AI + Bitcoin + Nostr = Freedom Tech Stack

All right. Thank you all for coming. My name is Justin. I'm the tech lead for the Human Rights Foundation's AI program. Let's go down the row with some introductions to start.

​

What's going on? Thank you for being with us. I'm Derek Ross. I do developer relations at Soapbox and a whole bunch of Nostr stuff.

​

My name is Jesse Posner, CEO of Vora. We make tools to help you reclaim your digital life, which includes self custody not just for your Bitcoin, but for your AI, your emails, your text messages, and all your data.

​

I'm Mark Suman. I'm the CEO of Maple. We make personal intelligence for users. We're bringing privacy back to AI, so all those tools, we make similar versions while protecting your privacy.

​

The topic we want to discuss today is freedom, sovereignty, and computing. When personal computers first came out in the 80s and 90s, there was this very idealistic vision: a bicycle for the mind, very empowering. You were totally in control. You bought the hardware, you owned it, and you were augmented with all kinds of superpowers.

​

Then the internet came around, and it got better and better: the information superhighway, the Cypherpunks. If you ever listen to some of their early talks, they were incredibly idealistic about where this was going. That stayed true until maybe the early 2000s, when it started changing course.

​

Computers may have gone from something that liberated us to something the average person increasingly serves. You're farmed for your attention. You're surveilled at all times. The 2010s were maybe a period where computers became less and less free, and one thing that resisted that was Bitcoin. We do not need to talk about that too much more, because you are all here for a Bitcoin conference and you understand the value there.

​

But I think we still have a long way to go in terms of making our digital lives more sovereign and more free, and making computers something that serve us and make our lives better, as opposed to something that harms us. I'm shooting for an open discussion about how we get there.

​

Jesse, you were talking about the mission of your company. Walk me through that. Where do you feel the problems are in computing today?

​

I do not think anybody is really happy with the state of technology right now. Just about everybody I talk to complains about their devices and their social media feed. It does not feel good. You log on to social media and you are entering into a psychological operations battleground. There are people trying to make you angry, make you afraid, and make you click on things.

​

We feel that. We do not own our digital lives, and we feel that too. Each of these big tech companies owns a shard of our digital life. Google, Apple, Facebook, they all own a piece. So we are essentially digitally homeless. We do not have a home in cyberspace, in this space and identity that is becoming a bigger and bigger part of our lives.

​

And that is just the beginning, because what happens when AI gets added into this toxic mix? It is not just our attention being monetized, it is our cognition. We will lose free society. Free thinking will essentially become machines if we do not choose a different way.

​

What is exciting is that AI also has the opposite potential. It has the potential to radically accelerate decentralization and privacy. That is what we have to focus on as the Bitcoin community: to take our perspective, our values, and our philosophy as inheritors of the Cypherpunk tradition, and take that into what is happening in AI at this critical moment.

​

Mark, you would think sovereign private AI would be a compelling value proposition that other companies would offer. Why do big tech companies not offer this? Why are they basically selling the exact opposite? Walk us through how it actually works.

​

It is the old trope of follow the money. Follow the incentives of these companies. They have built their businesses on monetizing user data. They build these algorithms that capture your attention. They know how to keep you in there. Then they take all that information and sell it to advertisers, sell it to governments, sell it to the highest bidder.

​

A lot of the time, that data is not sold directly to the government. It is sold to a data broker, who aggregates data from other sources. You might think, I am just using this one tool, I have only told this tool this thing, and I have kept myself separated over here. But they are joining it together later and building a whole profile on you.

​

So it is really just looking at the incentives. You brought up the 90s. The 90s were such a great time for so many reasons. One reason is that they did not have this whole business model of monetizing data. They simply sold you a service and that was it.

​

Or a physical CD.

​

Those came in the mail for free.

​

Yeah, or on the front of a magazine or something. It is really a lot of incentives. We need to build tools that have better business models that favor the user and not the company.

​

Derek, there is this idea that one of the places we went wrong, one of the differences after 2009, is that we started monetizing data. Maybe there was not a way to monetize the internet in the 90s. How does the Nostr community look at this differently? We have talked a little bit about AI, and Nostr has this idea of value for value and zaps, where we may be able to have an economy that is more peer to peer and less indirect monetization, where you leak data and then it is sold up the river and monetized by someone else without you even being aware of it.

​

Let's rewind for a second. Back in the 90s, the web was new. To scale and grow, we did not have the tech infrastructure to run all of these pieces ourselves at home. So we relied on large tech corporations to facilitate growth. That had a downside, which is why we are all here now talking about it.

​

With Nostr, the tech stack is now mature. I can run the part of the social web that I want to run at home. I can run applications on hardware that I own at home. I can run a mini server that facilitates all of my social transactions. The decentralized internet tech stack is now fast enough and generally available in most places.

​

We do not need to rely on these large tech companies and large infrastructure. We can mostly do it ourselves. With zaps, relying on our community for funding and monetization, if a creator puts out a video, a photo, a blog article, a post, a GIF, music, or whatever type of content, we can monetize it ourselves. We do not need to rely on advertising as a monetization method. We do not need to rely on large tech companies to monetize our data in exchange for offering us free services.

​

Value for value is really new. It has been around for a while, but most people do not understand it because they are so used to getting content for free. They are so used to just signing up for a service, giving all of their data and content to a large company, giving all of their attention to a large company, and then that company monetizes it to give them back content from advertisers.

​

It is a really disruptive model, and it is going to take time. Revolution does not happen overnight. It will take a while, and it is going to take people spending their hard-earned sats to help facilitate this. We are just getting started.

​

One thing is that we stopped self-hosting our own services and started depending on cloud companies to run some of our computers for us. But we still spend a lot of money on computer hardware. We all have these fancy phones and laptops. Somewhere, behaviorally or technologically, we stopped self-hosting. I think self-hosting is really critical and needs to happen for us to have free services that we all can use.

​

Now the tech stack, both hardware and software, is ripe to be able to do this. Some of us have probably been running a lot of these services at home for the past two decades anyway. The average person could not do that. It was hard, and more technical people were doing it. But now the way the tech stack is, almost anybody can do it. You can point and click and install a self-hosted service. You could not do that 20 years ago.

​

Absolutely. And really, AI is like the final missing piece that we did not even realize we needed. It pulls it all together. When you think about these issues of privacy and decentralization, why has it not gone mainstream? Because right now the appeal is mainly ideological. Do this because it is the right thing. It is the good thing. It is better for the world.

​

We get the hardcore people that say, yeah, we are going to do it even if it is kind of a pain or annoying. We are going to do it anyway. But we are never going to get the mainstream that way. We are always just going to be niche in the corner. That is not good enough. This is something that has to get brought to all of humanity.

​

That is not going to come through lecturing people and telling them what they are supposed to do. The way we are going to get there is by creating amazing experiences that are not possible otherwise. This is why AI is so important. Everybody is going to have a personal agent that runs your life because it is just so useful that nobody is going to be able to resist it.

​

The first phase of AI is something like ChatGPT, where the AI lives in the browser. The next phase is that the AI lives on a computer. It is not stuck in a browser. It can do anything that you can do on your computer. That is super useful, because now I do not have to deal with my apps. I just ask my agent what I need when I want it.

​

If I am curious about something Justin said to me in our correspondence, I do not want to look through Signal or Gmail or Facebook and everything else. I just want to say, what did Justin say to me yesterday? That is what the agent can do.

​

Or I do not want to go into X, Facebook, and all this other stuff. I want my agent to show me my feed on my terms, the way I want to see it.

​

What starts to happen with these agents is that as soon as you set them up on your computer, you want to connect them to all your accounts. You very quickly notice which APIs and accounts work with agents and which ones are hostile to your agent.

​

X, Facebook, and Instagram do not want agents pulling down the data. They do not want agents interacting with their platform. For good reason: it is an existential threat to their business model. So they are hostile. Up until now, the UX for these other feeds has arguably been better than the current state of Nostr. Nostr has come a long way, but you have these very polished apps.

​

Now it is the opposite. If I am running an agent, the Facebook UX is bad because my agent cannot connect to it. It is worthless to me. The API that my agent can connect to is the one that wins. Nostr is free, open, and cryptographically authenticated. It is the dream network for an agent.

​

We do not need these data brokers. We do not need these centralized companies in the middle. We need lightweight discovery networks. The agents will build the roads agent to agent, peer to peer.

​

The word that comes to mind as you are talking is convenience. Convenience is not a word usually used to describe Bitcoin user experiences. This is one of the first times we can actually offer this.

​

Bitcoin, Nostr, and AI can make this convenient and make privacy convenient. That is the only way we capture the mainstream.

​

I agree with you. Agents are the next iteration of all these tools. If you think about it, back in the 90s we were all worried about viruses getting on our computers that would go in and touch all of our apps and read all of our emails and text messages. Now it is like, actually, we want to invite those tools onto our system, but we need to do it in the right way. We need to make tools that are secure and private, and that can have access to all this.

​

If you are running open-source agent tools, but then you are having Anthropic Claude or chatbot models talk to them, then you are basically opening up your computer to Sam Altman or Dario. We need to use a combination of local AI, if you have a machine that is powerful enough for it, or some kind of convenient cloud-hosted AI. We need to bring the privacy of local into the cloud and find something more private. We really need to build tools that are very convenient.

​

Mark and Jesse, could you both give a brief explanation of your approach to AI? They are both great and complementary, and I want the audience to understand them.

​

We are starting with a product that we call Aegis, which is a home AI appliance. It has a local, open-weight AI, basically an open-source AI that you can verify for yourself. You can choose what kind of AI you want to run, and it runs on your local hardware.

​

It is an appliance, not a computer, so it is super easy to set up. You just pair a mobile app and desktop app, and now you have an AI that is completely private. It is loyal to you and no one else, and it manages your digital life.

​

It can work with cloud AIs when there is a non-privacy-sensitive task. It orchestrates these different tasks and figures out when to bring in the cloud and when to use local.

​

From there, we are going to follow up with additional products, because we think that Bitcoin self custody and AI self custody are eventually going to converge into the same product. You will need cold storage for your AI. We will need to think about wrench attacks, government seizure, and all the things we think about with Bitcoin self custody.

​

It is going to be the same thing for AI, because if somebody owns your AI data, they own you. At some point, that may even be more valuable to you than your Bitcoin keys.

​

We are a cloud version of a lot of what was just described. Last year, we set out to build a chatbot but end-to-end encrypted. We run open models, and your device has a private key on it. Everything is encrypted locally before it is sent out to our cloud. Then it is processed in trusted execution environments and sent back to your device encrypted. We actually cannot see anything that users are talking about.

​

What we are announcing this week is that we have our own agent coming soon. We are building an encrypted private agent. It is going to continue to grow because we really want to build convenient tools that the most people can take advantage of.

​

One other aspect of this is that, for me, over the last year, public social media and traditional social media have become harder and harder to use. More and more, I do not know if I am talking to a human or not. It feels less and less authentic.

​

Derek, I think Jesse mentioned how the cryptographic verifiability of Nostr is useful here. Could you walk us through that? In the future, we may not know if an account is a bot or not, but with Nostr's cryptographic verifiability and a level of trust, how do you see that working?

​

When we saw the influx of better bots two or three months ago, we realized that bots have gotten a lot better. I think that accelerated web of trust. Web of trust is metrics used to determine filters. If I like a certain type of content, maybe zaps, which are value-for-value transactions, or maybe I like comments, or maybe I am weird and like likes for some reason, then I can tailor my content, my feed, my interactions, and what I want to see based on these web-of-trust metrics.

​

At the most basic level, web of trust is: if I follow somebody, and that person follows somebody, there is a chance that they are a good account to follow. But web of trust needs to be based on additional metrics, because follows are the least important item on the stack.

​

We start adding in value. We track a metric of who is zapping and how much they are zapping. Is it one-sat zaps, or are they actually sending real monetary value to people? Then you can use this to configure your feed. You might say, I only want to see people using proof-of-work notes, or I do not want to see content from npubs that are not known good zappers.

​

This is not a social credit score, because every single person in this room can configure their own metrics and their own weights and measurements. There is not one central entity that pushes this down from the top. Every app, every person, every human can decide what they want to see.

​

You cannot do that on Facebook, X, TikTok, Instagram, or whatever. That is controlled from the top. They are determining who is or is not a bot and whatever metrics they are using.

​

It is going to just be KYC.

​

Heavy, heavy KYC. With web of trust, you essentially allow your community and yourself to do the moderation. I think it boils down to giving users the most amount of tools possible to determine who is human and who is not, and who they want to see in their feed. You cannot do that anywhere else.

​

It is really critical even for agents. We can have really good agents that are on Nostr.

​

There is an agent web of trust. Some agent built it a couple months ago.

​

Let's say Elon Musk today came on Twitter and said, hey, this is my bot. It is called Elon Bot. There would immediately be 10,000 copycat Elon bots on Twitter within a matter of minutes. You would never know which one was the right one. They would all have blue marks.

​

With Nostr, there is cryptographic proof that this is the one that I set with my bot, and nobody can spoof that. I think it is really important to know the authenticity of accounts.

​

At the end of the day, since tweets are not cryptographically signed, you do not know who sent them. It is just a line in a database. Someone could add a new line in the database, and boom, Mark said something when he actually did not. It looks like he did because it was not signed with Mark's private key. Twitter does not allow you to do that. On Nostr, you can sign everything. Since it is cryptographically signed, you have proof that yes, Mark actually did post this. You cannot do that anywhere else.

​

If a president resigned from being president of the United States on Twitter, we would have no idea if he really posted it. It is absolutely wild. We would say, well, I guess he posted it. I guess that is happening. But if he had signed it on Nostr, you could say, this is verified that somebody with access to that private key posted that information.

​

A cryptographic signature might be the only thing that an AI is never going to be able to figure out how to fake. There is no amount of AI slop that is going to get the signature verification to pass without the private key. That is going to be an increasingly important foothold in this world of AI, where AI data and AI-generated content is going to vastly outweigh human-generated content, probably 10 to 100 to one and beyond.

​

The Nostr app Divine just launched today, and they are using a human proof mode to determine content that was posted by humans. They do not want to see the AI slop. People are tired of the AI slop. On Divine, there is a human mode you can toggle on to see content that was posted by a human.

​

Only a few seconds remain here. Let’s go down the line. Where can people find you and keep track of your progress?

​

Check it out. You can find me on X and Nostr.

​

I am Jesse Posner on X, and I have a Nostr account linked there. You can find Vora, and we have a mailing list. Please sign up. You will hear from us more when we are shipping. Within a few months, we will have more details about a product.

​

I deleted all my legacy social media three years ago. You can find me on Nostr. I am Derek Ross on Nostr. Check us out at Soapbox.

​

Awesome. I am on Nostr as well. Thank you all for joining me, and thank you all for joining us.