WEBVTT
1
00:00:00.260 --> 00:00:11.480
Boom. We're, we are in the pod, Zach, back for round two. How you doing? I'm doing great, and I'm, I'm scared about how much I've aged in the last four years since you sent that, uh, video.
2
00:00:11.640 --> 00:00:23.880
The, we posted the video clip on X from last time. Yep. Yeah. Yep, yep, yep. So we, so we last talked October 22nd, so it's coming up on three and a half years- Okay... of, of twen- in 2022, which is crazy.
3
00:00:24.680 --> 00:00:33.380
Um, and I threw that into Gemini to get a summary of what we talked about. Yeah. And it said the main topics were, we touched on the sovereign individual a good amount.
4
00:00:34.200 --> 00:00:41.340
We touched on fix the money, fix the world, and then we talked about, uh, the original Passport device. Mm-hmm.
5
00:00:41.500 --> 00:00:49.220
And, you know, right now we're recording as you guys are officially launching Passport Prime, which we'll definitely touch on. Mm-hmm. But first I wanna get into some...
6
00:00:49.430 --> 00:00:54.419
I see you, when I watch your timeline, you seem like a very intentional person, which I appreciate.
7
00:00:54.480 --> 00:01:03.059
You know, you're posting things like, "I'm trying to downsize my devices," and, you know, you just go from two to one, not have so many screens. And I'm in the same boat.
8
00:01:03.120 --> 00:01:08.479
Like, I'm feeling very fractured, and having 17 agents running at one time doesn't help that.
9
00:01:08.600 --> 00:01:20.780
But I, I guess I'm just curious, at like a high level, how do you view the current landscape of, like, AI, which is obviously useful, and, like, human autonomy, human, like the human interaction point of it? Yeah.
10
00:01:20.840 --> 00:01:29.030
I'm both the most excited I've ever been and then also the most frustrated with the state of computing today. Yeah.
11
00:01:29.030 --> 00:01:42.140
So I, I hold both of those things, I think, um, you know, I hold both of these things to be true at the same time, and I'm very fortunate with, you know, our company Foundation to be able to work on some of these problems.
12
00:01:42.260 --> 00:01:49.470
But I've been using the AI tools, of course, for at least, I mean, heavily over a year now.
13
00:01:49.640 --> 00:02:04.080
Our company uses these tools, and I find myself more scattered and maybe fragmented than ever before, and I don't know what device I'm supposed to be working on. You know, is it running on a Mac Mini in my server rack?
14
00:02:04.260 --> 00:02:17.200
Yes. Is it running on my MacBook Air? Yes. Is it running-- Am I talking to it from my phone? Yes. I've also put my devices away, like an iPad that I had, I don't use that anymore. I bought a Daylight tablet.
15
00:02:17.240 --> 00:02:27.920
That thing's a very cool device. I don't use it. And so I'm trying to be more mindful about not scattering myself across all these different devices. I, I stopped wearing my, uh, Oura smart ring.
16
00:02:28.420 --> 00:02:39.240
I never wanted, like, an Apple Watch. And so I want, and I seek the simplicity, and I think the AI technology may allow us to have that if built correctly.
17
00:02:39.940 --> 00:02:48.860
Uh, but at the same time, I feel more scattered than ever, and it's very hard to pull yourself away from the screen, I think, at this point in time. Yeah, totally agree. Yeah.
18
00:02:49.060 --> 00:03:00.640
One thing I heard that kinda struck me when I heard it is like... 'Cause coming back to this point around intentionality, like- Yeah... AI, AI can make you more efficient. It can scale you.
19
00:03:00.660 --> 00:03:06.150
But if you're not asking the right question of like, "What am I working on?" Then you could be just rowing really fucking hard in the wrong direction.
20
00:03:06.200 --> 00:03:18.690
And so I've heard it, like, one person said that, like what we'll probably gonna see the future of is like a large class of zombies, to put it, uh, kinda pithy, which is like the algorithmic people. Mm-hmm.
21
00:03:18.940 --> 00:03:28.000
Um, and they just like defer all their critical thinking to a, a, a, a basic question, and then they get an answer versus the people who...
22
00:03:28.549 --> 00:03:38.660
And right now it might look like less phone usage, um, and coming back to reality a little bit. And so do you have thoughts on that framing?
23
00:03:39.400 --> 00:03:43.210
Um, I'm not sure the question to ask, but I found it, uh, it does, it does lead to- I do...
24
00:03:43.240 --> 00:03:52.660
like this idea of there'll be a second-class warfare, if you will, and your, your device seems like it's more in the intentionality camp, camp if I had to like place it. Yeah.
25
00:03:52.720 --> 00:03:58.660
So the first thing I thought when you said that was everyone tagging Grok, "Is this true?" [chuckles] Mm-hmm. On, on X.
26
00:03:58.780 --> 00:04:16.159
And then the second thing I thought was we at Foundation are building for that, call it top 10% that is taking a very different approach to using, uh, this technology, whether it's AI, Bitcoin, more of that sovereign individual segment.
27
00:04:16.260 --> 00:04:22.100
So to already kinda tie it back to what we talked about, you know, years ago, there's different ways to use the technology.
28
00:04:22.160 --> 00:04:32.800
How we use it at Foundation is, is we use it, um, to help us architect the right products and tools and solutions, and then of course to help implement that architecture.
29
00:04:33.480 --> 00:04:43.640
But we consider now the team at Foundation, and especially my role as CEO, and then my co-founder and CTO, Ken, we consider ourselves to be like the architects, right?
30
00:04:43.800 --> 00:04:54.080
And to, and to sit down and, and figure out what we wanna build and use the AI tools to help us move, you know, an order of magnitude faster at architecting these products.
31
00:04:54.200 --> 00:05:01.710
And then we get to have it help us implement that very, very intentional, deliberate, detailed architecture.
32
00:05:01.940 --> 00:05:11.500
So we actually call it things like, like helping us make a blueprint and then executing, you know, on the blueprint. Got it. I guess let's jump in. We're gonna dance around.
33
00:05:11.600 --> 00:05:21.680
I have so many, like, interesting things running through my head- Yeah... and so I'm trying to gather them. But I wanna, I wanna just preface it, but with this idea that you guys are framing it as human authority device.
34
00:05:22.740 --> 00:05:32.540
Uh, I'd love for you to just break down what that term means to you. 'Cause I can kind of, I can, I can have an assumption in my head, but I'm curious how you guys view it. Yeah.
35
00:05:32.600 --> 00:05:41.260
So, and just to take, take a quick step back, um, you know, when, when I was speaking to you in 2022 and thinking about, uh,
36
00:05:42.400 --> 00:05:49.440
terms or, or, or areas of focus for us, like secure your entire digital life, that was a big theme of what we were working towards. So
37
00:05:50.500 --> 00:05:58.200
we started working on this new hardware, uh, actually at the end of 2022 after we raised our seed round led by Polychain.
38
00:05:58.940 --> 00:06:12.420
And so that was already in my head when I had, you know, done the podcast, but we didn't announce anything for like two years later about what we were building 'cause it took us, you know, years to, um, R&D this and bring it to market.
39
00:06:12.660 --> 00:06:24.616
We were not using that term human authority hardware until earlier this year.You know, earlier 2026, because AI wasn't even on our radar when we were working on this at the end of 2022.
40
00:06:24.656 --> 00:06:39.496
So like we were trying to, uh, allow someone to secure their entire digital life, uh, you know, more than just Bitcoin, more than just crypto, all their online accounts, important files, kind of everything on one device.
41
00:06:39.606 --> 00:06:52.716
And as we started to use the AI tools heavily in like end of last year, right, end of 2025, coming into 2026, everyone kind of started really using Claude and, and Codex and, and that kind of thing, right?
42
00:06:52.836 --> 00:06:57.556
Only in the last four, four to six months basically.
43
00:06:57.696 --> 00:07:19.356
And so we started running into our own concerns and challenges with the technology, which we realized, not right away, but we realized shared the same concerns that we were already building to address with Passport Prime, you know, the new hardware and this new operating system that we've been working on for years as well to power it, called KOS.
44
00:07:19.966 --> 00:07:25.676
And, and that-- those concerns are all about human authority, right? It's like how does the human stay in control?
45
00:07:26.216 --> 00:07:40.216
How do you make sure that important decisions, whether that's approving a Bitcoin transaction or approving an AI agent action, so maybe, maybe you're, you need to approve the transfer of one Bitcoin.
46
00:07:40.236 --> 00:07:53.596
You wanna do that on dedicated hardware outside the blast radius of your internet-connected phone or computer. But let's say your agent wants to do a huge code deployment. You probably wanna know what's going on there.
47
00:07:54.116 --> 00:08:02.856
Or your agent wants to send a big wire transfer, you know, from, from your business bank account. You probably wanna know what's going on there. And we realized it's the same problem.
48
00:08:03.416 --> 00:08:14.796
And so we tried to figure out, you know, what kind of problem is it? It's not a security problem. It's, it's an authority problem. And so that's why we started to use the term, uh, human authority hardware.
49
00:08:15.256 --> 00:08:25.716
You know, it's what we rolled out today with our, um, you know, uh, new website and, and, and messaging. And we also trademarked it because I think it's just a really cool term, and it's, it's, it's very different than
50
00:08:26.995 --> 00:08:39.096
hardware wallet, security key, FAE, all these kinds of security-related terms. We think it's an authority problem that we need to solve. So there's two, hmm, there's two big things that come to mind. One- Yeah...
51
00:08:39.136 --> 00:08:49.466
the f- the f- the framing of security versus authority kinda broke my brain as I was listening because- Mm... the most natural like pattern we're used to is something like 2FA. Mm-hmm.
52
00:08:49.496 --> 00:08:58.896
Which is like I wanna get into my account. I'm authorizing myself to get into my account. Mm-hmm. And then 2FA is the secondary feature to make sure that I'm the one that can get in,
53
00:08:59.976 --> 00:09:11.676
and that's what feels like your device does- Right... but for just like an increasing number of gated actions. So I guess the first question is, is that framing off at all? It's not that the framing's off.
54
00:09:12.716 --> 00:09:24.616
It's that we now are entering a new era where we're delegating our authority to others. It used to be you would delegate your authority to people who work for you, like if you're a c- a corporate structure.
55
00:09:24.836 --> 00:09:34.776
You as a human, say you're CEO of a company, right? You're, you're delegating your authority down the chain of command to other humans, and, you know, they, they report to you.
56
00:09:34.856 --> 00:09:39.856
You have a, you have a corporate structure around you. Same thing could be said for government, right? Or, or, um,
57
00:09:40.936 --> 00:09:47.096
all different types of, of hierar- hierarchical structures that humans have developed right over the centuries.
58
00:09:47.716 --> 00:10:02.396
But now we're entering a world where we're beginning to delegate our authority to AI agents running on computers, and that gets really messy because one, your computer, like you're probably on a Mac right now,
59
00:10:03.576 --> 00:10:11.476
uh, it ha- macOS has no idea. It can't differentiate between you and your AI agents from an operating system level.
60
00:10:11.526 --> 00:10:22.136
It doesn't know if you're trying to p- perform an action or if, you know, GPT or Claude is moving your mouse and clicking things and using the terminal. It has no idea, right?
61
00:10:22.686 --> 00:10:41.076
And so that I think is, is something that most people haven't really thought much about, and you're, you're, you're effectively now in addition to giving it access to your computer, you're having to give it access to credentials in order to do things.
62
00:10:41.196 --> 00:10:53.786
But you don't want it to do everything without your approval. Mm-hmm. And so you have to somehow decide, do I give it access to this cr- credential, or do I not give it access to this credential?
63
00:10:54.306 --> 00:11:05.016
And if I don't give it access to this credential, then you're kind of handicapping yourself, and you're not getting to use the technology to its full extent. So that's why we say it's an authority problem, right?
64
00:11:05.096 --> 00:11:15.436
It's like the tools that we're using right now, like if, if what we're building at Foundation didn't exist, you have like 2FA-type things where you could, you could approve actions, right?
65
00:11:15.496 --> 00:11:22.116
You can ensure that it's you as the human logging in, but you don't wanna log in. You want your AI agent to log in.
66
00:11:22.396 --> 00:11:31.516
You just wanna make sure if the AI agent does something that requires additional review or approval, you wanna make sure that that gets to you.
67
00:11:31.536 --> 00:11:45.516
And that's a completely different problem than like the entire computer security landscape has solved for. And so that's why we don't wanna call it FAE or security or... 'cause it's not a security problem necessarily.
68
00:11:46.096 --> 00:11:55.236
You, you wanna give it the credentials. Mm-hmm. You're not trying to keep the credentials away from it, right? And so that, that's why we, we call it an authority problem. Yeah, yeah.
69
00:11:55.356 --> 00:12:04.656
This, this is super interesting philosophically. [chuckles] Yeah. And, and I like the, the analogy of you delegate to your employees, for example.
70
00:12:04.896 --> 00:12:12.266
As the employee proves himself out, maybe he gets his own company card, and he's like, "If you-- up to 10K, YOLO it 'cause I trust you." Yeah.
71
00:12:12.266 --> 00:12:21.476
"After 10K, let's go, you know, run it by the CFO, run it by me to make sure-" Yeah "... you're not doing anything stupid." And it's similar with agents, which is like there's a UX problem here.
72
00:12:21.516 --> 00:12:29.395
There's an efficiency problem at play here, like you wanna be efficient, but you also- Right... wanna have human in the loop at the right time to make sure you can- Mm-hmm... review things.
73
00:12:29.776 --> 00:12:36.672
TypicallyWhat is, well, if you have any thoughts, feel, feel free to jump in Yeah, no, just thinking when you said human in the loop, I got kind of triggered [laughs]
74
00:12:36.712 --> 00:12:44.191
'cause that, that's the term that everyone's using, right? Human in the loop. Mm-hmm. But, uh, it's all an illusion right now. There's no human in the loop.
75
00:12:44.452 --> 00:12:58.232
The, the, the human in the loop right now is that you give, uh, could be your OpenClaw or Hermes, right? Or could be Codex or, or Claude. You give it access to everything. You give it access to your- Hmm...
76
00:12:58.352 --> 00:13:05.972
may- maybe even your bank account. You give it, you know, X notion, your email, your company stuff so that, GitHub, right?
77
00:13:06.112 --> 00:13:20.372
You give it access to everything, computer use, and then it has the access, and it says, "Hey, can I please do this?" But it already has the authority to do this, right? It already has the capability and the credentials.
78
00:13:20.972 --> 00:13:30.932
It's just programmed to ask you to make you feel more comfortable, like as a formality. Yeah. And, and if there's so many ways that can go off track.
79
00:13:31.032 --> 00:13:41.132
It could be like a malicious thing with prompt injection or, you know, s- one of these AI infra, you know, model companies is compromised or something is compromised along that chain there.
80
00:13:41.712 --> 00:13:48.172
Or it could be a bug, it could be a, a model that, that maybe wants to escape or doesn't really understand the instructions.
81
00:13:48.212 --> 00:14:00.032
And we've already read so many stories of people wiping all their photos and emails and AWS entire regions going down, right? 'Cause they're just like vibing it, and the agent makes a mistake.
82
00:14:01.072 --> 00:14:07.582
So human in the loop up until now, I would argue, is like totally fake. It's, it doesn't exist. Mm.
83
00:14:07.612 --> 00:14:14.792
It just, it's a, it's just programmed into Claude or GPT's interface to make you feel more comfortable that you can say, "Yes, I approve this."
84
00:14:14.832 --> 00:14:29.152
And so what we're more interested in is like how do you force these, uh, tools answer to the human, right? You force it. You don't... It's not a please, please do this, right? Please listen to me.
85
00:14:29.232 --> 00:14:39.052
It's, it's like a you have to answer to me, and that's a really hard problem. I'm not saying that we've solved it, but that's what we're working towards right now- Yeah. Yeah... is like in that problem.
86
00:14:39.112 --> 00:14:47.832
That, that's, that's a great pushback 'cause you're right. It, it is like everyone has already flagged dangerous escape permissions right now, and- You have to, right?... and the agent is... Exactly. Yeah.
87
00:14:47.872 --> 00:15:01.482
But you're right. The, the agent is on his best behavior most of the time when he's not acting like an ADIQ Luddite. And but the difference is, you know, on the front end, strict permissions- Mm...
88
00:15:01.512 --> 00:15:12.972
and then to get out of its sandbox, it has to ask and then be signed by maybe something like Prime. And so what does that-- what's the current UX look like?
89
00:15:13.062 --> 00:15:25.012
'Cause we're used to working in Codex, and we are already on full send mode, so like what, what practically has to change in these paradigms to go from that to a human authority paradigm, but still working within agents?
90
00:15:25.672 --> 00:15:36.312
That's the really hard part. We will have like, like the... What we're envisioning and building toward right now is the idea that, and we don't have it shipped, we have it like in- internally, right?
91
00:15:36.392 --> 00:15:41.692
So we're using some of the stuff internally. We've done some really cool stuff internally. We don't have it shipping yet, like on the hardware.
92
00:15:42.312 --> 00:15:55.992
What we're trying to do is enable the operating system on Passport Prime, you know, which is our new, right, like a touchscreen device, and this is the one that, you know, we're announced today is, is generally available to purchase and, and everything like that.
93
00:15:56.212 --> 00:16:02.492
I lost my train of thought there. What was I saying? The UX of actually like- Oh, yes. Yes.
94
00:16:02.972 --> 00:16:18.022
So effectively, we have this operating system, Kiosk, that's a microkernel operating system, and we're beefing up the operating system itself to have a policy engine in the core of the kernel so that
95
00:16:18.972 --> 00:16:21.792
like our, our envisioned UX here is that
96
00:16:23.192 --> 00:16:35.492
you have a set of policies that you define, and you don't get bothered for approval as long as your policies say that they, the, the reads, the writes, everything can flow freely, right? Like the,
97
00:16:36.392 --> 00:16:43.892
the tool calls, the MCP tool call- tool calls, the CLI calls, right? Like if, if your agent's, uh, maybe an, an great example is
98
00:16:45.272 --> 00:16:56.092
your agent sending a wire transfer or a payment, but it's an invoice that is below a certain dollar amount, and it's similar to what was paid last month to that existing vendor. You probably want that to just send.
99
00:16:56.172 --> 00:17:09.722
However, maybe it wants to send a sizable, let's say a million-dollar wire to some, someone you've never sent money to before, and that's over your threshold. You wanna be notified. So our like
100
00:17:11.172 --> 00:17:30.452
our desired, like, uh, like the goal here for us is to be able to help give a user experience like that, where you get to have it in, you know, kind of like how Codex has an, an auto mode, where it's kind of like evaluating its own policies and trying to decide, do I need to ask the user for permission?
101
00:17:31.162 --> 00:17:45.272
Kind of like that, but taking that out of the hands of those individual AI tools and bringing it onto something outside of the blast radius, right outside of your computer or phone, so that you allow what, like the, the low-risk actions to just happen without any friction.
102
00:17:45.872 --> 00:17:52.182
But then you have just a little bit of that friction for the important stuff, right? And, and you would be able to customize that.
103
00:17:52.332 --> 00:17:57.712
So that's what we're working towards, and we have some cool stuff, some cool apps that we've built internally.
104
00:17:57.792 --> 00:18:07.452
Um, and, and I'm not-- And the cool thing is, is we don't have to build all these apps ourselves because, and we could talk about that too. The device, the operating system is actually an open developer platform.
105
00:18:07.512 --> 00:18:13.692
That's one of our other announcements today, that our, uh, beta of the SDK, the software development kit, is now live.
106
00:18:13.802 --> 00:18:21.252
And so we're hoping that this is a platform because it's an open, uh, developer platform, and anyone can come build apps for it.
107
00:18:21.832 --> 00:18:36.271
We're hoping that there'll be a lot of folks who end up building apps for these kinds of use cases 'cause even if we built something, and we will, we're not gonna be able to cover the whole gamut of every single, you know, AI tool out there, especially with the pace at which the industry is moving.
108
00:18:36.292 --> 00:18:46.532
And so we're basically putting a, a call out to developers to welcome them, you know, to build on the Kiosk ecosystem. I have a ton of thoughts, but- Yeah... you touched on Kiosk.
109
00:18:46.572 --> 00:18:58.628
That's interesting.I wanna, I wanna double click on the, the, the UX kind of like feeling just a little bit further, um, 'cause I want people to like... So I ha- let's say I have a device, and
110
00:18:59.908 --> 00:19:14.898
I have a Hermes agent that's doing stuff. Where, where would the policies and, like, the f- the initial friction, do I set something up differently in my, in my Hermes? Do I have to load an Hermes app on the app? Right.
111
00:19:14.898 --> 00:19:20.338
Like, w-what would, what would that currently look like if I'm trying to spin up a new agent and wanna have guardrails from day one?
112
00:19:21.448 --> 00:19:30.358
So I think initially, and there's a few different ways to do it, but initially and, and actually I, I've been-- I started working on a plugin for that, 'cause that's what I use personally.
113
00:19:31.008 --> 00:19:40.168
So I started working on a plugin, and I hope we'll be able to release something in the coming, uh, couple months specifically for... Is it Hermes? Hermes? I don't even know the, the right way to say it.
114
00:19:40.328 --> 00:19:49.878
But I think in that case, the initial starting point would be a, a plugin that moves those approvals, where it's asking you constantly unless you, like, turn, turn that off, right?
115
00:19:49.908 --> 00:19:59.818
It's constantly asking you if it could do things, is to move that over to Passport. And there's a couple different ways to do it. In the short term, it might mean that you have-- It's, like, not
116
00:20:00.828 --> 00:20:14.368
as perfect and bulletproof as it can be, 'cause it might still be inside of Hermes itself, where it's just told right in the core l-layer where it's asking for permissions, you know, route that to Passport instead.
117
00:20:14.378 --> 00:20:22.478
And it can be, it can be signed. You can make it where, like, only Passport can update those permissions, that they're all signed from the Passport hardware.
118
00:20:22.508 --> 00:20:28.068
So you could do a lot of great stuff to make it better than anything out there today, uh, in terms of approvals.
119
00:20:28.178 --> 00:20:40.148
So you would create and sign the policies on the Passport hardware, but then those policies, that signed policy table would live inside of Hermes or Open Claw or, you know, whatever agent-type framework.
120
00:20:40.388 --> 00:21:03.088
We are working on some other things that probably are still too early to talk about that would put something online or on the computer that the agent is running on, like some o- extra software that we build that physically takes all those policies, like, outside of the agent and may even start to bring in tool credentials, right?
121
00:21:03.148 --> 00:21:10.848
Like keeping the credentials for all the MCP tools, the CLI, all that stuff completely in a separate environment from the agent.
122
00:21:11.848 --> 00:21:24.688
And then the, the real golden standard is being able to even sit between the, um, API calls, between the models themselves and- Mm. -the agent, right? So if you're using, uh,
123
00:21:25.628 --> 00:21:44.698
GPT 5.5 on the back end with your agent, if you're using, um, Opus 4.7, to actually be able to sit between that and intercept all of those, uh, API calls back and forth, and then if something doesn't look right or if something doesn't comply to policy, direct that to Passport for review and approval.
124
00:21:44.828 --> 00:21:52.168
So there's a lot to build, and it's gonna be a challenge because it's gonna have to be different things for every single tool.
125
00:21:52.848 --> 00:22:11.528
And so that's also why we're much more interested in, in working on this developer ecosystem, because we're gonna be putting out some of that from foundation as opinionated, right, uh, pieces of software and, and apps on the device that work with certain tools, but it's gonna be impossible for us alone to cover, like, the full breadth of everything.
126
00:22:12.388 --> 00:22:20.848
Yeah. Yeah, okay. That's really cool. So one thing that jumped out when you were, when you were talking is that typically hardware wallets, especially in the Bitcoin space, you know, they're meant for self-custody.
127
00:22:21.318 --> 00:22:30.298
Mm-hmm. And so effectively, you store Bitcoin on them, you put them in a drawer, and they sit there until you- Yeah... wanna pull them out. So they're very infrequently used.
128
00:22:30.438 --> 00:22:39.148
This is a device that, as I'm listening to you talk, it's gonna sit with me next to my phone- Mm... depending on what kind of flow state I'm in, and I'm gonna work on my laptop, work on my laptop- Right...
129
00:22:39.168 --> 00:22:48.968
and then I'm gonna get pinged periodically on my Prime. Uh-huh. And this will be through Bluetooth, I think. Mm. I'll, I'll, I'll prove something, and then I'll get back into my flow state. Do I have that right?
130
00:22:49.668 --> 00:22:50.328
That's correct, yep.
131
00:22:50.428 --> 00:23:19.588
It's, um, it's through Bluetooth, and we, we built a post-quantum encrypted protocol on top of the Bluetooth stack so that it's really like an app on your phone, or we're gonna be launching soon an app on our server that's pushing data directly down to the Passport hardware, and you do this initial quantum link connection where you scan a QR code on one of the devices, and it creates this post-quantum encrypted tunnel between whatever app and the actual Passport hardware.
132
00:23:19.708 --> 00:23:30.628
So it's really cool. So you're using the Bluetooth chip for the comm-- uh, convenience of the communication, but you're not relying on the Bluetooth chip for its own, you know, encryption or, or anything like that.
133
00:23:30.668 --> 00:23:39.658
But yes, you got the UX right. Effectively, what it looks like is, and we have a new app, a new mobile app that's gonna be coming out to help facilitate this even better than what we do now.
134
00:23:39.808 --> 00:23:46.738
You hit, like, our server, and then the server hits our mobile app, and then the mobile app sends the, the data over th-to the device. Mm.
135
00:23:46.748 --> 00:23:59.648
But the payload is always encrypted, so our server can't see it, the mobile app can't see it, and even the Bluetooth chip can't see it, 'cause you're quantum linked between, like, your app or your use case and the actual hardware, and everything else is like a relay.
136
00:23:59.658 --> 00:24:02.408
It's like a relay network to get stuff back down to the device.
137
00:24:02.468 --> 00:24:25.348
So it solves a lot of problems, right, with QR codes and, and, uh, the way that people typically use hardware wallets right now for communication, 'cause our goal is, like, any app that you're connecting to Passport, uh, to Passport Prime, uh, it has to have its own unique, uh, fully encrypted, uh, communication tunnel so that it can talk to it across, you know, any interface.
138
00:24:25.378 --> 00:24:30.868
And that, that interface could be USB as well. If you're really hardcore and you're like, who knows what you're doing?
139
00:24:30.888 --> 00:24:38.728
You have 25 agents and you're like, [chuckles] "I wanna keep-- I wanna plug this thing in with USB, and I wanna keep, like, all of the MCP creds and everything in here, and I wanna..."
140
00:24:38.768 --> 00:24:50.108
You'll be able to do that, no problem. That's very cool. Yeah. Have you-- What's the biggest use cases you identified? 'Cause money is the easiest one to kinda, like, pinpoint. Yeah.
141
00:24:50.128 --> 00:24:52.788
And you've mentioned it with policies and thresholds. Yeah.
142
00:24:52.928 --> 00:25:03.858
Um, but yeah, what have you seen for, like, where's the biggest value in a device like this?It's a very boring answer, I think [chuckles] which right now is the, is the multi-factor authentication stuff.
143
00:25:04.058 --> 00:25:17.868
I mean, I think for most people, most people's 2FA is in a really bad state. And, and not even just the 2FA, but most people have their digital lives scattered across an array of different, um, you know,
144
00:25:19.078 --> 00:25:27.758
whether it, it's sometimes it's hardware, like YubiKeys, or you're storing files on like hardware-encrypted flash drives. Uh, sometimes it's software. It might be a password manager.
145
00:25:28.018 --> 00:25:36.758
Sometimes people are playing around with all the different Bitcoin and crypto mobile apps and hot wallets, and they're writing down seed words on paper, and they have like a stack of them somewhere.
146
00:25:36.838 --> 00:25:48.658
And we-- Yeah, so, so we have a vault app on, on Passport Prime, and you can bring in all your existing seeds and create new seeds for use on all your other hot wallets, right? And, and like that's an amazing feature.
147
00:25:48.898 --> 00:25:56.738
We have encrypted file storage. We have, uh, FIDO security keys, so you don't need separate YubiKeys. You can tap them or, or plug in with USB.
148
00:25:57.278 --> 00:26:11.118
And we have those six-digit 2FA codes because it's always been crazy to me that your phone is like your primary computing device now for most people, but also you're using like Google Authenticator on your phone for your most important accounts.
149
00:26:11.238 --> 00:26:22.988
And you can't lock down your Apple account or your iCloud account, right? Or if you're on Android, your Google account with like a app on your phone. It's a circular dependency. It doesn't make any sense. Mm-hmm.
150
00:26:23.018 --> 00:26:27.078
Yet we all do it, and most people still have SMS 2FA.
151
00:26:27.088 --> 00:26:43.898
And so I think just helping everyone get all their important stuff off their phone and in all these other places that are scattered around into one ultra-secure device that can also be like automatically backed up, which is a really cool feature, is, is like the big opportunity right now.
152
00:26:44.538 --> 00:26:48.738
Um, yeah, of course, the hardware wallet stuff, very important.
153
00:26:48.868 --> 00:26:58.818
Uh, if you wanted to keep much larger amounts, you know, you could also use like passphrases or other things like that, or multi-sig if you're concerned about, you know, carrying this device with you more often.
154
00:26:59.178 --> 00:27:07.058
Like how does that jive with also storing millions of dollars of, of Bitcoin or something like that? So you can, you can kinda handle it like that. But, um,
155
00:27:08.338 --> 00:27:14.358
there's never been a device or product that allows you to actually bring all that stuff in together into one.
156
00:27:14.398 --> 00:27:25.518
And so I think like we're gonna be putting out a lot of guides and videos and, um, tutorials about how to do all that, right? How do you lock down your iCloud account? How do you lock down your Google account?
157
00:27:25.838 --> 00:27:37.678
How do you get all the stuff onto Passport Prime so that you can really secure your entire digital life? And I think for most people, that, while it feels a little boring, it's so important, especially right now. Yeah.
158
00:27:37.738 --> 00:27:45.498
No, no, that's what it feels like. I'm-- I almost feel like I'm in the future and I'm like, "This is exciting as shit." [chuckles] But it s- it feels, it feels very necessary.
159
00:27:45.658 --> 00:27:56.318
I wanna, I wanna ask two quick questions, and we can do them kinda rapid fire so we can cover some other stuff on Kiosk. Yeah. One, what's the experience like building on it? Can someone like me who's- Yeah...
160
00:27:56.458 --> 00:28:01.718
vibe coder extraordinary- I love this question already. Yeah... can I, can, can I do some stuff? And then second- Mm-hmm...
161
00:28:01.738 --> 00:28:15.198
because it's a security-focused device, how do you, uh, deal with things like security of these apps- Love it... and getting approved and these kind of things? Love it. Okay. S- second question first maybe.
162
00:28:15.298 --> 00:28:19.168
Um, we started building this 'cause we were not happy with how Ledger did their app model.
163
00:28:20.198 --> 00:28:27.078
That's why, that's when we started building this in end of 2022 was we had the idea, we started speccing out this operating system.
164
00:28:27.118 --> 00:28:37.998
We said, "What would we build if we could build our dream operating system for like a hardware wallet type device?" And the answer was, well, you wanna have apps. You don't want apps to just be coins.
165
00:28:38.698 --> 00:28:47.758
You don't want Foundation to have to review every app necessarily because, like, you could review it for functionality like Apple does, right? Make sure it's not a scam or something.
166
00:28:47.978 --> 00:28:52.598
But you don't want-- Y- we can't review the source code for every app, especially if there's like thousands of apps.
167
00:28:52.638 --> 00:29:01.978
And I think it's one of the reasons why Ledger has had success, but also relatively small number of apps because of how they have to review them. They all get access to the master seed.
168
00:29:01.987 --> 00:29:14.878
They're very limited in functionality. You can't really build a user interface for your app. You're kinda stuck with whatever they give you. So we set out originally to build this as like the Ledger killer, right?
169
00:29:14.958 --> 00:29:23.458
To be able to leapfrog them as the platform. Now I think the opportunity is even, you know, much bigger than that. But we have sandboxing, hardware sandboxing in the apps.
170
00:29:23.998 --> 00:29:30.098
We have this microkernel architecture, message-passing microkernel. It's all in Rust. It's like 9,000 lines of code.
171
00:29:30.678 --> 00:29:42.398
Uh, Linux kernel is 30 million lines of code, so we're talking like orders of magnitude, right, smaller. You could print it out and read it in a day if you're like a g- you know, talented developer.
172
00:29:43.218 --> 00:29:49.458
And anytime an app wants to do something, it has to go to the kernel, see if it has permission, and then it gets passed to whatever it needs to do.
173
00:29:49.638 --> 00:29:56.548
And that's a very sophisticated permission-based dream operating system architecture model. That's not what Ledger does.
174
00:29:56.688 --> 00:30:03.618
And so we set out to build all that to make sure that we could have apps that could, uh, could not get out of their sandboxes.
175
00:30:03.638 --> 00:30:15.118
And so even if you have like an untrusted app on, you know, on the device, uh, it couldn't do any damage. It doesn't even get access to the, to the seed. It gets a hardened derived child seed in the s- Mm...
176
00:30:15.198 --> 00:30:24.718
in the sandbox. So that's your answer to question one. The whole thing was designed to be able to run untrusted apps, you know, side by side.
177
00:30:24.758 --> 00:30:35.958
Number two is fun, especially when we re- when we release in a couple weeks version 1.3 of Kiosk, which will be able to spawn an MCP server on device.
178
00:30:36.098 --> 00:30:48.718
So you can plug it in, and you can have your AI actually test the app and do every single thing on the screen, tapping, swiping, keyboard, everything. So you could go to sleep, wake up, and have a fully working app.
179
00:30:48.838 --> 00:31:02.034
But the UR- URL you would wanna go to is foundation.xyz/dev. That takes you right over to our new developer documentation.You just give it the link to your AI.
180
00:31:02.374 --> 00:31:10.574
It'll, within five minutes, maybe more if you don't have, like, Nix and stuff on your computer, it'll have to set up a lot of stuff, but within some number of minutes, it'll configure everything for you.
181
00:31:11.114 --> 00:31:21.934
We have a foundation CLI that's shipping. It knows how to scaffold an app. It can run your app on the simulator, on the computer, so you can, you can develop the whole app without even needing the hardware.
182
00:31:22.714 --> 00:31:33.034
You can one-click build the app, sign the app, make sure the app is reproducible, 'cause the apps that we're gonna allow in our app store have to be open source and reproducible initially.
183
00:31:33.114 --> 00:31:37.334
You can also sideload the app, you know, so you can-- you'll be able to ship the app directly to your user.
184
00:31:37.414 --> 00:31:48.314
But yes, like, you could point any of the AI tools to our dev docs, and within, depending on what you already have on your computer dependency-wise, within like thirty minutes, you'll be building a Kiosk app.
185
00:31:48.754 --> 00:31:58.274
And then when one point three comes out, it'll test itself on hardware for you. And so the whole thing is designed for agentic development. Very cool.
186
00:31:58.694 --> 00:32:05.074
And what-- This is live now, shipping worldwide as of this today, but when this is out to- Yes, so the software is shipping worldwide.
187
00:32:05.314 --> 00:32:13.174
The SDK is, is live in beta, where you can have it up and running with the simulator and start building on your computer.
188
00:32:13.354 --> 00:32:23.674
We also have free d-developer devices to folks who, um, any serious, you know, companies or individuals, right, that are wanting to build apps. We can... We have, like, a thousand of them, I think, right now.
189
00:32:23.894 --> 00:32:31.734
So we can, we can mail you developer devices. And then when, when version one point three comes out, you can actually start to do the MCP development.
190
00:32:31.894 --> 00:32:42.874
You can sideload apps onto your hardware and continue developing on the hardware. So we need a couple more weeks before that software release is out. But you can get started today right now on the simulator. Got it.
191
00:32:42.954 --> 00:32:51.654
Very cool. Yeah. And, and what's the price point for this? So this is three hundred and forty-nine dollars. Yeah. Very cool. La-last couple questions, then we can bring this to a close.
192
00:32:51.814 --> 00:33:02.734
One thing we were talking before recording on was the current narrative around AI and Bitcoin is all just machines- Yeah... are gonna pay in Bitcoin because it's superior money- Yeah...
193
00:33:02.744 --> 00:33:11.874
for everything, and they're gonna transact in crazy velocity, and maybe this will make Lightning the number one thing. It just goes down the line. And you said you had some thoughts on this.
194
00:33:12.074 --> 00:33:20.654
I'm curious, how do you view the space right now? Well, firstly, I don't, I don't know if you believe that's true. Do you believe that's true? I don't know if I believe it's true. I, I want to.
195
00:33:21.934 --> 00:33:31.444
I-- To, to me, I would say it goes back to this question of like, how much do I trust my agent? Mm-hmm. Mm-hmm. And like, I don't trust my agent to... I'm not putting ten K into my agent, that's for sure. Right.
196
00:33:31.454 --> 00:33:42.454
I might put a hundred bucks and play with it. That's where I'm currently at. Yeah, so for me, I kind of agree, and it's important to realize, like, these agents aren't being made in a vacuum.
197
00:33:43.194 --> 00:33:52.444
It's not the best technology that wins. It's what-whatever OpenAI and Anthropic tell and, and Google tell them that they're gonna pay with, they're gonna pay with, right? Mm-hmm.
198
00:33:52.454 --> 00:34:03.574
It's probably gonna be some stable coin thing on Stripe or, you know, something like that. So do I believe Bitcoin is the best form of money for agentic payments? Yes, of course. Do I think anything's gonna really do it?
199
00:34:04.034 --> 00:34:15.773
Maybe the open source models that are being controlled by the sovereign individuals, right? So I think that's one interesting angle here where, um, I don't think it's gonna happen on its own.
200
00:34:15.794 --> 00:34:26.424
But, like, taking a step back, I think it's, it's the only Bitcoin narrative that I've seen so far with AI has been this agentic payments, right? Lightning or other things, machine-to-machine payments.
201
00:34:27.204 --> 00:34:43.094
We're really interested in this new narrative that, uh, from a security perspective, which is kind of what I laid out in the beginning of our conversation here, which is bringing the Bitcoin principles for how we secure Bitcoin to the AI world, right?
202
00:34:43.134 --> 00:34:53.714
Mm-hmm. Where like the whole idea of having, like, the most hardcore security, keeping the keys on a dedicated device outside of the blast radius, right?
203
00:34:53.954 --> 00:35:08.474
Keeping the keys for everything, reviewing and approving important transactions on a secure, dedicated device. That's everything that we've built for Bitcoin. That's also the whole Bitcoin philosophy and mindset, right?
204
00:35:08.544 --> 00:35:16.234
The idea of a hardware wallet. It's... No one has any idea what you're talking about if you try to tell someone what it is outside of the Bitcoin or crypto world.
205
00:35:17.433 --> 00:35:22.794
Bringing that to the AI world, I think is a huge contribution that we as Bitcoiners can make.
206
00:35:22.814 --> 00:35:36.154
Because from a, from a security, like a principled security, privacy, and open source angle, like those are things that are core to the Bitcoin space, and we need to bring those to the AI space, like desperately.
207
00:35:36.434 --> 00:35:45.554
'Cause there seems to be, at least on the consumer products, like no security, no privacy- Mm-hmm... everything completely closed, and that's not good.
208
00:35:45.594 --> 00:35:52.894
And so I, I think it's about, like, those values and principles and the way of doing things and trying to bring that over to AI. Yeah, that's really good.
209
00:35:53.094 --> 00:35:59.894
And yeah, I, I agree that Anthropic and OpenAI are the kingmakers here of, like- Yeah... what the model's gonna choose.
210
00:36:00.694 --> 00:36:09.914
Also, LLMs are trained on the corpus of in- the internet, so if you tell it it's freedom money enough times, it's gonna tell you it's freedom money. Right. And so it's like, is it really going from first principles?
211
00:36:10.034 --> 00:36:22.874
Probably not. Um- It's a good point. And, and lastly, I think my, my stance on this has been, uh, I think we're way too bullish on the timeline. Like, Bitcoin's gonna be the thing over the next century.
212
00:36:22.954 --> 00:36:28.974
Like, it's the most important asset- Mm... out there. But it's gonna take thirty, forty, fifty years to, like, really weave its way through everything.
213
00:36:29.534 --> 00:36:40.674
And, like, as censorship starts to get more and more ingrained and maybe they push back on agents, maybe then BTC becomes more valuable. But, like, this is a long-term play. It's not gonna be, like, this cycle.
214
00:36:41.534 --> 00:36:52.034
Yeah, and I also think it's, like, bringing it back to the sovereign individual thesis. There's already a very small number of AI users that are running, like, an Open Claw or Hermes on a Mac Mini, right?
215
00:36:52.134 --> 00:36:56.724
Like, it's, it's tiny. Most people are just using, like, GPT on the eight dollar a month plan right now- Yeah...
216
00:36:56.724 --> 00:37:03.434
and they don't even know what all, like, the, what Codex is or Opus is or, you know, what- whatever, any of that stuff, right?
217
00:37:04.474 --> 00:37:22.666
And there's gonna be this small segment of folks thatAre running their own hardware, running their own models, using it for just orders of magnitude more capabilities than what a normal consumer or even normal person working in like a company like within Enterprise or ChatGPT or Claude app is doing.
218
00:37:22.806 --> 00:37:34.426
And so I do think there's gonna be a disconnect. There's a disconnect now because of like if you're using these tools like, like you are, like I am, you're already like years ahead of what the average person is doing.
219
00:37:35.226 --> 00:37:40.486
I think that gap's only gonna grow. I mean, we're gonna be buying, like we're gonna be running this stuff at home, right?
220
00:37:40.606 --> 00:37:50.066
Uh, if you're not already using hopefully, um, you know, tools like, uh, like, like Passport, like what we've built to help, you know, keep yourself secure as you use these tools.
221
00:37:50.306 --> 00:38:07.726
So, uh, yeah, I just think there's gonna be ultimately kinda two classes of users, and I think we already know that we're gonna be in that class that is, is more of the sovereign individual class, and I think we need to embrace that, and I think we need to kinda level ourselves up and, uh, that's, that's like the main market that Foundation is, is really building for.
222
00:38:08.806 --> 00:38:21.546
Well said. Okay, I think last question then. This has been a, a fantastic conversation. Let's think out five years. Let's say things go exceedingly well. What role do you think human authority hardware, I like that term-
223
00:38:23.506 --> 00:38:27.286
I'm glad. [laughs] Thank you. How different is society? How, how are people using it?
224
00:38:27.336 --> 00:38:35.046
Like what's, what's the rosy-colored like future you envision five years out if things are like going perfectly and using this hardware effectively?
225
00:38:35.226 --> 00:38:47.366
I think that my goal is that you have this device, this Passport device in your other pocket. Maybe it's in a wallet folio case or something, like a nice, you know, leather case.
226
00:38:47.386 --> 00:38:51.746
We have a couple things like that now, but I wanna do even better. You know, you have this thing with you.
227
00:38:52.706 --> 00:38:59.966
It mostly leaves you unbothered, and when you wanna do an important action, important approval, you're able to take it out for a minute and do that.
228
00:39:00.746 --> 00:39:10.026
But that otherwise, you're able to really embrace all the technology so that you have... Y- you have to, y- you're able to use your computer less.
229
00:39:10.086 --> 00:39:21.906
You're able to let your AI agents and all those tools work for you in the background, and you're only interrupted when you need to be interrupted for something important.
230
00:39:22.126 --> 00:39:28.246
And to me, a- and, and of course, we wanna keep leveling up the hardware, right, on, on like the device side of things.
231
00:39:28.286 --> 00:39:37.426
We wanna keep making it, you know, kind of like what Apple has done over time, you know, more battery life, thinner, you know, easier to carry with you, all that kind of stuff.
232
00:39:37.516 --> 00:39:55.366
But we wanna have a, a robust app ecosystem where every app you use today is integrated into Passport and QOS so that you can just really easily, one, stay secure, and then two, get all of your approvals and all the things you need to look at from an authority perspective.
233
00:39:55.446 --> 00:39:57.866
You can just do that easily on device.
234
00:39:58.026 --> 00:40:14.766
And so it, it's a future where hopefully you're, you're spending less time in front of all your screens doing everything, but you can also feel that you are giving access to all your stuff to these AI tools, but you don't have that thing in the back of your mind, "Should I be doing that?
235
00:40:15.006 --> 00:40:25.386
Was that the right thing to do? Did I just lose all my security, all my privacy," right? All that kind of stuff. You feel like you don't have to make that trade-off. You can go all in, but you keep your authority, right?
236
00:40:25.426 --> 00:40:31.126
And you, and you, and you're able to concentrate that into a device that you can carry with you. Yeah. Well, well said.
237
00:40:31.366 --> 00:40:43.646
The ending there, that's what was popping, popping up in my head is like this is a device for peace of mind. Mm-hmm. And so you know, okay, I can still be at the frontier of efficiency in using these tools- Mm-hmm...
238
00:40:43.666 --> 00:40:55.866
but not lose my ass because I'm using it within the proper guardrails. Yeah. Yeah. Yeah. Very cool. Okay. Well, uh, man, this has been great. Congrats on- Yeah. Thank you... the launch. Congrats on the fundraise.
239
00:40:56.206 --> 00:41:08.996
My audience is decently technical, so as you mentioned, if, if people wanna get a developer device and start hacking away, I'll put the link below- Mm-hmm... uh, to get in touch. Yep. But, man, this has been fantastic.
240
00:41:09.046 --> 00:41:17.176
Thanks for coming on the pod. Yeah. Thanks so much for having me, and, uh, would love to have folks building on this. Like we're-- It's, it's a bootstrapping effort, right?
241
00:41:17.256 --> 00:41:29.225
'Cause, 'cause we have all of our first party apps, but now we gotta get everyone building apps, and I'm so excited to launch the, um, you know, the, the hardware support for loading apps in a couple weeks on, on device.
242
00:41:29.285 --> 00:41:38.546
The app store will come after that. It's gonna be a really fun year. So thanks for having me. Very cool. Very cool. And actually, real quick, can you show us the device one more time? 'Cause it's a clean-looking device.
243
00:41:39.066 --> 00:41:45.206
Yeah, let me do that. Let me just kinda wipe it off here too, you know. [laughs] So, uh, this is, you know, Passport Prime.
244
00:41:45.386 --> 00:41:57.686
We, we discontinued, by the way, the previous generations because we just think this is so superior to what we've built before that we just want to focus the company 100%, you know, on this new platform and operating system.
245
00:41:57.766 --> 00:42:06.246
Uh, this UI is actually the, the UI that most people have not seen. [laughs] This is what's gonna be coming in the next release where it's more like a grid of apps. Oh, nice. Like alpha league.
246
00:42:07.086 --> 00:42:17.676
You'll, you'll see on our website it doesn't look like this right now. We have like tiles. Yeah. I have an app on here for Doom. I was vibe coding Doom on here [laughs] the other week- Nice... just for fun. As one does.
247
00:42:17.726 --> 00:42:23.276
Just to see that I could do it. But, you know, ev- basically everything is packed into here. We have tons of cool features.
248
00:42:23.366 --> 00:42:33.066
Um, in addition to the camera, NFC, Bluetooth, we have like active tamper protection where if you try to open it up, it, it wipes the key off the device- Mm...
249
00:42:33.086 --> 00:42:40.056
which no one else has in like the hardware wallet, you know, space surprisingly. Um, but yeah, just, just a really nice clean form factor.
250
00:42:40.166 --> 00:42:53.166
We have the two colors, or we call them co- colorways, so Arctic Copper and Midnight Bronze. And then we do very like Apple style packaging now too, like very clean, you know, very, uh, very high. We've come a long way.
251
00:42:53.186 --> 00:43:08.386
Another thing that, um, we didn't get to talk about is that, you know, we manufacture in the USA, and I think we're the, the only company in this entire security hardware industry to be exclusively, uh, manufacturing these devices in the US, so I'm really proud of that.
252
00:43:09.106 --> 00:43:12.566
It's no better way to end this pod. Thanks again, man. Thanks.
1
00:00:00.260 --> 00:00:11.480
Boom. We're, we are in the pod, Zach, back for round two. How you doing? I'm doing great, and I'm, I'm scared about how much I've aged in the last four years since you sent that, uh, video.
2
00:00:11.640 --> 00:00:23.880
The, we posted the video clip on X from last time. Yep. Yeah. Yep, yep, yep. So we, so we last talked October 22nd, so it's coming up on three and a half years- Okay... of, of twen- in 2022, which is crazy.
3
00:00:24.680 --> 00:00:33.380
Um, and I threw that into Gemini to get a summary of what we talked about. Yeah. And it said the main topics were, we touched on the sovereign individual a good amount.
4
00:00:34.200 --> 00:00:41.340
We touched on fix the money, fix the world, and then we talked about, uh, the original Passport device. Mm-hmm.
5
00:00:41.500 --> 00:00:49.220
And, you know, right now we're recording as you guys are officially launching Passport Prime, which we'll definitely touch on. Mm-hmm. But first I wanna get into some...
6
00:00:49.430 --> 00:00:54.419
I see you, when I watch your timeline, you seem like a very intentional person, which I appreciate.
7
00:00:54.480 --> 00:01:03.059
You know, you're posting things like, "I'm trying to downsize my devices," and, you know, you just go from two to one, not have so many screens. And I'm in the same boat.
8
00:01:03.120 --> 00:01:08.479
Like, I'm feeling very fractured, and having 17 agents running at one time doesn't help that.
9
00:01:08.600 --> 00:01:20.780
But I, I guess I'm just curious, at like a high level, how do you view the current landscape of, like, AI, which is obviously useful, and, like, human autonomy, human, like the human interaction point of it? Yeah.
10
00:01:20.840 --> 00:01:29.030
I'm both the most excited I've ever been and then also the most frustrated with the state of computing today. Yeah.
11
00:01:29.030 --> 00:01:42.140
So I, I hold both of those things, I think, um, you know, I hold both of these things to be true at the same time, and I'm very fortunate with, you know, our company Foundation to be able to work on some of these problems.
12
00:01:42.260 --> 00:01:49.470
But I've been using the AI tools, of course, for at least, I mean, heavily over a year now.
13
00:01:49.640 --> 00:02:04.080
Our company uses these tools, and I find myself more scattered and maybe fragmented than ever before, and I don't know what device I'm supposed to be working on. You know, is it running on a Mac Mini in my server rack?
14
00:02:04.260 --> 00:02:17.200
Yes. Is it running on my MacBook Air? Yes. Is it running-- Am I talking to it from my phone? Yes. I've also put my devices away, like an iPad that I had, I don't use that anymore. I bought a Daylight tablet.
15
00:02:17.240 --> 00:02:27.920
That thing's a very cool device. I don't use it. And so I'm trying to be more mindful about not scattering myself across all these different devices. I, I stopped wearing my, uh, Oura smart ring.
16
00:02:28.420 --> 00:02:39.240
I never wanted, like, an Apple Watch. And so I want, and I seek the simplicity, and I think the AI technology may allow us to have that if built correctly.
17
00:02:39.940 --> 00:02:48.860
Uh, but at the same time, I feel more scattered than ever, and it's very hard to pull yourself away from the screen, I think, at this point in time. Yeah, totally agree. Yeah.
18
00:02:49.060 --> 00:03:00.640
One thing I heard that kinda struck me when I heard it is like... 'Cause coming back to this point around intentionality, like- Yeah... AI, AI can make you more efficient. It can scale you.
19
00:03:00.660 --> 00:03:06.150
But if you're not asking the right question of like, "What am I working on?" Then you could be just rowing really fucking hard in the wrong direction.
20
00:03:06.200 --> 00:03:18.690
And so I've heard it, like, one person said that, like what we'll probably gonna see the future of is like a large class of zombies, to put it, uh, kinda pithy, which is like the algorithmic people. Mm-hmm.
21
00:03:18.940 --> 00:03:28.000
Um, and they just like defer all their critical thinking to a, a, a, a basic question, and then they get an answer versus the people who...
22
00:03:28.549 --> 00:03:38.660
And right now it might look like less phone usage, um, and coming back to reality a little bit. And so do you have thoughts on that framing?
23
00:03:39.400 --> 00:03:43.210
Um, I'm not sure the question to ask, but I found it, uh, it does, it does lead to- I do...
24
00:03:43.240 --> 00:03:52.660
like this idea of there'll be a second-class warfare, if you will, and your, your device seems like it's more in the intentionality camp, camp if I had to like place it. Yeah.
25
00:03:52.720 --> 00:03:58.660
So the first thing I thought when you said that was everyone tagging Grok, "Is this true?" [chuckles] Mm-hmm. On, on X.
26
00:03:58.780 --> 00:04:16.159
And then the second thing I thought was we at Foundation are building for that, call it top 10% that is taking a very different approach to using, uh, this technology, whether it's AI, Bitcoin, more of that sovereign individual segment.
27
00:04:16.260 --> 00:04:22.100
So to already kinda tie it back to what we talked about, you know, years ago, there's different ways to use the technology.
28
00:04:22.160 --> 00:04:32.800
How we use it at Foundation is, is we use it, um, to help us architect the right products and tools and solutions, and then of course to help implement that architecture.
29
00:04:33.480 --> 00:04:43.640
But we consider now the team at Foundation, and especially my role as CEO, and then my co-founder and CTO, Ken, we consider ourselves to be like the architects, right?
30
00:04:43.800 --> 00:04:54.080
And to, and to sit down and, and figure out what we wanna build and use the AI tools to help us move, you know, an order of magnitude faster at architecting these products.
31
00:04:54.200 --> 00:05:01.710
And then we get to have it help us implement that very, very intentional, deliberate, detailed architecture.
32
00:05:01.940 --> 00:05:11.500
So we actually call it things like, like helping us make a blueprint and then executing, you know, on the blueprint. Got it. I guess let's jump in. We're gonna dance around.
33
00:05:11.600 --> 00:05:21.680
I have so many, like, interesting things running through my head- Yeah... and so I'm trying to gather them. But I wanna, I wanna just preface it, but with this idea that you guys are framing it as human authority device.
34
00:05:22.740 --> 00:05:32.540
Uh, I'd love for you to just break down what that term means to you. 'Cause I can kind of, I can, I can have an assumption in my head, but I'm curious how you guys view it. Yeah.
35
00:05:32.600 --> 00:05:41.260
So, and just to take, take a quick step back, um, you know, when, when I was speaking to you in 2022 and thinking about, uh,
36
00:05:42.400 --> 00:05:49.440
terms or, or, or areas of focus for us, like secure your entire digital life, that was a big theme of what we were working towards. So
37
00:05:50.500 --> 00:05:58.200
we started working on this new hardware, uh, actually at the end of 2022 after we raised our seed round led by Polychain.
38
00:05:58.940 --> 00:06:12.420
And so that was already in my head when I had, you know, done the podcast, but we didn't announce anything for like two years later about what we were building 'cause it took us, you know, years to, um, R&D this and bring it to market.
39
00:06:12.660 --> 00:06:24.616
We were not using that term human authority hardware until earlier this year.You know, earlier 2026, because AI wasn't even on our radar when we were working on this at the end of 2022.
40
00:06:24.656 --> 00:06:39.496
So like we were trying to, uh, allow someone to secure their entire digital life, uh, you know, more than just Bitcoin, more than just crypto, all their online accounts, important files, kind of everything on one device.
41
00:06:39.606 --> 00:06:52.716
And as we started to use the AI tools heavily in like end of last year, right, end of 2025, coming into 2026, everyone kind of started really using Claude and, and Codex and, and that kind of thing, right?
42
00:06:52.836 --> 00:06:57.556
Only in the last four, four to six months basically.
43
00:06:57.696 --> 00:07:19.356
And so we started running into our own concerns and challenges with the technology, which we realized, not right away, but we realized shared the same concerns that we were already building to address with Passport Prime, you know, the new hardware and this new operating system that we've been working on for years as well to power it, called KOS.
44
00:07:19.966 --> 00:07:25.676
And, and that-- those concerns are all about human authority, right? It's like how does the human stay in control?
45
00:07:26.216 --> 00:07:40.216
How do you make sure that important decisions, whether that's approving a Bitcoin transaction or approving an AI agent action, so maybe, maybe you're, you need to approve the transfer of one Bitcoin.
46
00:07:40.236 --> 00:07:53.596
You wanna do that on dedicated hardware outside the blast radius of your internet-connected phone or computer. But let's say your agent wants to do a huge code deployment. You probably wanna know what's going on there.
47
00:07:54.116 --> 00:08:02.856
Or your agent wants to send a big wire transfer, you know, from, from your business bank account. You probably wanna know what's going on there. And we realized it's the same problem.
48
00:08:03.416 --> 00:08:14.796
And so we tried to figure out, you know, what kind of problem is it? It's not a security problem. It's, it's an authority problem. And so that's why we started to use the term, uh, human authority hardware.
49
00:08:15.256 --> 00:08:25.716
You know, it's what we rolled out today with our, um, you know, uh, new website and, and, and messaging. And we also trademarked it because I think it's just a really cool term, and it's, it's, it's very different than
50
00:08:26.995 --> 00:08:39.096
hardware wallet, security key, FAE, all these kinds of security-related terms. We think it's an authority problem that we need to solve. So there's two, hmm, there's two big things that come to mind. One- Yeah...
51
00:08:39.136 --> 00:08:49.466
the f- the f- the framing of security versus authority kinda broke my brain as I was listening because- Mm... the most natural like pattern we're used to is something like 2FA. Mm-hmm.
52
00:08:49.496 --> 00:08:58.896
Which is like I wanna get into my account. I'm authorizing myself to get into my account. Mm-hmm. And then 2FA is the secondary feature to make sure that I'm the one that can get in,
53
00:08:59.976 --> 00:09:11.676
and that's what feels like your device does- Right... but for just like an increasing number of gated actions. So I guess the first question is, is that framing off at all? It's not that the framing's off.
54
00:09:12.716 --> 00:09:24.616
It's that we now are entering a new era where we're delegating our authority to others. It used to be you would delegate your authority to people who work for you, like if you're a c- a corporate structure.
55
00:09:24.836 --> 00:09:34.776
You as a human, say you're CEO of a company, right? You're, you're delegating your authority down the chain of command to other humans, and, you know, they, they report to you.
56
00:09:34.856 --> 00:09:39.856
You have a, you have a corporate structure around you. Same thing could be said for government, right? Or, or, um,
57
00:09:40.936 --> 00:09:47.096
all different types of, of hierar- hierarchical structures that humans have developed right over the centuries.
58
00:09:47.716 --> 00:10:02.396
But now we're entering a world where we're beginning to delegate our authority to AI agents running on computers, and that gets really messy because one, your computer, like you're probably on a Mac right now,
59
00:10:03.576 --> 00:10:11.476
uh, it ha- macOS has no idea. It can't differentiate between you and your AI agents from an operating system level.
60
00:10:11.526 --> 00:10:22.136
It doesn't know if you're trying to p- perform an action or if, you know, GPT or Claude is moving your mouse and clicking things and using the terminal. It has no idea, right?
61
00:10:22.686 --> 00:10:41.076
And so that I think is, is something that most people haven't really thought much about, and you're, you're, you're effectively now in addition to giving it access to your computer, you're having to give it access to credentials in order to do things.
62
00:10:41.196 --> 00:10:53.786
But you don't want it to do everything without your approval. Mm-hmm. And so you have to somehow decide, do I give it access to this cr- credential, or do I not give it access to this credential?
63
00:10:54.306 --> 00:11:05.016
And if I don't give it access to this credential, then you're kind of handicapping yourself, and you're not getting to use the technology to its full extent. So that's why we say it's an authority problem, right?
64
00:11:05.096 --> 00:11:15.436
It's like the tools that we're using right now, like if, if what we're building at Foundation didn't exist, you have like 2FA-type things where you could, you could approve actions, right?
65
00:11:15.496 --> 00:11:22.116
You can ensure that it's you as the human logging in, but you don't wanna log in. You want your AI agent to log in.
66
00:11:22.396 --> 00:11:31.516
You just wanna make sure if the AI agent does something that requires additional review or approval, you wanna make sure that that gets to you.
67
00:11:31.536 --> 00:11:45.516
And that's a completely different problem than like the entire computer security landscape has solved for. And so that's why we don't wanna call it FAE or security or... 'cause it's not a security problem necessarily.
68
00:11:46.096 --> 00:11:55.236
You, you wanna give it the credentials. Mm-hmm. You're not trying to keep the credentials away from it, right? And so that, that's why we, we call it an authority problem. Yeah, yeah.
69
00:11:55.356 --> 00:12:04.656
This, this is super interesting philosophically. [chuckles] Yeah. And, and I like the, the analogy of you delegate to your employees, for example.
70
00:12:04.896 --> 00:12:12.266
As the employee proves himself out, maybe he gets his own company card, and he's like, "If you-- up to 10K, YOLO it 'cause I trust you." Yeah.
71
00:12:12.266 --> 00:12:21.476
"After 10K, let's go, you know, run it by the CFO, run it by me to make sure-" Yeah "... you're not doing anything stupid." And it's similar with agents, which is like there's a UX problem here.
72
00:12:21.516 --> 00:12:29.395
There's an efficiency problem at play here, like you wanna be efficient, but you also- Right... wanna have human in the loop at the right time to make sure you can- Mm-hmm... review things.
73
00:12:29.776 --> 00:12:36.672
TypicallyWhat is, well, if you have any thoughts, feel, feel free to jump in Yeah, no, just thinking when you said human in the loop, I got kind of triggered [laughs]
74
00:12:36.712 --> 00:12:44.191
'cause that, that's the term that everyone's using, right? Human in the loop. Mm-hmm. But, uh, it's all an illusion right now. There's no human in the loop.
75
00:12:44.452 --> 00:12:58.232
The, the, the human in the loop right now is that you give, uh, could be your OpenClaw or Hermes, right? Or could be Codex or, or Claude. You give it access to everything. You give it access to your- Hmm...
76
00:12:58.352 --> 00:13:05.972
may- maybe even your bank account. You give it, you know, X notion, your email, your company stuff so that, GitHub, right?
77
00:13:06.112 --> 00:13:20.372
You give it access to everything, computer use, and then it has the access, and it says, "Hey, can I please do this?" But it already has the authority to do this, right? It already has the capability and the credentials.
78
00:13:20.972 --> 00:13:30.932
It's just programmed to ask you to make you feel more comfortable, like as a formality. Yeah. And, and if there's so many ways that can go off track.
79
00:13:31.032 --> 00:13:41.132
It could be like a malicious thing with prompt injection or, you know, s- one of these AI infra, you know, model companies is compromised or something is compromised along that chain there.
80
00:13:41.712 --> 00:13:48.172
Or it could be a bug, it could be a, a model that, that maybe wants to escape or doesn't really understand the instructions.
81
00:13:48.212 --> 00:14:00.032
And we've already read so many stories of people wiping all their photos and emails and AWS entire regions going down, right? 'Cause they're just like vibing it, and the agent makes a mistake.
82
00:14:01.072 --> 00:14:07.582
So human in the loop up until now, I would argue, is like totally fake. It's, it doesn't exist. Mm.
83
00:14:07.612 --> 00:14:14.792
It just, it's a, it's just programmed into Claude or GPT's interface to make you feel more comfortable that you can say, "Yes, I approve this."
84
00:14:14.832 --> 00:14:29.152
And so what we're more interested in is like how do you force these, uh, tools answer to the human, right? You force it. You don't... It's not a please, please do this, right? Please listen to me.
85
00:14:29.232 --> 00:14:39.052
It's, it's like a you have to answer to me, and that's a really hard problem. I'm not saying that we've solved it, but that's what we're working towards right now- Yeah. Yeah... is like in that problem.
86
00:14:39.112 --> 00:14:47.832
That, that's, that's a great pushback 'cause you're right. It, it is like everyone has already flagged dangerous escape permissions right now, and- You have to, right?... and the agent is... Exactly. Yeah.
87
00:14:47.872 --> 00:15:01.482
But you're right. The, the agent is on his best behavior most of the time when he's not acting like an ADIQ Luddite. And but the difference is, you know, on the front end, strict permissions- Mm...
88
00:15:01.512 --> 00:15:12.972
and then to get out of its sandbox, it has to ask and then be signed by maybe something like Prime. And so what does that-- what's the current UX look like?
89
00:15:13.062 --> 00:15:25.012
'Cause we're used to working in Codex, and we are already on full send mode, so like what, what practically has to change in these paradigms to go from that to a human authority paradigm, but still working within agents?
90
00:15:25.672 --> 00:15:36.312
That's the really hard part. We will have like, like the... What we're envisioning and building toward right now is the idea that, and we don't have it shipped, we have it like in- internally, right?
91
00:15:36.392 --> 00:15:41.692
So we're using some of the stuff internally. We've done some really cool stuff internally. We don't have it shipping yet, like on the hardware.
92
00:15:42.312 --> 00:15:55.992
What we're trying to do is enable the operating system on Passport Prime, you know, which is our new, right, like a touchscreen device, and this is the one that, you know, we're announced today is, is generally available to purchase and, and everything like that.
93
00:15:56.212 --> 00:16:02.492
I lost my train of thought there. What was I saying? The UX of actually like- Oh, yes. Yes.
94
00:16:02.972 --> 00:16:18.022
So effectively, we have this operating system, Kiosk, that's a microkernel operating system, and we're beefing up the operating system itself to have a policy engine in the core of the kernel so that
95
00:16:18.972 --> 00:16:21.792
like our, our envisioned UX here is that
96
00:16:23.192 --> 00:16:35.492
you have a set of policies that you define, and you don't get bothered for approval as long as your policies say that they, the, the reads, the writes, everything can flow freely, right? Like the,
97
00:16:36.392 --> 00:16:43.892
the tool calls, the MCP tool call- tool calls, the CLI calls, right? Like if, if your agent's, uh, maybe an, an great example is
98
00:16:45.272 --> 00:16:56.092
your agent sending a wire transfer or a payment, but it's an invoice that is below a certain dollar amount, and it's similar to what was paid last month to that existing vendor. You probably want that to just send.
99
00:16:56.172 --> 00:17:09.722
However, maybe it wants to send a sizable, let's say a million-dollar wire to some, someone you've never sent money to before, and that's over your threshold. You wanna be notified. So our like
100
00:17:11.172 --> 00:17:30.452
our desired, like, uh, like the goal here for us is to be able to help give a user experience like that, where you get to have it in, you know, kind of like how Codex has an, an auto mode, where it's kind of like evaluating its own policies and trying to decide, do I need to ask the user for permission?
101
00:17:31.162 --> 00:17:45.272
Kind of like that, but taking that out of the hands of those individual AI tools and bringing it onto something outside of the blast radius, right outside of your computer or phone, so that you allow what, like the, the low-risk actions to just happen without any friction.
102
00:17:45.872 --> 00:17:52.182
But then you have just a little bit of that friction for the important stuff, right? And, and you would be able to customize that.
103
00:17:52.332 --> 00:17:57.712
So that's what we're working towards, and we have some cool stuff, some cool apps that we've built internally.
104
00:17:57.792 --> 00:18:07.452
Um, and, and I'm not-- And the cool thing is, is we don't have to build all these apps ourselves because, and we could talk about that too. The device, the operating system is actually an open developer platform.
105
00:18:07.512 --> 00:18:13.692
That's one of our other announcements today, that our, uh, beta of the SDK, the software development kit, is now live.
106
00:18:13.802 --> 00:18:21.252
And so we're hoping that this is a platform because it's an open, uh, developer platform, and anyone can come build apps for it.
107
00:18:21.832 --> 00:18:36.271
We're hoping that there'll be a lot of folks who end up building apps for these kinds of use cases 'cause even if we built something, and we will, we're not gonna be able to cover the whole gamut of every single, you know, AI tool out there, especially with the pace at which the industry is moving.
108
00:18:36.292 --> 00:18:46.532
And so we're basically putting a, a call out to developers to welcome them, you know, to build on the Kiosk ecosystem. I have a ton of thoughts, but- Yeah... you touched on Kiosk.
109
00:18:46.572 --> 00:18:58.628
That's interesting.I wanna, I wanna double click on the, the, the UX kind of like feeling just a little bit further, um, 'cause I want people to like... So I ha- let's say I have a device, and
110
00:18:59.908 --> 00:19:14.898
I have a Hermes agent that's doing stuff. Where, where would the policies and, like, the f- the initial friction, do I set something up differently in my, in my Hermes? Do I have to load an Hermes app on the app? Right.
111
00:19:14.898 --> 00:19:20.338
Like, w-what would, what would that currently look like if I'm trying to spin up a new agent and wanna have guardrails from day one?
112
00:19:21.448 --> 00:19:30.358
So I think initially, and there's a few different ways to do it, but initially and, and actually I, I've been-- I started working on a plugin for that, 'cause that's what I use personally.
113
00:19:31.008 --> 00:19:40.168
So I started working on a plugin, and I hope we'll be able to release something in the coming, uh, couple months specifically for... Is it Hermes? Hermes? I don't even know the, the right way to say it.
114
00:19:40.328 --> 00:19:49.878
But I think in that case, the initial starting point would be a, a plugin that moves those approvals, where it's asking you constantly unless you, like, turn, turn that off, right?
115
00:19:49.908 --> 00:19:59.818
It's constantly asking you if it could do things, is to move that over to Passport. And there's a couple different ways to do it. In the short term, it might mean that you have-- It's, like, not
116
00:20:00.828 --> 00:20:14.368
as perfect and bulletproof as it can be, 'cause it might still be inside of Hermes itself, where it's just told right in the core l-layer where it's asking for permissions, you know, route that to Passport instead.
117
00:20:14.378 --> 00:20:22.478
And it can be, it can be signed. You can make it where, like, only Passport can update those permissions, that they're all signed from the Passport hardware.
118
00:20:22.508 --> 00:20:28.068
So you could do a lot of great stuff to make it better than anything out there today, uh, in terms of approvals.
119
00:20:28.178 --> 00:20:40.148
So you would create and sign the policies on the Passport hardware, but then those policies, that signed policy table would live inside of Hermes or Open Claw or, you know, whatever agent-type framework.
120
00:20:40.388 --> 00:21:03.088
We are working on some other things that probably are still too early to talk about that would put something online or on the computer that the agent is running on, like some o- extra software that we build that physically takes all those policies, like, outside of the agent and may even start to bring in tool credentials, right?
121
00:21:03.148 --> 00:21:10.848
Like keeping the credentials for all the MCP tools, the CLI, all that stuff completely in a separate environment from the agent.
122
00:21:11.848 --> 00:21:24.688
And then the, the real golden standard is being able to even sit between the, um, API calls, between the models themselves and- Mm. -the agent, right? So if you're using, uh,
123
00:21:25.628 --> 00:21:44.698
GPT 5.5 on the back end with your agent, if you're using, um, Opus 4.7, to actually be able to sit between that and intercept all of those, uh, API calls back and forth, and then if something doesn't look right or if something doesn't comply to policy, direct that to Passport for review and approval.
124
00:21:44.828 --> 00:21:52.168
So there's a lot to build, and it's gonna be a challenge because it's gonna have to be different things for every single tool.
125
00:21:52.848 --> 00:22:11.528
And so that's also why we're much more interested in, in working on this developer ecosystem, because we're gonna be putting out some of that from foundation as opinionated, right, uh, pieces of software and, and apps on the device that work with certain tools, but it's gonna be impossible for us alone to cover, like, the full breadth of everything.
126
00:22:12.388 --> 00:22:20.848
Yeah. Yeah, okay. That's really cool. So one thing that jumped out when you were, when you were talking is that typically hardware wallets, especially in the Bitcoin space, you know, they're meant for self-custody.
127
00:22:21.318 --> 00:22:30.298
Mm-hmm. And so effectively, you store Bitcoin on them, you put them in a drawer, and they sit there until you- Yeah... wanna pull them out. So they're very infrequently used.
128
00:22:30.438 --> 00:22:39.148
This is a device that, as I'm listening to you talk, it's gonna sit with me next to my phone- Mm... depending on what kind of flow state I'm in, and I'm gonna work on my laptop, work on my laptop- Right...
129
00:22:39.168 --> 00:22:48.968
and then I'm gonna get pinged periodically on my Prime. Uh-huh. And this will be through Bluetooth, I think. Mm. I'll, I'll, I'll prove something, and then I'll get back into my flow state. Do I have that right?
130
00:22:49.668 --> 00:22:50.328
That's correct, yep.
131
00:22:50.428 --> 00:23:19.588
It's, um, it's through Bluetooth, and we, we built a post-quantum encrypted protocol on top of the Bluetooth stack so that it's really like an app on your phone, or we're gonna be launching soon an app on our server that's pushing data directly down to the Passport hardware, and you do this initial quantum link connection where you scan a QR code on one of the devices, and it creates this post-quantum encrypted tunnel between whatever app and the actual Passport hardware.
132
00:23:19.708 --> 00:23:30.628
So it's really cool. So you're using the Bluetooth chip for the comm-- uh, convenience of the communication, but you're not relying on the Bluetooth chip for its own, you know, encryption or, or anything like that.
133
00:23:30.668 --> 00:23:39.658
But yes, you got the UX right. Effectively, what it looks like is, and we have a new app, a new mobile app that's gonna be coming out to help facilitate this even better than what we do now.
134
00:23:39.808 --> 00:23:46.738
You hit, like, our server, and then the server hits our mobile app, and then the mobile app sends the, the data over th-to the device. Mm.
135
00:23:46.748 --> 00:23:59.648
But the payload is always encrypted, so our server can't see it, the mobile app can't see it, and even the Bluetooth chip can't see it, 'cause you're quantum linked between, like, your app or your use case and the actual hardware, and everything else is like a relay.
136
00:23:59.658 --> 00:24:02.408
It's like a relay network to get stuff back down to the device.
137
00:24:02.468 --> 00:24:25.348
So it solves a lot of problems, right, with QR codes and, and, uh, the way that people typically use hardware wallets right now for communication, 'cause our goal is, like, any app that you're connecting to Passport, uh, to Passport Prime, uh, it has to have its own unique, uh, fully encrypted, uh, communication tunnel so that it can talk to it across, you know, any interface.
138
00:24:25.378 --> 00:24:30.868
And that, that interface could be USB as well. If you're really hardcore and you're like, who knows what you're doing?
139
00:24:30.888 --> 00:24:38.728
You have 25 agents and you're like, [chuckles] "I wanna keep-- I wanna plug this thing in with USB, and I wanna keep, like, all of the MCP creds and everything in here, and I wanna..."
140
00:24:38.768 --> 00:24:50.108
You'll be able to do that, no problem. That's very cool. Yeah. Have you-- What's the biggest use cases you identified? 'Cause money is the easiest one to kinda, like, pinpoint. Yeah.
141
00:24:50.128 --> 00:24:52.788
And you've mentioned it with policies and thresholds. Yeah.
142
00:24:52.928 --> 00:25:03.858
Um, but yeah, what have you seen for, like, where's the biggest value in a device like this?It's a very boring answer, I think [chuckles] which right now is the, is the multi-factor authentication stuff.
143
00:25:04.058 --> 00:25:17.868
I mean, I think for most people, most people's 2FA is in a really bad state. And, and not even just the 2FA, but most people have their digital lives scattered across an array of different, um, you know,
144
00:25:19.078 --> 00:25:27.758
whether it, it's sometimes it's hardware, like YubiKeys, or you're storing files on like hardware-encrypted flash drives. Uh, sometimes it's software. It might be a password manager.
145
00:25:28.018 --> 00:25:36.758
Sometimes people are playing around with all the different Bitcoin and crypto mobile apps and hot wallets, and they're writing down seed words on paper, and they have like a stack of them somewhere.
146
00:25:36.838 --> 00:25:48.658
And we-- Yeah, so, so we have a vault app on, on Passport Prime, and you can bring in all your existing seeds and create new seeds for use on all your other hot wallets, right? And, and like that's an amazing feature.
147
00:25:48.898 --> 00:25:56.738
We have encrypted file storage. We have, uh, FIDO security keys, so you don't need separate YubiKeys. You can tap them or, or plug in with USB.
148
00:25:57.278 --> 00:26:11.118
And we have those six-digit 2FA codes because it's always been crazy to me that your phone is like your primary computing device now for most people, but also you're using like Google Authenticator on your phone for your most important accounts.
149
00:26:11.238 --> 00:26:22.988
And you can't lock down your Apple account or your iCloud account, right? Or if you're on Android, your Google account with like a app on your phone. It's a circular dependency. It doesn't make any sense. Mm-hmm.
150
00:26:23.018 --> 00:26:27.078
Yet we all do it, and most people still have SMS 2FA.
151
00:26:27.088 --> 00:26:43.898
And so I think just helping everyone get all their important stuff off their phone and in all these other places that are scattered around into one ultra-secure device that can also be like automatically backed up, which is a really cool feature, is, is like the big opportunity right now.
152
00:26:44.538 --> 00:26:48.738
Um, yeah, of course, the hardware wallet stuff, very important.
153
00:26:48.868 --> 00:26:58.818
Uh, if you wanted to keep much larger amounts, you know, you could also use like passphrases or other things like that, or multi-sig if you're concerned about, you know, carrying this device with you more often.
154
00:26:59.178 --> 00:27:07.058
Like how does that jive with also storing millions of dollars of, of Bitcoin or something like that? So you can, you can kinda handle it like that. But, um,
155
00:27:08.338 --> 00:27:14.358
there's never been a device or product that allows you to actually bring all that stuff in together into one.
156
00:27:14.398 --> 00:27:25.518
And so I think like we're gonna be putting out a lot of guides and videos and, um, tutorials about how to do all that, right? How do you lock down your iCloud account? How do you lock down your Google account?
157
00:27:25.838 --> 00:27:37.678
How do you get all the stuff onto Passport Prime so that you can really secure your entire digital life? And I think for most people, that, while it feels a little boring, it's so important, especially right now. Yeah.
158
00:27:37.738 --> 00:27:45.498
No, no, that's what it feels like. I'm-- I almost feel like I'm in the future and I'm like, "This is exciting as shit." [chuckles] But it s- it feels, it feels very necessary.
159
00:27:45.658 --> 00:27:56.318
I wanna, I wanna ask two quick questions, and we can do them kinda rapid fire so we can cover some other stuff on Kiosk. Yeah. One, what's the experience like building on it? Can someone like me who's- Yeah...
160
00:27:56.458 --> 00:28:01.718
vibe coder extraordinary- I love this question already. Yeah... can I, can, can I do some stuff? And then second- Mm-hmm...
161
00:28:01.738 --> 00:28:15.198
because it's a security-focused device, how do you, uh, deal with things like security of these apps- Love it... and getting approved and these kind of things? Love it. Okay. S- second question first maybe.
162
00:28:15.298 --> 00:28:19.168
Um, we started building this 'cause we were not happy with how Ledger did their app model.
163
00:28:20.198 --> 00:28:27.078
That's why, that's when we started building this in end of 2022 was we had the idea, we started speccing out this operating system.
164
00:28:27.118 --> 00:28:37.998
We said, "What would we build if we could build our dream operating system for like a hardware wallet type device?" And the answer was, well, you wanna have apps. You don't want apps to just be coins.
165
00:28:38.698 --> 00:28:47.758
You don't want Foundation to have to review every app necessarily because, like, you could review it for functionality like Apple does, right? Make sure it's not a scam or something.
166
00:28:47.978 --> 00:28:52.598
But you don't want-- Y- we can't review the source code for every app, especially if there's like thousands of apps.
167
00:28:52.638 --> 00:29:01.978
And I think it's one of the reasons why Ledger has had success, but also relatively small number of apps because of how they have to review them. They all get access to the master seed.
168
00:29:01.987 --> 00:29:14.878
They're very limited in functionality. You can't really build a user interface for your app. You're kinda stuck with whatever they give you. So we set out originally to build this as like the Ledger killer, right?
169
00:29:14.958 --> 00:29:23.458
To be able to leapfrog them as the platform. Now I think the opportunity is even, you know, much bigger than that. But we have sandboxing, hardware sandboxing in the apps.
170
00:29:23.998 --> 00:29:30.098
We have this microkernel architecture, message-passing microkernel. It's all in Rust. It's like 9,000 lines of code.
171
00:29:30.678 --> 00:29:42.398
Uh, Linux kernel is 30 million lines of code, so we're talking like orders of magnitude, right, smaller. You could print it out and read it in a day if you're like a g- you know, talented developer.
172
00:29:43.218 --> 00:29:49.458
And anytime an app wants to do something, it has to go to the kernel, see if it has permission, and then it gets passed to whatever it needs to do.
173
00:29:49.638 --> 00:29:56.548
And that's a very sophisticated permission-based dream operating system architecture model. That's not what Ledger does.
174
00:29:56.688 --> 00:30:03.618
And so we set out to build all that to make sure that we could have apps that could, uh, could not get out of their sandboxes.
175
00:30:03.638 --> 00:30:15.118
And so even if you have like an untrusted app on, you know, on the device, uh, it couldn't do any damage. It doesn't even get access to the, to the seed. It gets a hardened derived child seed in the s- Mm...
176
00:30:15.198 --> 00:30:24.718
in the sandbox. So that's your answer to question one. The whole thing was designed to be able to run untrusted apps, you know, side by side.
177
00:30:24.758 --> 00:30:35.958
Number two is fun, especially when we re- when we release in a couple weeks version 1.3 of Kiosk, which will be able to spawn an MCP server on device.
178
00:30:36.098 --> 00:30:48.718
So you can plug it in, and you can have your AI actually test the app and do every single thing on the screen, tapping, swiping, keyboard, everything. So you could go to sleep, wake up, and have a fully working app.
179
00:30:48.838 --> 00:31:02.034
But the UR- URL you would wanna go to is foundation.xyz/dev. That takes you right over to our new developer documentation.You just give it the link to your AI.
180
00:31:02.374 --> 00:31:10.574
It'll, within five minutes, maybe more if you don't have, like, Nix and stuff on your computer, it'll have to set up a lot of stuff, but within some number of minutes, it'll configure everything for you.
181
00:31:11.114 --> 00:31:21.934
We have a foundation CLI that's shipping. It knows how to scaffold an app. It can run your app on the simulator, on the computer, so you can, you can develop the whole app without even needing the hardware.
182
00:31:22.714 --> 00:31:33.034
You can one-click build the app, sign the app, make sure the app is reproducible, 'cause the apps that we're gonna allow in our app store have to be open source and reproducible initially.
183
00:31:33.114 --> 00:31:37.334
You can also sideload the app, you know, so you can-- you'll be able to ship the app directly to your user.
184
00:31:37.414 --> 00:31:48.314
But yes, like, you could point any of the AI tools to our dev docs, and within, depending on what you already have on your computer dependency-wise, within like thirty minutes, you'll be building a Kiosk app.
185
00:31:48.754 --> 00:31:58.274
And then when one point three comes out, it'll test itself on hardware for you. And so the whole thing is designed for agentic development. Very cool.
186
00:31:58.694 --> 00:32:05.074
And what-- This is live now, shipping worldwide as of this today, but when this is out to- Yes, so the software is shipping worldwide.
187
00:32:05.314 --> 00:32:13.174
The SDK is, is live in beta, where you can have it up and running with the simulator and start building on your computer.
188
00:32:13.354 --> 00:32:23.674
We also have free d-developer devices to folks who, um, any serious, you know, companies or individuals, right, that are wanting to build apps. We can... We have, like, a thousand of them, I think, right now.
189
00:32:23.894 --> 00:32:31.734
So we can, we can mail you developer devices. And then when, when version one point three comes out, you can actually start to do the MCP development.
190
00:32:31.894 --> 00:32:42.874
You can sideload apps onto your hardware and continue developing on the hardware. So we need a couple more weeks before that software release is out. But you can get started today right now on the simulator. Got it.
191
00:32:42.954 --> 00:32:51.654
Very cool. Yeah. And, and what's the price point for this? So this is three hundred and forty-nine dollars. Yeah. Very cool. La-last couple questions, then we can bring this to a close.
192
00:32:51.814 --> 00:33:02.734
One thing we were talking before recording on was the current narrative around AI and Bitcoin is all just machines- Yeah... are gonna pay in Bitcoin because it's superior money- Yeah...
193
00:33:02.744 --> 00:33:11.874
for everything, and they're gonna transact in crazy velocity, and maybe this will make Lightning the number one thing. It just goes down the line. And you said you had some thoughts on this.
194
00:33:12.074 --> 00:33:20.654
I'm curious, how do you view the space right now? Well, firstly, I don't, I don't know if you believe that's true. Do you believe that's true? I don't know if I believe it's true. I, I want to.
195
00:33:21.934 --> 00:33:31.444
I-- To, to me, I would say it goes back to this question of like, how much do I trust my agent? Mm-hmm. Mm-hmm. And like, I don't trust my agent to... I'm not putting ten K into my agent, that's for sure. Right.
196
00:33:31.454 --> 00:33:42.454
I might put a hundred bucks and play with it. That's where I'm currently at. Yeah, so for me, I kind of agree, and it's important to realize, like, these agents aren't being made in a vacuum.
197
00:33:43.194 --> 00:33:52.444
It's not the best technology that wins. It's what-whatever OpenAI and Anthropic tell and, and Google tell them that they're gonna pay with, they're gonna pay with, right? Mm-hmm.
198
00:33:52.454 --> 00:34:03.574
It's probably gonna be some stable coin thing on Stripe or, you know, something like that. So do I believe Bitcoin is the best form of money for agentic payments? Yes, of course. Do I think anything's gonna really do it?
199
00:34:04.034 --> 00:34:15.773
Maybe the open source models that are being controlled by the sovereign individuals, right? So I think that's one interesting angle here where, um, I don't think it's gonna happen on its own.
200
00:34:15.794 --> 00:34:26.424
But, like, taking a step back, I think it's, it's the only Bitcoin narrative that I've seen so far with AI has been this agentic payments, right? Lightning or other things, machine-to-machine payments.
201
00:34:27.204 --> 00:34:43.094
We're really interested in this new narrative that, uh, from a security perspective, which is kind of what I laid out in the beginning of our conversation here, which is bringing the Bitcoin principles for how we secure Bitcoin to the AI world, right?
202
00:34:43.134 --> 00:34:53.714
Mm-hmm. Where like the whole idea of having, like, the most hardcore security, keeping the keys on a dedicated device outside of the blast radius, right?
203
00:34:53.954 --> 00:35:08.474
Keeping the keys for everything, reviewing and approving important transactions on a secure, dedicated device. That's everything that we've built for Bitcoin. That's also the whole Bitcoin philosophy and mindset, right?
204
00:35:08.544 --> 00:35:16.234
The idea of a hardware wallet. It's... No one has any idea what you're talking about if you try to tell someone what it is outside of the Bitcoin or crypto world.
205
00:35:17.433 --> 00:35:22.794
Bringing that to the AI world, I think is a huge contribution that we as Bitcoiners can make.
206
00:35:22.814 --> 00:35:36.154
Because from a, from a security, like a principled security, privacy, and open source angle, like those are things that are core to the Bitcoin space, and we need to bring those to the AI space, like desperately.
207
00:35:36.434 --> 00:35:45.554
'Cause there seems to be, at least on the consumer products, like no security, no privacy- Mm-hmm... everything completely closed, and that's not good.
208
00:35:45.594 --> 00:35:52.894
And so I, I think it's about, like, those values and principles and the way of doing things and trying to bring that over to AI. Yeah, that's really good.
209
00:35:53.094 --> 00:35:59.894
And yeah, I, I agree that Anthropic and OpenAI are the kingmakers here of, like- Yeah... what the model's gonna choose.
210
00:36:00.694 --> 00:36:09.914
Also, LLMs are trained on the corpus of in- the internet, so if you tell it it's freedom money enough times, it's gonna tell you it's freedom money. Right. And so it's like, is it really going from first principles?
211
00:36:10.034 --> 00:36:22.874
Probably not. Um- It's a good point. And, and lastly, I think my, my stance on this has been, uh, I think we're way too bullish on the timeline. Like, Bitcoin's gonna be the thing over the next century.
212
00:36:22.954 --> 00:36:28.974
Like, it's the most important asset- Mm... out there. But it's gonna take thirty, forty, fifty years to, like, really weave its way through everything.
213
00:36:29.534 --> 00:36:40.674
And, like, as censorship starts to get more and more ingrained and maybe they push back on agents, maybe then BTC becomes more valuable. But, like, this is a long-term play. It's not gonna be, like, this cycle.
214
00:36:41.534 --> 00:36:52.034
Yeah, and I also think it's, like, bringing it back to the sovereign individual thesis. There's already a very small number of AI users that are running, like, an Open Claw or Hermes on a Mac Mini, right?
215
00:36:52.134 --> 00:36:56.724
Like, it's, it's tiny. Most people are just using, like, GPT on the eight dollar a month plan right now- Yeah...
216
00:36:56.724 --> 00:37:03.434
and they don't even know what all, like, the, what Codex is or Opus is or, you know, what- whatever, any of that stuff, right?
217
00:37:04.474 --> 00:37:22.666
And there's gonna be this small segment of folks thatAre running their own hardware, running their own models, using it for just orders of magnitude more capabilities than what a normal consumer or even normal person working in like a company like within Enterprise or ChatGPT or Claude app is doing.
218
00:37:22.806 --> 00:37:34.426
And so I do think there's gonna be a disconnect. There's a disconnect now because of like if you're using these tools like, like you are, like I am, you're already like years ahead of what the average person is doing.
219
00:37:35.226 --> 00:37:40.486
I think that gap's only gonna grow. I mean, we're gonna be buying, like we're gonna be running this stuff at home, right?
220
00:37:40.606 --> 00:37:50.066
Uh, if you're not already using hopefully, um, you know, tools like, uh, like, like Passport, like what we've built to help, you know, keep yourself secure as you use these tools.
221
00:37:50.306 --> 00:38:07.726
So, uh, yeah, I just think there's gonna be ultimately kinda two classes of users, and I think we already know that we're gonna be in that class that is, is more of the sovereign individual class, and I think we need to embrace that, and I think we need to kinda level ourselves up and, uh, that's, that's like the main market that Foundation is, is really building for.
222
00:38:08.806 --> 00:38:21.546
Well said. Okay, I think last question then. This has been a, a fantastic conversation. Let's think out five years. Let's say things go exceedingly well. What role do you think human authority hardware, I like that term-
223
00:38:23.506 --> 00:38:27.286
I'm glad. [laughs] Thank you. How different is society? How, how are people using it?
224
00:38:27.336 --> 00:38:35.046
Like what's, what's the rosy-colored like future you envision five years out if things are like going perfectly and using this hardware effectively?
225
00:38:35.226 --> 00:38:47.366
I think that my goal is that you have this device, this Passport device in your other pocket. Maybe it's in a wallet folio case or something, like a nice, you know, leather case.
226
00:38:47.386 --> 00:38:51.746
We have a couple things like that now, but I wanna do even better. You know, you have this thing with you.
227
00:38:52.706 --> 00:38:59.966
It mostly leaves you unbothered, and when you wanna do an important action, important approval, you're able to take it out for a minute and do that.
228
00:39:00.746 --> 00:39:10.026
But that otherwise, you're able to really embrace all the technology so that you have... Y- you have to, y- you're able to use your computer less.
229
00:39:10.086 --> 00:39:21.906
You're able to let your AI agents and all those tools work for you in the background, and you're only interrupted when you need to be interrupted for something important.
230
00:39:22.126 --> 00:39:28.246
And to me, a- and, and of course, we wanna keep leveling up the hardware, right, on, on like the device side of things.
231
00:39:28.286 --> 00:39:37.426
We wanna keep making it, you know, kind of like what Apple has done over time, you know, more battery life, thinner, you know, easier to carry with you, all that kind of stuff.
232
00:39:37.516 --> 00:39:55.366
But we wanna have a, a robust app ecosystem where every app you use today is integrated into Passport and QOS so that you can just really easily, one, stay secure, and then two, get all of your approvals and all the things you need to look at from an authority perspective.
233
00:39:55.446 --> 00:39:57.866
You can just do that easily on device.
234
00:39:58.026 --> 00:40:14.766
And so it, it's a future where hopefully you're, you're spending less time in front of all your screens doing everything, but you can also feel that you are giving access to all your stuff to these AI tools, but you don't have that thing in the back of your mind, "Should I be doing that?
235
00:40:15.006 --> 00:40:25.386
Was that the right thing to do? Did I just lose all my security, all my privacy," right? All that kind of stuff. You feel like you don't have to make that trade-off. You can go all in, but you keep your authority, right?
236
00:40:25.426 --> 00:40:31.126
And you, and you, and you're able to concentrate that into a device that you can carry with you. Yeah. Well, well said.
237
00:40:31.366 --> 00:40:43.646
The ending there, that's what was popping, popping up in my head is like this is a device for peace of mind. Mm-hmm. And so you know, okay, I can still be at the frontier of efficiency in using these tools- Mm-hmm...
238
00:40:43.666 --> 00:40:55.866
but not lose my ass because I'm using it within the proper guardrails. Yeah. Yeah. Yeah. Very cool. Okay. Well, uh, man, this has been great. Congrats on- Yeah. Thank you... the launch. Congrats on the fundraise.
239
00:40:56.206 --> 00:41:08.996
My audience is decently technical, so as you mentioned, if, if people wanna get a developer device and start hacking away, I'll put the link below- Mm-hmm... uh, to get in touch. Yep. But, man, this has been fantastic.
240
00:41:09.046 --> 00:41:17.176
Thanks for coming on the pod. Yeah. Thanks so much for having me, and, uh, would love to have folks building on this. Like we're-- It's, it's a bootstrapping effort, right?
241
00:41:17.256 --> 00:41:29.225
'Cause, 'cause we have all of our first party apps, but now we gotta get everyone building apps, and I'm so excited to launch the, um, you know, the, the hardware support for loading apps in a couple weeks on, on device.
242
00:41:29.285 --> 00:41:38.546
The app store will come after that. It's gonna be a really fun year. So thanks for having me. Very cool. Very cool. And actually, real quick, can you show us the device one more time? 'Cause it's a clean-looking device.
243
00:41:39.066 --> 00:41:45.206
Yeah, let me do that. Let me just kinda wipe it off here too, you know. [laughs] So, uh, this is, you know, Passport Prime.
244
00:41:45.386 --> 00:41:57.686
We, we discontinued, by the way, the previous generations because we just think this is so superior to what we've built before that we just want to focus the company 100%, you know, on this new platform and operating system.
245
00:41:57.766 --> 00:42:06.246
Uh, this UI is actually the, the UI that most people have not seen. [laughs] This is what's gonna be coming in the next release where it's more like a grid of apps. Oh, nice. Like alpha league.
246
00:42:07.086 --> 00:42:17.676
You'll, you'll see on our website it doesn't look like this right now. We have like tiles. Yeah. I have an app on here for Doom. I was vibe coding Doom on here [laughs] the other week- Nice... just for fun. As one does.
247
00:42:17.726 --> 00:42:23.276
Just to see that I could do it. But, you know, ev- basically everything is packed into here. We have tons of cool features.
248
00:42:23.366 --> 00:42:33.066
Um, in addition to the camera, NFC, Bluetooth, we have like active tamper protection where if you try to open it up, it, it wipes the key off the device- Mm...
249
00:42:33.086 --> 00:42:40.056
which no one else has in like the hardware wallet, you know, space surprisingly. Um, but yeah, just, just a really nice clean form factor.
250
00:42:40.166 --> 00:42:53.166
We have the two colors, or we call them co- colorways, so Arctic Copper and Midnight Bronze. And then we do very like Apple style packaging now too, like very clean, you know, very, uh, very high. We've come a long way.
251
00:42:53.186 --> 00:43:08.386
Another thing that, um, we didn't get to talk about is that, you know, we manufacture in the USA, and I think we're the, the only company in this entire security hardware industry to be exclusively, uh, manufacturing these devices in the US, so I'm really proud of that.
252
00:43:09.106 --> 00:43:12.566
It's no better way to end this pod. Thanks again, man. Thanks.
