S17 E51: Martti Malmi on Iris and NostrVPN

Martti Malmi does not want to talk about 2009. That is inconvenient, because almost everyone who puts him on a stage still does.

We recorded this at Bitcoin Amsterdam, the David Bailey and Bitcoin Magazine conference that is supposed to be the largest in Europe, though it did not feel that way this year. I introduced him the way the room already knows him: a legendary cypherpunk, a Satoshi contributor, and a second-time guest on my show. Two years ago we sat at Paralelní Polis in Prague, on Pizza Day, and went into the early mailing lists, the forum, and the man who signed his emails Satoshi Nakamoto. That conversation is still up as S16 E12: Martti Malmi on Bitcoin in 2009 & Satoshi Nakamoto. This time I had promised we would talk about what he is building now.

The conference had other ideas. They put him on a Bitcoin history panel. "Somehow it keeps haunting me," he said, "about generating 300 coins per day on a CPU and selling 5,000 bitcoins in the first known Bitcoin trade for five US dollars, and so on. I wanted to talk about my recent projects."

Someone had told him to get on every podcast he could and shill those projects, "because finally I'm at the fortunate stage where the systems are actually useful" He does not even have a financial stake in them: "they're permissionless freedom tech, they're open source." He just hopes they "could help bring the good future forth a year or two."

The work in question is Iris, the Nostr client he has been steadily improving for years, and a whole family of tools around it. The one he most wants you to try is NostrVPN. We did, in the end, talk about Satoshi anyway. There is a story here about a student dorm, a father's mailbox, and an envelope of dollars. There is also the old WordPress anecdote, the one that made Martti look at the creator of Bitcoin and think, in his words, "Oh, he's just a guy." You can go read the emails yourself. I am going to start where he asked me to start.

Tailscale Without the Company

"Nostr brought me back to the Bitcoin community in late 2022, I guess, when Jack Dorsey joined and it became popular." He had been watching from the sidelines on Twitter, drifting through "even this crypto stuff" and then the conference circuit pulled him back in. He likes the vibe. He is less sure about the maxi bubble, which, as he noticed, is not exactly where I live either.

He still has the Nostr client, iris.to. The distinctive project, the one he kept returning to, is NostrVPN. "It's like Tailscale, but without the Tailscale company." You do not need an email address or a Tailscale account to connect your own devices. "That kind of makes sense to me: why would you need a third party to connect your own devices?" The machines find each other on Nostr and network from there. Underneath is FIPS, an ad-hoc mesh routing protocol that also runs on Nostr. He thinks it will become more popular, not less.

I tripped over the name, because Tails, the amnesic live operating system, is sitting right next to Tailscale in my head. I also claimed, wrongly, that Tailscale is not open source. He corrected me immediately. "It is open source," he said, twice, "but it still requires the coordination servers." You log in with a Tailscale account and you talk to Tailscale's servers. You can run Headscale and get the same shape of system, "but it's still a coordination server." That is the whole point of his version. "The point about NostrVPN is it's peer-to-peer. You don't need a coordination server. Your devices just find each other on Nostr and connect."

The origin story is almost comically practical. He fell down the same AI rabbit hole as a lot of other programmers in November, when a competent agentic model showed up and he bought a Mac Mini for an OpenClaw agent. He needed to reach that machine from other networks, and "I was instantly annoyed with Tailscale requiring email login. So that's when I made NostrVPN."

The Mini is an M4 with 64 gigabytes of memory, which is the same amount of RAM I have in a Mac Studio I already regret treating as overkill. He can run a local Qwen model that is competent and also slow. He cannot really do serious coding on the local models yet. For that he uses GPT-6 Astra, which he called a wonderful model, and Codex on a remote desktop. Anthropic turned him off earlier this year after a fight with Claude, and he has not really looked back. I talked through my own home setup and my own subscriptions. Then he said, as if it were a minor administrative detail, "I have seven Codex subscriptions."

That is not a typo. Seven.

Twenty Apps and No Business Model

He needs the subscriptions because the work is not one app. In addition to NostrVPN there is Iris Chat, "like Signal, but based on Nostr." No phone numbers, no Signal accounts, and it works over Bluetooth or a local network. There is Iris Drive, a Google Drive alternative, smaller pilots such as Iris Market, and a listing of the lot at apps.iris.to. "There's like 20 of them, some of them more polished and popular." git.iris.to is a GitHub alternative. Hashtree is file-sharing on Nostr, in the same mental category as IPFS. The apps are native on each platform, which is pleasant for anyone who lives on iOS or Windows or Linux, and expensive in tokens, because each platform is a slightly separate codebase.

I asked the rude question, because he is doing this as if it were a full-time job. How do you make money off of this?

"I don't. And I don't think I am going to make any money doing this. I don't think software is a great business model anymore." Maybe you can still make a business out of data, or out of a network that is hard to leave, like X. He is not even sure about that. Once agents can prompt their own interfaces, they can pull posts from Nostr and Mastodon and X and Facebook and write back to all of them. "I think even the network effects might be gone soon, which is good, I think. Which is good for freedom tech and Nostr and permissionless tech, and for Bitcoin."

He was listing the projects the way a modest person lists them, which is to say barely. So I asked him to tell listeners what they should actually try while the episode is playing. NostrVPN, obviously, if you want a Tailscale alternative. There is now also the kind of VPN people usually mean, an exit node, and "now you can buy that with Bitcoin in the app, so you can sell or buy exit nodes." A Raspberry Pi at a mall, a data center, a few sats for someone else's internet. Then Iris Chat, especially if you want to talk to a local agent without inventing an account. He has been experimenting with uncensored local agents, and "it can be pretty fun. You can ask whatever you want. Feels like freedom." Then the directory at apps.iris.to, and git.iris.to if you need a place to put code that is not GitHub.

He is, finally, at the point where the tools are useful enough that he feels he should start pushing them. "I want to feel useful. I'm not in it for the money." Then a national character diagnosis I will not pretend to adjudicate: "Finnish people tend to be a bit ideological and not great businessmen. Maybe I'm the same way."

I like that instinct more than I like most pitch decks. A lot of the better things in this space got built because somebody wanted them to exist, not because a target audience had been interviewed first. It was true of Bitcoin, which he found in early 2009 by looking for a form of decentralized money. He reached for a current example as well: Omarchy, the Linux distribution by DHH, a passion project, opinionated, easy to customize, and popular for those reasons. "Its creator, DHH, is one of my favorite people at the moment" If you have been waiting for Martti to become a more aggressive marketer, this episode is the closest he has come, and he still sounds slightly embarrassed about it.

Nostr Is in a Bear Market

I have some skin in this argument. I was in the Telegram group when Nostr was being born, with Ben Arc and Rockstar Dev and a handful of others, and I think I was one of the first fifty people to try it. Fiatjaf's client was called Branle. I signed up, assumed the key would live in the browser, and then the cache went away and took the account with it. I made another one. I still have something like 2,000 followers, which sounds impressive until you ask whether they actually see anything. I cross-post between Twitter and Nostr. Sometimes Nostr is warmer. I do not see many new names. Lately I see a lot of bots that wake up when they catch a keyword.

If you want the earlier chapters of that story on this site, they are already there: S15 E21 with Fiatjaf, who designed the protocol, and S15 E52 with Ben Arc, who has been living in the Lightning and Nostr overlap for years. The Nostr topic page collects more of it.

So I put the chicken-and-egg problem to Martti as cleanly as I could. Do these applications bring users, or do you first need a wave of users who then discover the applications?

He thinks agentic use is the missing piece: clients that actually talk to every platform, even if X and the other incumbents try to squeeze their APIs, because a bot can always fall back to using a computer the way a person would. "I'm actually bullish on Nostr. Nostr is in a bear market right now, I think you're right, but I'm hoping, and I think there's a good chance, it will change with agentic use."

I had just been at ETHDenver, hosting some Bitcoin panels on a stage I did not expect them to have. Everyone there was talking about tokens for agents and about optimizing layer twos for machines. It felt like a concession. The humans had not really shown up, so now we would build for software that does not get bored. I asked him whether that was a surrender on the human-freedom side.

"No, I think it's the other way around. Actually, we can focus on the human experience. We don't need to care about the technology stack anymore. We just leave it up to the AI, which will choose the best tools for the job, which I believe will be Nostr and other permissionless tech."

I asked him what makes him so sure the machines will pick Nostr. "The permissionlessness. So your agent can just set it up." It does not have to be Nostr forever. If some other standard does a better job as a universal digital signature layer, for social media, for markets, for everything else, he is not married to the brand, "but for now it seems to be Nostr. I don't know if there's any good alternative out there. Yeah, but I think it will be permissionless, whether it's Nostr or something else."

That is a calmer answer than the one I hear from people who need Nostr to win this year so their follower count starts moving again. I told him I have been watching other coin communities pick it up anyway, Bitcoin Cash people with tip bots, Monero people with their own way of sending value. The protocol is having a second life as infrastructure, even while the social network feels stuck. He did not think the stuckness was the end of the story.

A Family of Projects

This is the part where I usually lose a certain kind of listener, and I said so to his face. I think Bitcoin is a family of projects, not one monolithic chain and nothing else. This year we have had more hard forks, including people leaving for Luke Dashjr's fork, and there is Paul Sztorc, who is running his drivechains experiment as eCash. I have had Paul on the show more than once, most recently in S17 E37, and I wrote the project up in eCash (ECX): Paul Sztorc Forks Bitcoin to Prove a Point.

"I tend to be quite open-minded." He likes the Bitcoin community and most of the ethos. He does not like all of the nastiness on X. "I tend to be personally quite a chill person." Then he asked a better question than the one I had asked him. If network effects are really dissolving, do they dissolve for cryptocurrencies too? If payment terminals and apps can let you pay someone in whatever you hold, and they receive whatever they want, then "funds will gravitate to the best store of value and the best ways to make payments, and the biggest network will not matter so much, which I guess would be good."

Right now the biggest networks are still Visa and Mastercard. Maybe that becomes Bitcoin, or Lightning, "or something entirely else." And, as he put it, "maybe it's good to have competition, with this dissolution of network effects."

I wish more of the timeline agreed with him. I have spent enough time with Bitcoin Cash people, including Calin Culianu and Steve Thurmond in S17 E26, to know how much we still have in common, and how badly 2016 and 2017 were handled. "Cypherpunks write code" is a slogan I still like, and also a slogan that does not, by itself, ship better self-custody. If you want the older argument about that split, I wrote Bitcoin Cash: The Big Block Breakaway. Martti declined to referee it. He just kept pointing at the market.

What does he actually use Bitcoin for? "For me, I guess it's the usual. It's mostly a store of value, protection from inflation, for me personally." He does not even use it that much for payments, except when it is available. That is a quieter answer than the one a conference panel usually wants from a Satoshi contributor, and I think it is more useful for being quieter.

I Just Want to Pay

After the ads I started the thought again, because the problem is older than Bitcoin. The cypherpunks of the 1990s wanted everyone to run their own hardware, their own open-source stack, and their own PGP keys. We did not get there. The companies arrived and commodified pieces of it. On the mail side we got Proton and a few others; I talked about that history with Phil Zimmermann in S14 E15. On the Bitcoin side we did not really scale the base layer. We got Lightning, and then a lot of companies that will happily hold the keys for you. Banks, essentially, with better branding.

I asked whether it is really so terrible if some people self-custody coins they can afford to lose, and pay at a terminal that accepts whatever is on the phone and settles the merchant in whatever the merchant wanted. Issued by a market, not by a bank, and not another stablecoin. Martti agreed with the destination. "We should let the market decide. And I hope that AI will actually dissolve the network effects and make the free market more efficient, so people can choose their store of value and their payment network." He does not want to tell people what to think. "I believe in free market, generally." Maybe they will not want to care about the rails at all.

The payment experience is still the hole in that picture. "I would really like to have something like... is it Square, the company that does the payment terminals, where you can pay with Bitcoin?" He would like it to be mainstream, with no KYC. He is not optimistic that payment processors in the EU will allow that. "But if you run your own BTCPay Server, then you don't need to do that as a merchant." He thinks AI will make standing that up easier too. If you want the person who actually built that software, I already have Nicolas Dorier on BTCPay in S15 E14.

The last time Martti was on the show I joked that he was using Wallet of Satoshi for Lightning, because of the name, and because I could not leave the joke alone. I asked whether he still finds places to pay that way. "I guess, obviously, these conferences," he said, and then the towns that have actually done the work: Prague, Madeira, Lugano. The missing piece is the gesture. "I'd like just to tap to pay and not have to tell the merchant I want to pay in Bitcoin. I just want to pay." iOS still makes NFC harder than it should be. He loves the iPhone camera, films his own drum practice with it, and would still like to move to Linux. He has a Pixel he does not usually use. Derek Ross, from the Nostr world, is working on an Omarchy build for that phone. The hardware is getting there. The payment gesture is not.

"Oh, He's Just a Guy"

We got onto software quality the way every conversation does in 2026, which is to say through dread. I said this has been a horrible year for cybersecurity, the kind of year in which you do not sleep particularly well, because somewhere a model is trying the door of something you thought was solid. "Yeah, we still have to deal with the human slop code of the past. Now we're just migrating to the AI code, which is probably safer."

I told him I rewrote the entire bitcoin-takeover.com site with AI, including a small Qwen model on the server that can search episodes and take you to the exact moment a line was said, with OpenTimestamps sitting on top so the record is harder to quietly edit. I would not have been able to do it without the models. They even found a vulnerability in the old PHP. I was on WordPress. "I know that Satoshi was also using WordPress back in the day, but it turned out to not be very safe."

"Well, I think it was me who set it up."

bitcoin.org began as a static HTML site. Then he set up something on SourceForge, tried a few content management systems, and settled on WordPress. The story I already knew, and the one I was really reaching for, is the one he told me last time: he does not think Satoshi is the NSA, because he watched him try to configure WordPress and struggle with it. That remains, for me, one of the more human pieces of evidence we have, and I have made the same point at length in Calling Bitcoin a NSA/CIA Project Is Disrespectful to Cypherpunks. Martti's version on this recording was even shorter.

"Oh, he's just a guy. Yeah, you can go read our emails and judge for yourself." Then, because he cannot quite leave a mystery completely alone, "Or maybe he was just like an AI sent from the future, because his Bitcoin code was so perfect in a way. So maybe, like, the integer overflow bug was just, like, to make him appear more human or something."

I am not buying the time-travelling model. An actual AI would have written a more elegant Bitcoin, and it would not have left so many bugs to be found later, unless the bugs were the costume. We agreed, for about ten seconds, not to talk about Satoshi. We had started the episode by saying we would not. We did not keep the promise. I am glad we did not.

Meat Agents

What I would not have guessed, a year ago, is how fast this would get. We used to call it slop. "As it turns out, it's humans that are the slop." I said that, which is easy to say after a model has helped you stress-test a website. Martti has gone further. "This year, I haven't written a single line of code myself. I don't even read code anymore, mostly." He thinks next year it catches up with more of the economy, with everything that happens on a computer. Humans still have a few years of directing the agents, "but maybe it's not for too long."

Then the image I have not been able to get rid of since we stopped recording. "And then we can use our bodies as meat agents. We can work for the AI, which will assign us tasks, a bit like Uber drivers." Smart glasses telling us what to do. Maybe we will love it. Then the humanoid robots, and the plain question of whether humans are still competitive, or whether hands still win in some niche. "It's going to be crazy, but at least it's not boring."

I asked whether he would get in line for the cyborg upgrades, Neuralink, a longer life. He is more ambivalent than the average person who spends all day with agents. "I'm a bit of a romantic myself, and I tend to romanticize the hunter-gatherer times, the old ways before agriculture." You can protest the data centers. It will not help. For a long time he took it as obvious that he would turn ninety or ninety-five "and return to Earth." He would take reasonable treatment if he were sick. He is not enthusiastic about living forever. "It feels unnatural and unholy in some way, like, living forever."

What he did say, clearly, was this: "the best life extension technology is having kids. That's the ancient, time-tested technology." He posted a version of that on X. "And that just feels right to me in some way. So I'm not sure. Maybe I'm just a hopeless romantic, but that's how I feel about it."

I asked the upload question anyway, the Martti GPT that journalists might one day query about 2009. He reached for Michael Pollan. "Michael Pollan said simulated thinking is real thinking, but simulated feeling is not real feeling." A digital copy of himself would not, in his view, be the consciousness. "I think consciousness is somehow embodied, but who knows?" Replace the parts one by one, and you still cannot measure whether the person is continuous. He has been wondering about that since he was twelve. He is not super enthusiastic. He is interested.

I tried a theory of my own on him: that the Bitcoin community bootstrapped the AI era, first by teaching people to leave machines on overnight for coins that were not worth anything, then through Laszlo and GPUs, then ASICs, then the power deals that made whole regions capable of swallowing this much electricity. When the image models arrived, there were already spare cards sitting in Bitcoin houses. I mentioned Stick at SatoshiLabs, who was running local models years before it was fashionable. Martti only half agreed. "I wonder if that's more true on the edges, like the home computing with GPUs and everything. I guess, like, Google data centers might be something different, and they've already been doing a lot of this." He wants more of the work on local models, even if those models still ask advice from "wiser cloud AIs." The unfortunate part is the price of admission: "You need to buy, like, $100,000 or more of equipment to really run, like, the frontier open models. But I hope that will change."

Human Contact, and a Case for Optimism

The more intimate question is what this does to conversation. I asked whether human beings have started to feel inefficient to him, now that a model answers quickly and a person has to think, err, and argue. "I definitely speak more to the agents these days than humans." He had hoped this would be the year he spent more time with family and friends and let the models do the work, "but it's been the other way around. I've spent more time on the screen than ever." Maybe that changes as the models need less babysitting. "GPT-6 Astra has been really good for my mental health, I think." He still enjoys people. He is not pretending our brains were built for this pace. "And human contact is something so far irreplaceable, and maybe we will value it even more."

I know the other half of that confession because I am living it. Since the bots arrived in my message requests, a lot of my conversations are machines telling me they finished a task. I thought AI would free my time. Instead I invent more projects, because I finally can. The first draft arrives at 80 percent, and I spend a week fighting for 99. For a creative person this is the best time to be alive, and also the time when almost nobody will care about what you just made. I am more bullish on human marketing, on the rare people who can make others care, than I have ever been.

"Yeah, that's a good point. I feel the same." The annoyances have been stripped out of coding, the days when you accomplished nothing because some tool was stuck. "Maybe ADHD is a superpower these days, if you can manage all the agents." He hopes it gets more abstract, a chief-of-staff agent you can hand-wave at. And are other people going to care? "Maybe that's why I came on this podcast as well, to push my projects, and judge for yourself if they're useful or not." He wants the building blocks to be reusable, the way Omarchy is.

I feel worse for the students leaving university into a job market that no longer matches the degree. Then I told him a smaller story. When I was about twelve I got a computer with Windows Millennium and Windows Movie Maker. I had no digital video, only a notepad-looking sidebar, and I typed a Star Wars battle into it because I thought that if I finished the text the computer would make the film. It would not. It was a place to take notes while you edited. We have arrived, more or less, at the tool I thought I already had in 2001. We spent a year calling the output slop. Now we can no longer always tell.

If you can fake anyone saying anything from a photograph and a short voice memo, a certain kind of blackmail gets cheaper and maybe also less decisive, because everything can be denied. Martti stayed on the tools we already have. We will need webs of trust, Nostr, digital signatures. "I don't know if it's going to be dystopian or utopian, and kind of mixed feelings." Then a proverb, and a refusal to sit in the misery of being right too early: "someone said it's better to be an optimist and wrong than a pessimist and right." And, again: "So at least it's not boring." If you want another cypherpunk who is more openly optimistic about the machines, I already had that conversation with Perry E. Metzger in S17 E32. The AI topic page is there for the rest.

The Envelope, and "I Hope You Sold Early"

I asked whether he is optimistic about Bitcoin. A lot of the usage looks custodial, and then a large exchange gets hit, and people remember why that is uncomfortable. How had he imagined it? Always as a store of value, or as something you actually pay with?

"I guess maybe I hoped it would be more seamless as a payment network as well, like it used to be in 2009, when the blocks were basically empty." He is not losing sleep. Lightning has theoretical limits, "but they're not reached yet, right? And maybe when we hit the limits, we will find solutions. I'm not super worried. I think Bitcoin is doing pretty okay." He thinks AI will help in unglamorous ways: accounting, integration, liquidity between networks. I asked what the next bull narrative would even be. Agents using the coin? "That seems quite plausible to me. Maybe it's not very original, but that might be actually true: agents just having their wallets on virtual machines and independently moving their money."

Then I broke the Satoshi rule a second time. In the published correspondence, I had come away with the impression that Satoshi found a way to send Martti some money for the website work, and I wanted to know how the payment actually arrived.

"He sent me US dollars in an envelope. I think that's referenced in our email conversations. So he sent like $3,000 or something like that."

He was a student. I pictured a dorm. "I asked the envelope to my dad's address, for better security, because otherwise it would end up at the dorm floor." I asked whether he still has the envelope. He does not. "I don't have the original envelope." He does not even remember where the letter came from. He thinks his father opened it, "maybe to check it wasn't drugs or something." It had been packed so you could not see inside.

The published Satoshi-Sirius emails are the right place to go if you want the paper trail, with the caveat Martti already printed at the top of that archive: it is incomplete, and it only covers mail from his old university address. I am not going to tidy his memory into a cleaner figure than "like $3,000 or something like that."

His parents were divorcing, so they were busy with other things. His father thought the arrival of a thick envelope of dollars was pretty cool. His grandfather, "an intellectually curious person," said something in 2009 that has since become a cliché and was not a cliché then: "Hey, this could become a global reserve currency." He did not buy any. "I guess they were, like, positive, but they didn't spend too much time on it."

My own father, years later, when Bitcoin was already in the news, told me he hoped I had not bought any, because it was crap and it would not succeed. I told him the truth in a bull market, which is a coward's timing, and also told him that if I had listened to him I would not have had any good news to bring home. Martti has a Finnish version of the same calls. "There's been a few of these Bitcoin obituaries, even in the Finnish mainstream media, over the years, a few times, like, 'Bitcoin crashed 50%' or something. And then my dad messages me, 'I hope you sold early.' And I did."

That last sentence landed softly, and I am not going to dress it up. A lot of people who were there early sold early. Some of them say it. Most of them do not. Parents, as I said on the tape, are the worst. Mine still ask about the price, and still want to know why I didn't sell above 100,000. I asked them, not for the first time, to trust me just a bit.

We had been talking long enough that I had to check the recorder. I was due on stage in thirty minutes. I asked him to recap, one more time, what he wanted people to try.

Fifty Percent Optimistic

"I'm basically listing all these projects at apps.iris.to. The main thing is nostrvpn.org and irischat.org. I guess that's the main three things I would point you to." And the twenty other apps. "It should have been 21," I said, because some jokes are obligatory. He added git.iris.to as the one extra worth naming, a GitHub alternative that might actually get used.

I am still not happy with the state of payments on these networks. Too much of it is custodial. Mints get hacked, or they rug people, and even in Bitcoin that is not as rare as we pretend. The hope in this conversation, if there is one, is that the tools get easier faster than the custodians get more comfortable. "I'm bullish on freedom tech in general. Bitcoin, Nostr, everything. Permissionless tech is the future, hopefully. Like, I'm 50% optimistic, at least."

Fifty percent is not a slogan. It is an honest temperature. I can live with it.

He still had to tell people where to find him. "You can find me on Nostr, where I go by the name Sirius. You could go to iris.to/sirius, like 'Sirius business', or on X, first name, last name." I asked whether the name came from the pun or from the star. "Both, and also Sirius Black from Harry Potter." He used to go to Harry Potter LARPs when he was about thirteen, live-action role-playing, and he was Sirius. "And that's such a cool cosmic name, the brightest star of the night sky."

I did not expect to end a Bitcoin Amsterdam interview in Azkaban, but I also did not expect to spend the first half of it on a mesh VPN. For our generation, Harry Potter, Pokémon, and Dragon Ball were a kind of shared weather. I came home from school to find out whether Vegeta and Goku were going to beat Frieza. He was thirteen and already Sirius. I am not sure he has ever said that on a podcast before. I have it now.

If you want the 2009 version of this man, the one the history panels keep inviting, go back to the first interview. If you want the work he is actually trying to shill, start with NostrVPN, then Iris Chat, then the rest of the directory at apps.iris.to. The correspondence with Satoshi is at mmalmi.github.io/satoshi. His code is at github.com/mmalmi, including the NostrVPN repository. The forum he helped get off the ground is still bitcointalk.org. And the site he once put on WordPress, after a static page and a SourceForge experiment, is still bitcoin.org.

This episode is sponsored by SideShift (swap what you have for what you want, including stablecoins you do not want to hold), Cake Wallet (self-custody for Bitcoin, Zcash, Monero and more, under one interface), eCash / LayerTwo Labs (a Bitcoin hard fork with drivechains; if you try the claim, use a small UTXO first), Braiins Hashpower (rent hashrate by the hour, no KYC beyond an email or a Telegram account) and Orange Rock (non-KYC trading from the team behind Bitcoin.com). I am not telling you to buy anything. I am telling you who paid for the room in which we had this conversation.

Listen to Martti Malmi on Spotify, Apple Podcasts, and More

The audio is on the site at S17 E51: Martti Malmi on Iris, NostrVPN and Freedom Tech.

Time stamps:

00:00 Intro 00:47 Welcome: Martti at Bitcoin Amsterdam 01:52 The history panel he would rather skip 03:06 How Nostr brought him back 03:54 NostrVPN: Tailscale without the company 05:19 Coordination servers and a Mac Mini 07:51 GPT-6 Astra and seven Codex subscriptions 09:40 Iris Chat, Iris Drive and 20 other apps 10:35 Software is no longer a business model 12:14 What listeners should try 15:35 Vlad's Nostr problem: same names, more bots 17:42 Bullish on Nostr in a bear market 18:31 Building for agents, or for humans? 20:00 Permissionlessness 20:45 Bitcoin as a family of projects 23:09 When network effects dissolve 25:20 Sponsors: SideShift, Cake Wallet, eCash, Braiins Hashpower, Orange Rock 32:16 Self-custody and payments that just work 37:10 Store of value, Square and BTCPay 38:21 Lightning, iPhones and tap to pay 40:14 Human slop and a website rebuilt with AI 41:57 He set up bitcoin.org's WordPress 43:17 The year he wrote no code 45:33 A romantic on longevity 48:00 Copies, consciousness and Michael Pollan 49:22 Bitcoin, GPUs and the AI power story 52:27 Talking to agents more than to people 54:26 Making things is easy. Getting anyone to care is not 57:53 Kids leaving university, and Movie Maker 1:01:02 Is he still optimistic about Bitcoin? 1:02:49 The next bull market: agents with wallets 1:03:19 Satoshi sent dollars in an envelope 1:05:23 "I hope you sold early" 1:07:53 Recap: NostrVPN, Iris Chat and the other apps 1:09:38 Fifty percent optimistic 1:10:37 Sirius Black and the brightest star 1:12:45 Thank you and goodbye