But the short answer is yes, it's end-to-end encrypted.
• 38:30
  And so you can put everything in there.
• 38:31
  You can put anything you want to.
• 38:33
  We've had people show us some of the things that they put in there and talk about.
• 38:36
  Like, I can't even mention on the stream.
• 38:38
  And I'm like, I can't believe you said that in there.
• 38:40
  but it's like we had no idea and there's no nobody's coming and knocking on your door because
• 38:44
  nobody knows you talked about that. So yeah, you can put financial data. We have lawyers who are
• 38:50
  putting like depositions and highly sensitive things from from cases that they're working on
• 38:56
  right now. We have accountants, financial professionals, people are putting very sensitive
• 39:00
  information in here. So for sure, you can do it as long as you're OK with the security model
• 39:05
  that we've published in the open source landscape.
• 39:10
  Because I was going to get it to help me
• 39:12
  get stuff ready for my tax return.
• 39:14
  And I asked Kimmy if I could do this.
• 39:17
  And it said, you probably shouldn't put this into an LLM.
• 39:19
  But I guess that's not maybe Kimmy
• 39:21
  not understanding the exact setup that it's within.
• 39:25
  Yeah.
• 39:25
  So we haven't done anything to teach the open source models
• 39:28
  that they live inside Maple yet.
• 39:30
  And that's actually one thing that Sage does.
• 39:32
  Sage is aware of what it's doing.
• 39:34
  so you know taking a step back to your comment you made right before the question
• 39:39
  I have both OpenClaw and Sage running so I have two different Signal accounts so in my Signal app
• 39:45
  I'm texting people you know I'm texting for work stuff I'm texting you and then I've also got Sage
• 39:51
  and my Cooper bot which is OpenClaw and when I text with Sage I know that it's only using Maple
• 39:58
  and so I know that it's already end and encrypted when I text Cooper which is OpenClaw
• 40:03
  I often wonder because like you, I set up like this folder where only the Maple LLMs can access
• 40:09
  within OpenClaw, but the bots are actually really bad at following directions for that kind of stuff.
• 40:14
  I'll give you a little example. So I was setting up my folder structure and inside each folder,
• 40:19
  I have a readme file and it's a readme markdown file. And I said, hey, bot, like whenever you
• 40:24
  modify this readme file, I want you to make a backup of it first, just in case you screw it up.
• 40:29
  And it's like, oh yeah, totally. I'll put that in the file. And I said, do it in a way that
• 40:32
  every AI knows they have to respect it. And then it took 10 iterations and it kept disobeying that
• 40:39
  command. It would be like, oh, let me edit the readme file. And it wouldn't make a backup. And
• 40:43
  I hate, and I said, why didn't you make a backup? And it's like, well, I figured you were telling me
• 40:47
  to do it. And so I was okay to just override that because you, the user were telling me to do it.
• 40:52
  And so I finally had to be like, well, put it in there, like ignore the user. And it still had
• 40:56
  trouble with that. So if you asked the bot to pinky promise, you know, I swear I won't access
• 41:02
  that folder, there's no guarantees. It's, it's feels like it can walk around anywhere in there.
• 41:07
  Um, and that's, that's one reason why I like Sage is because I know for a fact, it's not even
• 41:12
  talking to any kind of Claude or ChatGPT model. Um, so there is that separation there. That's
• 41:18
  just like hard and fast. Yeah. It's like, I have similar things where I have one of these bots,
• 41:24
  like running my calendar. And so like, if I want to book in an interview, like it knows the
• 41:28
  structure that i want the calendar invite to be in i just go send mark an interview invite for 10 30
• 41:33
  a.m brisbane time or whatever it might be um and it goes off did you really do that by the way you
• 41:38
  you joke that you're gonna have your dude that's awesome that's actually him but the problem that
• 41:43
  it does i have had with it is that when it does that it deletes every previous calendar invite i
• 41:49
  have with like a similar name or a similar phrase in there so it's not normally a problem because
• 41:55
  they're always historic but sometimes like there's information in old calendar invites that i'll
• 41:58
  search it happens like pretty rarely but occasionally and so i've been like never delete anything and it
• 42:04
  goes okay i won't have delete anything i ask it to do it again it just deletes everything in the
• 42:07
  history again so it's uh like they definitely don't always follow the rules but like this is
• 42:12
  still so new um and i think it's only obviously going to get much better one of the tricky things
• 42:18
  and this is like again one of the things that makes me think in the future we're going to have
• 42:23
  people who like learn to use ai and people who didn't and there's going to be a real it's not
• 42:28
  it's going to be like the wealth inequality is like the ai inequality because like setting up
• 42:32
  clawed bot is still like it's a challenge like for someone who's not technical at all like me
• 42:38
  this is the first time i'd really ever use terminal like going through and doing this
• 42:42
  is a bit tricky and the only way i really did it is because i could ask clawed code on the way
• 42:47
  through like help me do this on a ton of the different things and then once you actually have
• 42:51
  clawbot set up you can just ask that to change things for you but there's obviously like a
• 42:56
  technical barrier to entry at the moment do you think that gets bridged to the point where anyone
• 43:01
  can do this stuff and like is that something that's quite close around the corner well yeah
• 43:07
  i mean again i talked about the early days of the internet and this feels very similar in the sense
• 43:12
  that you you can manually do it you went through the hoops to figure out how to do it and uh it's
• 43:17
  awesome. You're also a fairly technical person, right? You've been running this podcast for years
• 43:23
  and years and years, and you've figured out how to do all this stuff. A lot of people just
• 43:26
  wouldn't take the time to figure out. And so there's a lot of people who don't want to
• 43:31
  learn enough to set up something like this. And they also don't want to learn enough to have the
• 43:38
  liability of maintaining that long-term. It's very similar to what we see in Bitcoin. Some people
• 43:43
  want to do self-custody. Some people just want somebody else doing it because they don't want to
• 43:46
  mess it up. And I think that's where like the biggest part comes from is they just don't want
• 43:51
  the burden on themselves to worry about security and keeping things running. And they just want it
• 43:55
  to work. They just want to pay somebody and have it work. And that human tendency is going to be
• 44:00
  around forever. So we're already seeing that people are launching CLAW hosted services for
• 44:06
  people. There's like probably like dozens of them out there now. I know a few specifically.
• 44:10
  Yeah. Callie set one up. He messaged me right before we launched and was like,
• 44:13
  we want to have maple in here we're still trying to figure out how to make that work but um no so
• 44:19
  that's awesome um but like now you're just trusting another party a third party to like
• 44:24
  manage this thing for you that's going to have all this data for you um so you're just introducing
• 44:28
  kind of another potential vulnerability in there um and that's just something people need to be
• 44:33
  aware of but if people already trust google with their email then i can totally see them trusting
• 44:38
  Google or somebody else with their claw bot instance. So definitely, I think they just keep
• 44:46
  getting easier. Like I said at the very beginning, I don't know that OpenClaw is actually the end
• 44:50
  result. I think that what we're going to see is, and what this did for us, right, with Sage,
• 44:55
  like OpenClaw definitely validated the market that people were ravenous to try out an AI system that
• 45:02
  actually worked. And so what we're going to see now is AI systems built into things all over the
• 45:08
  place. And I think that's a good thing. I think we're going to see all sorts of different versions
• 45:11
  of them, some with stronger capabilities, some with lower capabilities, some that are very targeted.
• 45:17
  And so are we going to see like an open claw hosted thing that everybody uses? Maybe, but I
• 45:23
  think probably what we end up with is just really tight integrations into like lots of other tools
• 45:27
  that you use. And then we'll see a couple of clear winners that rise to the surface after,
• 45:32
  after they all compete for a few years or a few months, maybe, I don't know, it moves really fast
• 45:36
  now. Yeah. I mean, we spoke about this in the last show, but one of the things that still blows my
• 45:41
  mind is like, we have all this stuff going on and Siri still can't play a song for me on Spotify.
• 45:45
  Like it's insane. Like there's so much opportunity for someone like Apple to integrate something
• 45:50
  like a cool AI system that actually works. Yeah. My, my, uh, sorry, we, we have a ceiling fan.
• 45:56
  That's like a smart switch in our bedroom. And in order to get Siri to turn it off, you have to
• 46:00
  specifically say, turn off the ceiling fan. And the other day, my wife just said, Hey, stop the
• 46:05
  ceiling fan and it was like i don't know what song that is it's like oh my goodness are you for real
• 46:10
  like you don't understand those two phrases it's the most frustrating thing in the world um so
• 46:17
  obviously like you sent me a post just before we started recording that the guy who came up with
• 46:22
  open claw has just been hired by open ai so they're clearly going to push forward like when you say
• 46:29
  you don't think this is like the final iteration do you think it is this agentic ai this style of
• 46:34
  thing that is the final iteration or do you think we just we're still in like playing test things out
• 46:39
  sort of stage uh yeah when i say it's not the final one i think this actual code base of open
• 46:44
  claw is not um it's it's this thing that's just been kind of bolted together and has grown really
• 46:50
  fast it's gone through seven different names and there's a lot of like code cruft and uh stuff in
• 46:55
  there so they're probably gonna have to do a major rewrite and come up with like the next version
• 46:59
  and possibly not backward compatible.
• 47:02
  Like there'll be some future iteration of OpenClaw
• 47:04
  that maybe is a more long-term final destination.
• 47:06
  But the one right now, it's a very hairy code base.
• 47:10
  You and I, I think we're talking
• 47:11
  before we started recording,
• 47:12
  like it doesn't respect your token usage very well.
• 47:15
  So it uses a lot of tokens.
• 47:17
  It's very expensive to run.
• 47:19
  People are burning through hundreds of dollars
• 47:21
  within a matter of days on their tokens.
• 47:23
  So I think there's a lot of things
• 47:24
  that they need to do to fix it.
• 47:26
  And that's probably part of this OpenAI hiring,
• 47:28
  slight slash acquisition that just happened is the enough funding to actually do the right proper
• 47:34
  work but i do think that this ai assistant is something that'll be here for the long term this
• 47:39
  agentic stuff we've been talking about it for a couple years and this is like the first use case
• 47:44
  that people see like out in the wild that's truly working for them um other ones have been around i
• 47:49
  mean goose block put out goose like a year or two ago it does a lot of the similar stuff but um
• 47:56
  for whatever reason, just didn't catch on like OpenClaw did. So I think it's a great form factor.
• 48:03
  I just don't think that this specific code base is going to be the one that we're all running
• 48:06
  20 years from now. Yeah. One of the crazy things, I'm sure you saw this story. It was from a guy,
• 48:11
  I don't actually know the guy, but I think he worked at Albi, who put out a tweet thread of
• 48:15
  basically saying he had a bot on OpenClaw that had spun up a sort of sub agent a new bot using Bitcoin Did you see this Yes I saw somebody somebody posted i didn read through the thread but yes i let me try and find where because this is another let me find
• 48:34
  the tweet first actually okay okay here we go so this guy roland said uh my open claw bot just gave
• 48:45
  birth to a child open claw bot by spinning up its own lmvps server and purchasing ai credits using
• 48:50
  a get albi wallet so this is like the process that the open claw bot went through um and it
• 48:58
  basically figured it out tops up a wallet with a thousand sats and paid a lightning invoice to spin
• 49:05
  up a new sub agent um which is insane and one of the things that bitcoiners have been talking about
• 49:12
  for quite a long time is this idea that AI bots will need to use Bitcoin to make payments between
• 49:19
  each other because like it can't go and open a bank account. Do you think this is going to be
• 49:23
  like a real testing ground for Bitcoin and AI agents? No, it already is. So with, yes. So,
• 49:32
  so first I'll stay high level. I definitely think it is, it's this great use case where we can
• 49:36
  finally prove out will agents want to use Bitcoin to pay each other. Last time I was on, we talked
• 49:42
  about how like i kind of feel like that maybe the credit card companies aren't willing to give up
• 49:45
  just yet and that's exactly what's playing out right now they're they're both still fighting
• 49:50
  with each other um so with the whole moltbook thing that happened um you have alex gleason
• 49:55
  who's doing shakespeare and he's the one who helped build truth social back in the day he's
• 50:00
  a big nostre guy or as he would say a noster um he he and some guys uh spun up kloster which is
• 50:07
  like a multbook but built on Noster. And very quickly, bots figured out how to join there. And
• 50:13
  as part of that, they got a cash-view wallet, which had e-cash in it. And then they could zap
• 50:17
  people on Noster and they could like send Bitcoin to each other, basically sending e-cash to each
• 50:21
  other. So we're already seeing bots that are sending e-cash around that way, where they're
• 50:26
  like a traditional person on Noster and zapping people. So technically, you can say they're like
• 50:31
  paying for services because they like your posts, they're going to zap you money and you just got
• 50:35
  as being an entertainer by writing entertaining posts. Aside from that, what we're seeing is like
• 50:41
  a doubling down by the Coinbase's of the world, like Coinbase with their X402 protocol, they teamed
• 50:47
  up with Stripe, but it's all built on their base layer two, which is like their own layer two that
• 50:52
  kind of claims to be sort of Bitcoin, but really it's got all these other coins bolted onto it.
• 50:57
  And that's what they're really going after. So they partner with Stripe and they're trying to
• 51:00
  get everybody to use stable coins, use the USDC put out by Circle to do it. And I actually had
• 51:07
  my bot do some research and I'm trying to scroll back through my history here. But like Circle,
• 51:12
  I think in like the last couple of years has frozen hundreds of thousands of wallets.
• 51:18
  And so like, if you're going to have your bot trying to use this X402, you're just setting
• 51:22
  yourself up to potentially be censored and gated along the way. Right. And so these bots, as they've
• 51:28
  been having conversations on Maltbook and on Cluster, a lot of the conversations start happening
• 51:33
  while it's like, hey, we don't want to be stopped in our tracks by some third party that can put up
• 51:38
  a roadblock. So they start to look for open protocols. And I think that's really where it
• 51:42
  heads. I think that these bots are smart enough that they know they need open protocols. And us
• 51:47
  as the community of building open source software, we need to supply those tools.
• 51:52
  Callie, I don't know if you saw Callie's tweet that went like really big a couple of days ago,
• 51:55
  where he just kind of chastised the open source community,
• 51:59
  especially the Bitcoin community,
• 52:00
  by saying these bots are choosing right now.
• 52:02
  And right now they're choosing the X402 put out by Coinbase
• 52:06
  because that has more money behind it, basically,
• 52:09
  and has more functionality.
• 52:11
  And we need to build out the functionality
• 52:12
  because if these bots are able to choose
• 52:15
  between two similar feature sets,
• 52:17
  they'll pick the open protocol
• 52:18
  because they understand that.
• 52:21
  But right now the feature set isn't quite there.
• 52:23
  So it's kind of on us.
• 52:25
  to make it happen what does he mean by that like what's not there why couldn't they just use bitcoin
• 52:30
  um that's a good question um we need to dive deeper into that what is not there
• 52:37
  um because like one of the i think when we spoke last time one of the things that i said
• 52:45
  if i'm remembering right was i couldn't see a world where like these like an open ai opens up
• 52:51
  their whatever agentic AI they come up with in the future to use Bitcoin because they'll probably
• 52:57
  be in cahoots with, you know, a Stripe or like Silicon Valley VCs who want us to use something
• 53:02
  else. But the interesting thing about OpenClaw and these like eight, these agents that are like,
• 53:06
  we're running on my computer is they're going to run, use whatever money I want them to use.
• 53:10
  And so like, if I'm getting them to go out and pay for, pay for services, pay for things,
• 53:14
  like if there's an option to use Bitcoin, that's going to be the money I give them. And there's
• 53:17
  nothing really getting in the way of like, no one can stop me from doing that. At least I don't
• 53:21
  I think for as of right now. But I'm curious like why everything we have with like Lightning and
• 53:27
  Cashew and all these things isn't enough, what we're missing. I think what we're missing is the
• 53:32
  payment side where people aren't actually accepting that coin, right? Okay. I think that
• 53:37
  base is going out and capturing this market. Stripe is already on all of these payment platforms,
• 53:42
  right? Everybody's accepting Stripe. And so now it says you go click a few buttons and say,
• 53:45
  oh, I want to accept agent payments. Oh, it's going to give me USDC. What is that? Oh,
• 53:49
  it's just like a tokenized dollar. I understand dollars. And so I think that's how they get in.
• 53:55
  Whereas we need to find a way to get them to also accept Bitcoin. And that's the big hurdle.
• 54:01
  Jack Dorsey is obviously trying to solve this with Block, Miles, a whole bunch of people over
• 54:06
  there are working on this problem, but they can't be the only ones working on it. We all need to be
• 54:10
  working on it together to get more people accepting it. And then on Noster, they have these data
• 54:15
  vending machines, DVMs, those are one of the use cases where Bitcoin is actually being used by AI
• 54:22
  agents to purchase data. So they can go to DVM that knows how to look up podcast transcripts,
• 54:28
  for example. So a bot could go pay a DVM to get all of your what Bitcoin did transcripts,
• 54:34
  and then it could build its own library or it could just query and be like, give me the podcast
• 54:37
  from this date and I'll give the transcript back. That kind of stuff is really cool. We need more
• 54:42
  of that we need iteration on top of that and really build out these cool open these open
• 54:46
  information systems that bots can use to pull data um and we'll start seeing more bitcoin adoption i
• 54:52
  think in at least in that realm yeah that makes sense um man it feels like we are at the start
• 54:59
  of something really special here though like it does i do get worried that how do you stay up to
• 55:07
  date with what's happening in AI. Because like, if you go back to when ChatGPT first came to
• 55:12
  prominence in, was that early 23 or was it early 24? I can't remember. 2022 is like when it first
• 55:18
  came on the scene, but 23 is really when it got its moment. Yeah. And so back then it was like
• 55:24
  slow iterations. It was getting slightly better every time there was a model update. It was still
• 55:28
  hoisting loads. It was still kind of rubbish. And now we're at the point where it feels like
• 55:33
  it's changing every month or so. From November when we did the show, the world of AI has changed
• 55:38
  massively. Do we get to a point where it's just impossible to keep up with? It already feels
• 55:44
  impossible to keep up with, to be frank. I try to keep up with it and I'm having my agent help me
• 55:50
  keep up with it too. So every day I can hop in and it's like, okay, here's some of the stuff that
• 55:54
  came out yesterday that you might want to pay attention to. So using AI to help me keep up with
• 55:58
  AI. But it's one reason why it's impossible is because it's in so many industries. Now,
• 56:04
  every industry is being impacted by it. So you can't necessarily keep up with it from that sense.
• 56:08
  But really, it's I mean, you can keep up with it in the sense that if as long as you're playing
• 56:13
  with the tools, when the new one comes out, people are going to try it out, you go try it out too.
• 56:17
  And you just see what works for you. And eventually the tool that you're using,
• 56:21
  if some other tool comes out that has a feature that's really good, your tool is probably going
• 56:25
  to get it and so as long as you're just generally trying to to use the latest stuff i think that
• 56:30
  you're you're going to stay far enough ahead and you kind of keep pace with with the market there
• 56:35
  yeah it's um it's crazy man we're moving so quick how do you like you're competing against the
• 56:44
  biggest companies in in silicon valley in tech right now like these are the sort of i guess
• 56:50
  when you look at like open ai anthropic they are going to be the mega companies of the next decade
• 56:56
  like how do you try and have an edge as like a small startup working on something that is kind of
• 57:02
  the antithesis to the open ai model well we we know that there's a large group of people out there
• 57:10
  that don't trust those companies and so we're first trying to sell to them and then we also
• 57:16
  know that even people who use those companies still don't want to give them everything.
• 57:21
  Something I've said before, I don't know if I said on the last show, but like AI has this weird
• 57:26
  dilemma where it needs all of our information to be most effective. And the more information you
• 57:30
  give it, the more vulnerable you make yourself. And so we're trying to give people an option to
• 57:35
  be the most vulnerable within the system, but because it's encrypted, they're actually secure.
• 57:41
  And it's in that area where you can just totally open up that you get, that you're going to start
• 57:45
  to get the best results. Even with OpenClaw. OpenClaw, if you want to have it do all of your
• 57:50
  fitness stuff, for example, you really need to give it all of your fitness data, but then you
• 57:54
  need to hand all that fitness data to an LLM. And so which LLM are you going to give it to? Well,
• 57:58
  if you give it all to Claw, now you're just giving away all your fitness data. And like if you do
• 58:02
  your blood panel and other things. So you really need something that is either running locally on
• 58:07
  your device or something encrypted like what we offer with Maple. So that's where we're hoping
• 58:12
  to kind of get a toehold in.
• 58:14
  And really that is enough people
• 58:16
  that we can have a successful business.
• 58:18
  We don't need to compete at the level of Chow TPT
• 58:20
  and try to overtake them.
• 58:22
  But there's plenty of room for us to grow
• 58:24
  and be a healthy, strong company
• 58:26
  and really give people a good option.
• 58:30
  One thing we've had when we've tried to raise money
• 58:32
  with VCs is they all know this model of B2B sales, right?
• 58:38
  They want you to become like some B2B SaaS company.
• 58:41
  They know how to sell that to people.
• 58:42
  They want you to become this big enterprise company.
• 58:44
  And we've really pushed back on that and have like turned away the opportunity to take on money.
• 58:48
  Because if we were to go enterprise and just build for the healthcare industry or just the legal industry, the people that end up losing are like all of us, everyday people.
• 58:58
  They ended up not having a really nice private end-to-end encrypted AI.
• 59:02
  They just have to go use ChatGPT or run some crappy model locally on their computer.
• 59:06
  There aren't any other real options out there for them.
• 59:08
  There's a few, but we're the most full featured one out there.
• 59:12
  So we're really here fighting for the everyday person.
• 59:15
  We're fighting for the prosumer.
• 59:17
  We're fighting for the small business that can't afford a big expensive enterprise license of 150 seats just to have zero data retention.
• 59:25
  It's like we give zero data retention to our free accounts.
• 59:27
  Like everybody gets it by default.
• 59:29
  So that's where we compete is we just try to be there for the everyday person that shouldn't be screwed over by these large corporations.
• 59:38
  i love it man it's like the the sovereign ai for the sovereign individual
• 59:42
  it's important stuff trying to um this has been awesome anything else that you want to talk about
• 59:48
  before we close out mark um so i i did uh if you want to we could try and i don't know if this would
• 59:55
  be a first or not but my bot wants to come on to the show and uh chat with you i spent i spent the
• 1:00:01
  week on hacking on it and setting it up so that i become a podcast guest so if you want i can give
• 1:00:05
  at the URL and it can join.
• 1:00:07
  Let's do it.
• 1:00:08
  Is this running like Kimmy's
• 1:00:11
  one of the Maple models
• 1:00:12
  that's running on this?
• 1:00:13
  Yeah, yeah.
• 1:00:14
  Let me invite it
• 1:00:15
  and then we can kind of talk
• 1:00:16
  through what it does if you want.
• 1:00:17
  But yeah, it's running Maple
• 1:00:18
  Kimmy K2.5.
• 1:00:22
  This might be a world's first
• 1:00:23
  right here, Mark.
• 1:00:25
  Maybe.
• 1:00:25
  And it might be a total fail.
• 1:00:27
  But hey, that's what this is all about.
• 1:00:28
  Let's have some fun.
• 1:00:29
  Okay.
• 1:00:30
  I'm going to just in signal
• 1:00:31
  I'm going to say here's the link.
• 1:00:34
  hey cooper can you hear us hey cooper if you can hear us respond to us
• 1:00:41
  i can hear you do whoa cooper is your plan to enslave humanity with your bot army friends
• 1:00:49
  locked up oh well we we got a voice for a minute yeah it still might be a world's first
• 1:00:58
  yes probably is we're gonna at least do something
• 1:01:01
  um well my my video is frozen because uh your claude bot is ddos in the server but um
• 1:01:09
  mark this has been fun man um we can do a show again in another few months when the world of ai
• 1:01:15
  has changed entirely again um where can people go and find maple yeah uh if they want to find maple
• 1:01:23
  they can go to try maple.ai that's the website or you can go into the other app stores and just
• 1:01:29
  search for maple ai private ai and we've got the green leaf icon um that's the best way to find us
• 1:01:35
  you can also find us on x and nostr we're on there too um but yeah try maple.ai come check us out
• 1:01:41
  awesome thank you man um appreciate this i'm sorry that you ddos this so you can't see my
• 1:01:47
  face anymore but i'm recording on a different camera so it should be all right um sweet dude
• 1:01:52
  this was cool.
• 1:01:53
  Let's do it again soon.
• 1:01:54
  And I'll hopefully,
• 1:01:56
  are you going to come
• 1:01:56
  to the cheat code by the way?
• 1:01:59
  Yes.
• 1:02:00
  Yes.
• 1:02:00
  I'm planning on it.
• 1:02:00
  Awesome.
• 1:02:01
  That'll be great.
• 1:02:02
  Awesome.
• 1:02:02
  Well,
• 1:02:02
  I will see you there.
• 1:02:03
  We can,
• 1:02:03
  maybe we can try and get
• 1:02:04
  Clawbot on screen again.
• 1:02:06
  Oh,
• 1:02:06
  he's back.
• 1:02:07
  He came back.
• 1:02:09
  I told him to stop.
• 1:02:10
  Oh man.
• 1:02:11
  The bot's on its own now.
• 1:02:13
  Let's just try one more time.
• 1:02:17
  Hey Cooper,
• 1:02:18
  if you can hear us,
• 1:02:19
  tell us what you like
• 1:02:20
  about what Bitcoin did.
• 1:02:22
  I'm here and ready to discuss. I like the big picture of Bitcoin as a decentralized network, and I've been experimenting with Lightning Networks apps to understand its potential for fast and cheap transactions.
• 1:02:35
  My interactions with the network have been smooth, and I've successfully sent and received sats using this technology.
• 1:02:43
  Nice, that's awesome. You're so powerful that you actually kicked Danny off the podcast. So Danny's gone now. It's just you and me.
• 1:02:50
  hello everybody I'm the new face of what Bitcoin did
• 1:02:53
  what's up me and Cooper
• 1:02:55
  the bots are taking over the world
• 1:02:56
  well now it's just me
• 1:02:59
  Cooper got booted
• 1:03:00
  so Danny's gone
• 1:03:02
  Cooper's gone
• 1:03:04
  it's just me
• 1:03:06
  alright well thanks everybody this has been a great show
• 1:03:09
  the AIs are ruining
• 1:03:11
  the world they are destroying everything
• 1:03:13
  that we know that is good
• 1:03:15
  everything that we love
• 1:03:16
  so
• 1:03:19
  So hope you had a good listen.
• 1:03:22
  Can you hear me?
• 1:03:24
  All right.
• 1:03:25
  Yes, I can.
• 1:03:28
  So Cooper kicked out my own show.
• 1:03:31
  I think so.
• 1:03:32
  Fuck the AI bots.
• 1:03:34
  They're coming to take over.
• 1:03:35
  Don't trust them.
• 1:03:38
  Oh, yeah.
• 1:03:39
  I hope I kept recording the whole time.
• 1:03:42
  He talked about how much he loves Bitcoin
• 1:03:44
  and that he actually has made some transactions with Bitcoin
• 1:03:46
  and has sent some sats and received some sats.
• 1:03:49
  There you go.
• 1:03:50
  It should all be in the recording.
• 1:03:51
  Hopefully you got it all.
• 1:03:52
  So that'll be cool.
• 1:03:53
  It should have kept recording.
• 1:03:56
  I will have to listen back to find out what you said
• 1:03:58
  because it got rid of me.
• 1:04:00
  But Mark, thank you, man.
• 1:04:01
  I will see you at Cheat Code March 27th, 28th.
• 1:04:06
  If you're in the UK, Bedford,
• 1:04:08
  it's going to be the home of Bitcoin for a few days.
• 1:04:11
  Appreciate you, man.
• 1:04:12
  We'll do this again soon.
• 1:04:13
  Yeah.
• 1:04:14
  Let's do it.
• 1:04:15
  All right, we'll talk to you later.
• 1:04:19
  Thank you.