Hey everyone,
Been a minute and I have a new podcast episode I think you’ll find interesting.
I sat down with Zach Herbert, CEO of Foundation Devices, to talk about their new device, Passport Prime, and we ended up going deep on something I think is going to become a much bigger conversation:
What happens when AI agents increasingly act on our behalf?
Actually use our computers. Access accounts. Write and deploy code. Make API calls. Even moving money.
That’s exciting, but also… mildly terrifying.
Zach’s framing really stuck with me: this isn’t just a security problem. It’s an authority problem.
If an AI agent already has access to everything, then “human in the loop” might not mean as much as we want it to. The pop-ups we see from Claude asking for permission are fake prompts. They are not hard-coded guardrails that prevent it from doing something.
The real question is: can we build systems where AI gets the freedom to be useful, but humans still hold the final approval over anything that really matters?
That’s where Passport Prime comes in.
The idea is that hardware wallets may evolve beyond just protecting Bitcoin keys. They could become dedicated devices for approving high-stakes actions across your digital life, from Bitcoin transactions to a myriad of AI agent permissions.
In the episode we talk about:
Why rogue AI agents are a real concern
How Bitcoin-style self-custody applies to AI
Why your phone might be a bad place to keep your most important security
The limits of “human in the loop”
What Foundation is building with Passport Prime
Why sovereignty gets more important as automation gets better
This was one of those conversations where the topic felt futuristic, but also weirdly immediate.
AI agents are already here. The question is how much authority we’re comfortable giving them.
Watch the full episode here: LINK
Would love to hear what you think of this one.
- Jake