TFTC: A Bitcoin Podcast - #721: The Human Skills That Matter Now That AI Can Code with Calle

Episode Date: February 28, 2026

Marty sits down with Calle, prolific Bitcoin and Freedom Tech builder behind the Cashu eCash protocol, BitChat, and Clawi AI, to discuss the AI agent revolution transforming software development, the launch of Numo tap-to-pay for Bitcoin, eCash as the ideal payment rail for autonomous agents, and what it all means for the future of human creativity and work. Calle on X: https://x.com/callebtc Clawi: https://clawi.ai/ STACK SATS hat: https://tftcmerch.io/ Our newsletter: https://www.tftc.io/bitcoin-brief/ TFTC Elite (Ad-free & Discord): https://www.tftc.io/#/portal/signup/ Discord: https://discord.gg/VJ2dABShBz Opportunity Cost Extension: https://www.opportunitycost.app/ Shoutout to our sponsors: Bitkey https://bit.ly/4pOv2L4 Promo Code: TFTC99 Unchained https://unchained.com/tftc/ SLNT https://slnt.com/tftc Lygos: https://bit.ly/4koiJmB Salt of the Earth: https://drinksote.com/tftc Join the TFTC Movement: Main YT Channel https://www.youtube.com/c/TFTC21/videos Clips YT Channel https://www.youtube.com/channel/UCUQcW3jxfQfEUS8kqR5pJtQ Website https://tftc.io/ Newsletter tftc.io/bitcoin-brief/ Twitter https://twitter.com/tftc21 Instagram https://www.instagram.com/tftc.io/ Nostr https://primal.net/tftc Follow Marty Bent: Twitter https://twitter.com/martybent Nostr https://primal.net/martybent Newsletter https://tftc.io/martys-bent/ Podcast https://www.tftc.io/tag/podcasts/

you've had a dynamic where money's become freer than free if you talk about a fed just gone nuts all all the central banks going nuts so it's all acting like safe haven i believe that in a world where central bankers are tripping over themselves to devalue their currency bitcoin wins in the world of fiat currencies bitcoin is the victor i mean And that's part of the bull case for Bitcoin. If you're not paying attention, you probably should be. Kale, it's been a year and a month almost exactly to the date since we published our last podcast where we met in person in the jungle. And a lot has happened since then.

How have you been, sir? Oh, yeah. I've been doing great. And wow, that's been over a year. That's crazy. i wish we were in the jungle again though i've been looking out in uh 20 inches of snow uh the jungle is much preferred to this but for anybody who's listening who didn't catch that first episode doesn't know who cali is i'd be surprised if you don't cali is one of the most prolific and

active individual builders in the bitcoin and freedom tech ecosystem uh working on bitcoin the cashew e-cash protocol bit chat claw i ai and other open source dev projects you've got a phd in physics and i mean i've been following the sparse podcasts you've been doing throughout the year i think you were on citadel dispatch summer of last year uh and you're talking about this proliferation of ai tools and how vibe coding helps you build the android app for bit chat that is blown up talk about bit chat but i think um focusing in on what's happening now uh with claw eye ai and basically you launching agents to help you do your work you tweeted yesterday about a devops engineer that you launched that would typically cost 200 000 but you have it running autonomously

24 7. what uh what's going on yeah i i i don't know what's going on that the i i strongly believe that no one does where we're heading and it really feels like everything that we're doing right now is something that hasn't been done before so we are like 100 in discovery or uh in in a discovery mode or in a pioneer mode where you know any new idea on the table is basically is probably the first time that someone tries it out. As you made some examples, like I've been focusing a lot on AI agents recently, particularly focusing on OpenClaw,

and I like it a lot and try to make it as useful as possible and try to figure out ways how to integrate it into the projects that I'm working on to be more effective, more cost-effective, and faster chip more and higher quality code but we'll be talking about that i'm sure in this pot in more detail yeah i think i mean just dive into like how prolific this is for you somebody who actually understands like i was saying just before we hit record i'm having some sleepless nights pushing my open-claw agent to the edges and I barely know what I'm doing from

the systems admin back in engineering programming perspective at all and I can't imagine what it's like for you and in terms of like productivity wise and what you're able to do what's what how profound has it been for you well everything has completely changed and I would have never thought that I would be able to say that in my uh like programmer career so like many people i've been programming for decades at this point so i consider myself like a professional developer with lots of experience and and probably everyone has noticed by now that the work of being a programmer has flipped by like 180 degrees it has changed in a way that no one would have predicted even only two years

ago maybe so in the last one or two years maybe we'll start a little earlier so the way i see this in my mind is we have multiple stages of innovation in ai with regard to llms and agents so level one was using an LLM in a browser chat GPT that was the most pivotal moment in AI history so to speak when you're thinking when you think about how actual normal people interface interact with AI that was the biggest moment and I think also up to date comparing to the penetration speed of other technologies it has been the fastest growing and fastest adopted technology in history was chat gpt so um it has conquered the entire world that's level one you open a browser you ask a question you get a response and we've been you know doing that uh for a year or two

and also programmers started noticing quite fast that you can copy codes from your code base and put it into chat gpt ask a few questions get some code back and put it back into your code base so there were many manual steps still involved when doing development work using AI. And so after level one, we reached level two around maybe one and a half years ago, which is the introduction of agents. And agents finally gave the LLMs that were previously confined in the browser gave them kind of legs and arms to act on your computer so that is for example cloud code is probably the most popular one and then there's codex and so on from chat from from open ai what i use the most is open code which is a llm agnostic framework

fully open source really high quality projects so um that is most of the code that i write today goes to these tools so um and now i'm talking to other software engineers all the time about this and it's it's very fascinating to see what people think about this and how they perceive their their job changing over time and you have like you have a small portion what feels like 10 to 20 percent of developers resisting the change and they they don't like ai and they think it's it's fallible and it's not as good as they are and then you have this way larger percentage of developers who uh use it and try to increase their productivity and um and what i see is you know those who value the manual aspect of coding and i compare them to artisans so people who take pride in

having done the work themselves and also enjoy the process of doing so and see some mastery in it or try to perfect their mastery and skills in it have a harder time giving up the job to an ai whereas and then there is kind of like the opposite of that is prolific coders who've been in the business for decades at this point who notice that they can increase their output by and i'm not exaggerating when i'm saying like five to ten x which is you If you are confronted with an efficiency boost of 5 to 10x, then it's really, really hard to say no. And the only good reason that remains to say no is extreme safety. So you're dealing with Bitcoin, for example, and you're not going to have a machine just write all the code for you that deals with money.

or you're in like an industrial sector where it's the critical infrastructure or something like that where you don't like where you need the human in the loop and or those who you know really want to perfect their art so there's like still some extremely good

a very tiny number of human workers voters who can't be matched by AI and you know they are getting paid large sums of money these days to be like the one big expert in COBOL or something like that working in the companies but for the rest you know the adoption of agents has led to an incredible output boost and productivity boost and so that's what I've been focusing on for the last a year and a half or two which is like how can i make use of these tools to improve all of these projects that i'm working on a lot uh you know a lot faster and also help all my peers to also

get on that train and basically making sure that anyone in our in our cashier ecosystem uses ai tools and realizes that if you're a software developer today and you're not using ai I'm willing to put money on the table that you will lose your job in five years. Like you won't have a job in five years. I am 99% sure that this is what's going to happen to most people if they don't adopt this technology and the economic realities will force you to do so.

So that is level two. It has changed the industry completely, which again is something that I would never have thought that I would experience this. And level three, now this is the level that we are currently in. This is the next evolution of AI agents are these proactive, always on running agents, personal AI agents that have a memory and a context about you that run either on your own home computer all the time or on a server somewhere in the cloud all the time. you're the only one who has access to that one computer and the agent has its own computer can do anything on that computer and with that power and with a loop that closes kind of like tells

the agent to work around the clock and with all these ingredients combined we've seen what the future of personalized AI agents will look like a couple weeks ago when you know open call blew up And when I first discovered OpenClaw, I had, like, we're talking about this before the recording, as you said, almost like two or three sleepless nights, just trying to process what this means. Because when you try technology for the first time, and then you get this intuition that, oh, my God, like, this is so powerful, this, everyone will have this in one year, this will be everywhere in one year, because it's too powerful. It's like this chat GPT moment again, I think where, you know, the first moment you try chat GPT, you are like, sure, this is so significant that there is no way this won't penetrate everywhere. This is how I felt again with kind of the latest iteration of level three autonomous AI agents. And yeah, that's been on my mind a lot lately. yeah and what would you say to people because i've um been using open claw for like i was i'm

an og i was using it when it was clawed by i had to set it up uh in the cloud via cli i didn't have is it clawy or claw.ai i call it clawy clawy i didn't have clawy i didn't have like one click deploy so i set it up like a like a caveman in a vps and i've been using and i've similarly had this like holy crap this is so powerful everybody's going to be using this eventually but there there are naysayers out there like it it's like cloud code like i've been using cloud code and cloud co-work and it's the exact same from what i understand um what are the differences and what is the difference between just using cloud code or cloud co-work um uh sort of in in a silo versus is creating an OpenClaw and giving it access to Cloud Code.

And the way I understand is OpenClaw is like a harness that really adds a bunch of power to Cloud Code and other models, for that matter. Yeah, so we definitely have to differentiate between what's the LLM, and that can be Cloud, Sonnet, or Opus, or GPT, or whatever it is, like the brain. And then the body, which is the cloud code, CLI, for example, or OpenClaw. So kind of where do you put that brain into?

And the biggest difference between the level two coding tools and the level three personal agents is the loop. Personal agents are alive all the time. They don't care whether you leave your keyboard or not. it's not meant to interact like they're not meant to be interacted with using that one single keyboard that you have on your computer and when you close your computer it's gone again but it's rather meant for something that that is active all the time can do something in the background and um has memory about everything you've done so far and that is what you know open call also does it it has a persistent memory sometimes it struggles with that memory

but at least you know tries to remember everything that you've been doing you can talk to it like a normal person via chat so you you usually interact with it with you know signal whatsapp telegram whatever you want and it can you can tell it to you know every day at 2 p.m make a report about all the activity that's been going on in this github repository for example this is one of the many examples that we were using in our dev organization which is we have one of these glories that basically just tracks our progress on github and gives us you know daily reports about hey these are the new pull requests that were open these are the new issues that were open sorted by severity and these people are working on that and there is no activity in this project

so maybe you should look at it or something like that you know having something like that that gives you automated responses every day for example it could be every hour it could be every every week whatever you want it could be news aggregation research on the latest topics that you're interested in or just something very banal and related to your personal life so there are clear differences in the capabilities between level two and level three i think you know it's very it's very normal that there will be skeptics about you know what is this useful for i've seen many technologies come and also go where you always have the same thing um and since we're so new and so early with this technology it basically the limit is your imagination right

now is if you cannot think of a way how this can be useful for you then think a little harder that's that's what i would suggest because i'm pretty sure everyone who deals with computers and does everyone as something that they don't like doing themselves or they would they would like someone else to do it like in the background for them and yeah so so you can probably come up with stuff yourself i'll just give you one uh one anecdote of mine is you know i've been on a conference and using like uber or bolt or something like that for for taxi rights so at the end of the conference i need to put all the invoices together for my accountant so they can you know be properly accounted for and i hate that step like it's it gives me anxiety just to think about that little

task that i have to do after every single conference is download all the pdfs from uber they're like have these weird file names and then i need to go into every pdf take out the date and rename it and send it to my accountants like these are like a couple of stupid steps that i need to do after every conference and this time i just like downloaded 15 invoices as a pdfs and i dropped it over to my chloe over signal and i told it like hey you know just you know look into the pdfs extract the dates rename the file send them back to me and i just could go on with my life and three minutes later i have i received 15 pdfs back perfectly formatted in a way that i can just drop and forward to my accountant and i'm done with the task you know this is something very

personal to me no one would build a product just for that because it's too too specific but we don't need to anymore we have now machines that can do basically everything and that's just mind blowing and another thing that i think is you know is is something that we're just getting adjusted to is how we talk to computers something is fundamentally changing in how we interact with computers because you know for example just the experience of installing software until now it was either you know go to a browser download a file and then click on it go through that set up blah blah blah and then you have a software or if you're a linux person you just you know you do it over the terminal you install it and then it's there and then maybe you need to configure it and

then you can use it and you know lately the way i've been installing most of the software over for an ai agent is basically just taking the link to the github or to the website of the thing that i want it to use i send it the link and say install this and i just turn around and do my thing it comes back after two minutes says it's done it's working i've tested it what do you want me to do with it and you know the i it's just it's mind-blowing that these basic these basic interactions that hadn't changed for over you know for decades since the invention of the personal computer everything has now changed now every single person can now use basically everything to a language-based interface.

And I think that's just amazing. It opens so many new doors and helps so many new people to use software and be more productive humans. This trip at TFTC was brought to you by our good friends at Unchained. If you want to get drunk on the beach and drink beers, the peace of mind that your Bitcoin's safe

in multi-sig cold storage, sign up for a multi-sig vault with Unchained. Go to unchained.com slash TFTC. Use the code TFTC10 for 10% off your first purchase. So freaks, if you've been listening for a while, you've probably heard us all talk about BitKey. It's a Bitcoin wallet built for people who want self-custody to actually fit into their lives.

BitKey is a private multi-sig wallet that removes the biggest point of failure in traditional self-custody, the seed phrase. No ceremony, nothing to hide, and no single mistake that can put your Bitcoin at risk. We all know things get lost. I lose things all the time. Phones get replaced.

Life happens. BitKey specifically is designed so your Bitcoin stays secure and recoverable without demanding constant attention or expertise because it's built for the long-term. Inheritance is built in so your Bitcoin can move securely into the next generation.

It's Bitcoin self-custody built for real life. And for February only, for you freaks, okay? February only. You got 28 days, not a leap year. I'm pretty sure, 28 days. You listeners, TFTC listeners, can get BitKey for $99 using code TFTC99 at bitkey.world.

That's B-I-T-K-E-Y.world. Use the code TFTC99 and you're going to get a bit key for 99 yeah it's i mean it's become clear to me over the last six weeks the form factor with how we interact with computers is changing right before i it's like voice to text voice to voice even um is how i mean a month and a half ago i would use voice to text for some things but really not much and now whenever i'm on my computer i'm hitting option space bar and just talking for five minutes and then sending sending that prompt or that conversation to to my agent and then just doing these tasks and

it's like we're projectionists now we can just think of something we want in your example of the uber receipts for your account like i need that and you can just project it into reality which is insane yes and and you know it goes even there's even it goes even deeper i'll tell you just a little story about how I started working on BitChat. Because when I saw Jack release BitChat for iOS initially, I was like, oh, no, this is iOS. I'm an Android guy. I would like this to exist for Android. Why did he only build it for iOS? And we're in times where you can just do it. But the thing is, I'm not an expert on Android. I've never made a real Android app before. And I've been searching for experts in several fields for so long.

For example, we're working a lot with NFC these days for Cashew. And try to find an Android expert that also has experience with NFC. You're looking for someone very specific. And even if you have the money to pay them, it's super hard to find them. And even, you know, if you say like, okay, I'm going to learn it. So you go out and search for documentation and, but, and every, all you see is codes from eight years ago that hasn't had been updated or like super old documentation or PDFs that look like technical sheets that you really don't like to read, but, or you

can't understand or something like that. it's like even for someone who is you know a good amount of skill in a certain field there are blocks blockers that would have cost you either a lot of money or time to find someone who can help you and now i can just like talk to the machine and tell like can you please explain to me how nfc on android works and it will figure everything out it it's the best expert that you can think of in several very specific domains in the thing that you are good at already. So it not only helps the uninitiated to raise the levels

and become more productive humans that can focus on their lives, but also the ones that are really deep into their own profession, helps them to broaden their scope and figure out things that they weren't able to figure out before. Oh, it's truly mind blowing. And I had a discussion with Justin Moon

a couple of weeks ago about what this means for freedom tech. And since you brought up Bitshat, I think Bitshat's the perfect example of something that's been accelerated via these A.I. tools and is a freedom tech project that found, I think it's safe to say, like mass adoption during the elections in Uganda, when they turned the Internet off in Iran, And when they had the hurricane in Jamaica, people have been turning to BitChat, which you've been working on. But I think talking about a broader topic, like how important are these AI tools for the proliferation of freedom tech generally?

And what does it mean for how we build these freedom tech apps moving forward? Does it give you more hope that we can actually win this battle? Yeah, 100%. And interestingly, it's a human factor. The answer is the human factor at the end of the day, because with AI, what effectively happens is that you increase the number of people who can reach a certain goal.

And previously, it was confined to a very small number of people. You need like the Android, NFC, Bluetooth expert who has worked on some project five years ago, and that's why he is the perfect person to build BitShed today, for example, right? So the chance of this happening is just way lower. But today, we're 100% at a point where you can build a solid and well, like useful application with a single person who's motivated enough to actually pull it off, right? So Jack has proven that.

He's basically said, I wanted to research how Mesh Networks works, and he's built this incredible app. Or another example is OpenClaw itself. It was built by a single person. It's unheard of. It's impossible. There's no way in the world one year ago that this would have been possible. And today, like only a year or maybe two years into this technology, we've already seen profound

effect profound impact of of highly motivated individuals who are now able to execute i still think you know it requires some upfront work from you as a human because it's not like everything is has now the same uh same difficulty for everyone it definitely equalizes everything but still being an expert in a certain field knowing what the pain points are having all that context that you only gain through experience that is what makes you kind of stand out in the sea of um you know equal capabilities and i think that like that's the move forward is is figuring out who's motivated and less like who's skilled it's more about you as a person what are your goals and your ideals and if you if you're a person who's like that's a doer who doesn't sit there

and wait until someone else does it but you're motivated enough to do it previously the biggest obstacle was skill and today it's not anymore and that's just that's amazing we we can't even start to comprehend what it can change i believe that software is one of the most important driving factors in this in today's world are we as humanity progress we see this in mass adopted software and social media and so like all these things have a massive impact on the world today so you know one plus one equals two is give the give them the tools and they will build well it goes both ways too right there's a it's a double-edged sword because it can be used for good and bad and so how do you how do you view that the that particular nature of this beast that

we're working with because it seems like we're in an arms race to avoid the digital panopticon ensure that we can leverage free and open source software tools that preserve privacy sovereignty and the ability to operate outside the walled gardens i mean it's i'm almost getting goosebumps thinking back to our previous conversation in the journal now because we're kind of closing the loop and it had a similar um we talked about similar topics uh a year and a half ago but things were so different back then that it felt like you know and here's what i want to say actually about your question is the good usually comes from the many whereas the bad comes from the few but the bad from the few usually has all the resources that they need to you know build a centralized social

media platform that conquers the entire world so although everyone has improved their skills i think it's like vastly disproportional in how much it has helped the many it we are way more people who want to see good in the world than those who want to see bad in the world and we as you know the mass of people have increased our ability disproportionately more than the large mega corporations who can also now use vibe coding all right okay you can use vibe coding now too but they were able to do anything that they want anyway because of you know a large amount of resources but this technological shift has benefited especially those with smaller resources or smaller teams and individuals.

And so my hope, I'm very hopeful because I hope that I know that there are more good people than bad people in the world. And those good people now have way more tools to express themselves, to actually change the world and to show us their vision of the world and inspire others. So I'm way more optimistic in that regard.

than one and a half years ago, because we have now the way more powerful weapons that we can use to fight against surveillance or repression and the digital panopticon, as you call it. Sup, freaks? This rip was brought to you by our good friends at Salt of the Earth. It's so delicious. This is my pink lemonade electrolyte mix. It's pink Himalayan salt.

there's no sugar it's way better than lmnt liquid iv i went through a a big electrolyte kick testing them all out and i landed on this settled on this been drinking it pretty consistently every day for two years now it is my favorite this is the pink lemonade they have orange here as you can see this one's ripped i had this on the train ride back from new york last night i'm always stocked up with these things they also have creatine packets very convenient particularly if you're traveling and you like to work out. You like to get those brain juices flowing. They have these creatine packets as well.

They're about to launch a bunch more flavors, lemon-lime, watermelon, and a few others as well. So go check it out. Go to drinksote, D-R-I-N-K-S-O-T-E.com. Use the code TFTC for 15% off. Seriously, game changer in our house. My wife drinks it. I drink it.

Our boys drink it. Stay hydrated. It's a beautiful thing. What's up, freaks? This rep is brought to you by good friends at silent silent creates everyday faraday gear that protects your hardware We're in bitcoin. We have a lot of hardware that we need to secure your wallet amid signals that can leave you vulnerable You want to pick up silence gear put your hardware in that I have a tap signer right here I got the silent card holder replaced my wallet. I was using ridge wallet because it secured against rfid signal jacking

Silent the card holder does the same thing. It's much sleeker fits in my pocket much easier I also have the faraday phone sleeve which you can put a hardware wallet in we're actually using it for our keys at the house too there's been a lot of robberies they have essential faraday slings faraday backpacks it's a bitcoin company they're running on a bitcoin standard they have a bitcoin treasury they accept bitcoin via strike so go to slnt.com slash tftc to get 15 off anything or simply just use the code tftc when shopping at slnt.com patented technology special operations approved it has free shipping as well so go check it out and to your point about just figuring out what your skill is when you're good at for example like when i view

my contribution to to this space is just the ability to observe uh and curate and so that's what i've been using my agent to do is just that the media company is observed and curate so i have it running all these um cron jobs to surface information that i think people aren't getting enough exposure to and then i can then curate that and put it in a podcast or a newsletter or share it on social media. And it's been incredibly impactful. Like the amount of sort of output that we've been able to push out

over the last couple of months has increased, I think, at least five, maybe 10x, just because we have these tools available. And so now when I'm thinking of TFTC, it's like, okay, I want a CBS, NBC, CNN, Fox News, but for freedom tech and, uh, sort of, uh, economic, um, economic storylines that people aren't getting in the mainstream. And I think it's, it's, um, I think it's successful. We're like two weeks into this five day a week newsletter, um, that we've been putting out

with the help of our agent. And I've gotten a ton of responses like, Hey, I love the mix of content. I haven't heard these stories. And so that's been a signal to me, like, okay, these tools are working for my particular niche in the media world um and i think on that note it took me like a few weeks to really like download a cloud bot started working with it it took me like a week or two of understanding how its memory works and basically what it could do pushing it to its limits exploring the territory which it operates in where um after a week and a half two weeks i like okay here's how i can use that this um but it took experimentation um the point i bring this up is to ask you like in terms of like practical first steps for anybody listening to this who

isn't experimenting with these technologies like how do you think they should approach using this stuff um i i want to start with especially like the security aspect first so um you should be aware that using agents on your computer uh or with your personal accounts can go terribly wrong so that is something you know that's the first thing to think about we're not at a place yet where you know some highly polished experience exists made by you know a large company or something that is uh fully safe to use and to be honest i also don't think that that will happen anytime you know very soon at least not very soon because his agents are just too kind of powerful to put them in a box yet so um so first what you do is you set it up you can set it up on your own

machine uh sorry on a separate machine let's say a mac mini is the is the trend or you can also use a service that runs it for you and one of them is chloe ai c l a w i dot ai and there the process is basically super simple to set up an agent you make a subscription you pay for it monthly and there is like ai credits involved and included and off you go and then it starts and the next step is to connect it to one of your messengers that you want to use in order to interact with it so then you're set up and you have your own personal agent that you can ask questions about and like what i what i do what i do is think about things that i wish i had someone like a like my own personal human agent basically do things for me so one of them could be accounting bookkeeping for example that's

one thing but uh we in casual we run a large organization as i said we need to basically have something that can observe progress that is also something that many people use is having a bot that just observes things and writes reports on it and in your case it can be the news for example and in my case is pull requests and issues on github for example and um and lately what we've been experimenting especially with chloe uh in order to make our operation more efficient is to pick off kind of jobs that we would have hired someone for, where we can carve out some specific aspect

where we think this is safe and nothing can go wrong there if you automate this part, for example, observing something and then build like a dedicated bot just with those skills. And in the case of Chloe, we run our infrastructure

on a system called Kubernetes. it's it's a fairly common advanced provisioning system where you know the cloud basically is kubernetes and um you have like experts for for that obviously there's like people who know kubernetes in and out and you can pay them to do a certain job for you but you also now we have these agents that know kubernetes really really well as well so and if you plug in a relatively good LLM into it, say you use Sonnet 4.6 or even Opus or something, you kind of get to a better than junior

person in that specific job that just works for you 24-7. And that's what we've been doing. So we have these DevOps agent that lives in a signal chat with us. And I can ask the bot in you real human language hey how's everything looking you know are there any problems with our cluster and then it'll say yeah there are like two two people whose agents have crashed and they haven't come back for like five hours maybe you should look at them and just gives me further hints to any problems that might have come uh while i was doing something else so like this is

for our operation it's extremely useful already today and you know i'm just thinking about let's you run a factory, right? So you have a factory with all sorts of sensors and monitoring and reporting and all of that needs to be observed. Like a human needs to sit there and look at that data basically to make use of it. And what we've been doing so far before AI is like dashboards. That's the way we typically try to condense information so it's useful for us as humans. But today, I mean, you can still do the dashboard, it's no problem, but you can also just plug an agent into that system that can read the raw data and it will probably come up with like even better ways to look at the data and report anything that is out of the ordinary for you so you know just

this one little thing again we're just talking about one thing that it can do that it was not particularly designed to do but it does exceptionally well and just this one thing is by itself already huge imagine what it means for the industry well the industry producing physical goods in real factories what it means to have agents plugged in into their system that can observe everything 24 7 all the time and that costs basically no money i mean you pay for the other them but the difference between a human doing that job and the bot doing that job is basically no money so um so if i can think of like two or three ways how this can be extremely useful for me i know there must be millions of ways for millions of billions of people how they can make

it useful for them as well and the nice thing about these agents is we don't need to know before what they're going to do because they're general purpose machines they can probably do way more than we think that they will and they will continue improving in in the coming months and years and we'll be just like mind blown for the next couple years i'm pretty sure about that yeah that's the crazy thing to to realize is how this thing is pretty good right now but it's probably the worst it's going to be 100 it's going to get faster and better at doing what it's doing and you know just from a software engineering perspective if you go back to level two right the agent that writes code for you in your computer right now the flow is you enter a

prompt and then you wait like a couple minutes and then you observe what the agent has produced that loop will become like 100 milliseconds at some point you will you will just enter it and press enter and it will be there instantly for you to observe and if you don't like it you can just redo everything instantly as well so you know just the flow of the talking to the thing and getting something back all that is going to compress a lot in the next coming years and even like just the speed aspect just the performance aspect itself will open unlock new ways of working with it again yeah i guess um diving into particulars of of building um products with it whether it's the cashew protocol bit chat um clawy.ai using the

to build that service um now numope uh so it seems like you've been shipping at an increased clip just observing publicly and i'm assuming that you're using all these tools to to do that i think it's a safe assumption um i guess let's just talk about some of the things particularly at the intersection of of bitcoin um in ecash with the cashew protocol what you've been able to to ship that's been on your mind for for many years i'm sure yeah so um i just hinted towards NFC. NFC is something that I've been researching for years at this point. And I'm very jealous about the fiat world in that sense, that tap-to-pay

is just an incredible user experience. And everyone loves it. People use it all the time, everywhere, globally. I remember a conversation I had with Miles Sutter from Cash App, where we were talking about Bitcoin adoption around the world and how we can improve you asked for bitcoin payments and he told me this story that doesn't leave my mind anymore which is he's somewhere in peru in some like distant uh isolated mountain somewhere and he finds this little shack that sells him a coke and in this like mountain somewhere in the nowhere the person hands him over a pos device and she takes his iphone and she taps it and it makes

beep and it takes like three seconds and the payment is done he said like that made that thing made him more bearish than anything else with regard to bitcoin and that just resonated completely with me because i like that is what makes me bearish about bitcoin as well is the user experience because if fiat feels so good like why do you even need bitcoin that's many what many people think how many people think so obviously we need to get better in bitcoin and offer at least the same experience the quality of experience or better and i think we can do better because we're an open source system and we can you know where improvements of one thing help also the other thing we don't need to compete we can also collaborate and that's what we're doing with

open protocols and and that's the whole idea about bitcoin basically is just we all work on the same thing so um nfc i think is a vital part tap to pay just works everyone loves it and um in order to build that i needed to get all this expert knowledge which i wasn't able to obtain over the years and i said like i was actively looking for people who can help me but found no one and so forth until um at some point i think this started all more than a year ago again with a little project i called nfc chat and that is kind of a if i think back it's kind of like a precursor of bit chat uh because it's like bit chat but with way less radius and distance it's just a simple chat app that you run on your phone and you write a message and you have to tap the other phone in

order for the message to go from one phone to the other so it's like a very stupid simple chat app where you need to touch the phones in order to send the message and i built that as a step towards kind of nfc payments because with cashew and that's the charming e-cash protocol that i'm working on money can be sent as a bearer asset bearer token so that means like if you can package it in some kind of a data package and you can send it over and that could be you know email or telegram or sms or whatever it is any communication channel you can make a payment so i wanted to see bitcoin payments happening with tap to pay using cashew by sending the e-cash token from one phone to another and that was the first proof of concept i was able to only achieve because of ai so i thought like

that's neat awesome that's great so this is exactly like the these are the building blocks that i was looking for for so long we have cashew as a protocol that we can build on that runs on on Bitcoin and with this incredible property that if you can send data, you can also send money, right? So that's what Cashew enables us. And at the same time, I just should clarify, you know, if we deal with money, for example, you know, the core libraries of Cashew, they're all handwritten because we're not at a point yet where I can safely just say that the agent does a better job than I do in, you know, figuring out what could go wrong or how you could lose money and so on.

So that's very important to us is that the critical parts are handwritten, normal programmed libraries. And that's what usually a cache library is. But once you have that, once it's running, once it works, then you can build all the scaffolding around it, vibe coding, like there's nothing that can go wrong, basically, because the library itself works. You know that you won't lose e-cache.

So now let's build an app around it. So we started with this NFC chat, sending around eCash tokens and then showing like, okay, this is possible. And then this ultimately culminated in something, in a project we call Numo. We've been working on this for a couple of months at this point in stealth mode. We haven't announced it to anyone until like yesterday. And Numo is our approach to introduce tap-to-pay to Bitcoin using eCash. Because with eCash, it has this very simple flow of data flowing from one phone, the payer, to the receiver, the POS device. So Numo runs on a normal Android POS. I have one here. I can show it, for example.

This is a standard POS that runs Android. but it can also be any Android device, any Android phone. So it needs to be an Android phone because iOS doesn't allow you to use the NFC chip in that way. So the POS device itself off the merchant runs Numo

and then any other phone, and now this can be iOS or Android, doesn't matter, can basically just tap to pay to it. And if it runs a cash audit, then the cash audit will see that there's an NFC chip and then take the token

and send it over to the PVETs. And why is this cool? Because first of all, it's using eCash generally. Cash use privacy inherits this incredible privacy by default. So this is, while the UX is amazing, you also have the best privacy possible,

which is a great combination. Usually you have a trade-off there. And the next best thing is that it works offline. So as a payer, so the device paying at the POS, could be your iPhone, your Android, doesn't need internet to make that actual payment because all you're doing is you take e-cash from your wallet and you send it over NFC without touching the internet. So that also gives you incredible privacy, obviously, because you're not even online, but it also means that it just works everywhere. And that's, again, one point that fiat payments already have been doing for years. If you do tap-to-pay with your credit card or if you do tap-to-pay with your phone, it also works while you're offline because your phone doesn't talk to the Internet.

It kind of hands something like a token from the phone to the POS device in fiat land. And so we're doing now the same in Bitcoin land, but with incredible privacy and open protocol. And so the way it works today is you can install Numo on your Android device or on your POS device. It can be any generic POS device. And you can set up a cashew mint. So cashew is a custodial system. You need to trust the mint operator with the Bitcoin that you receive. So you set up one or more mints.

And then any payer from any mint, it can be like, even if we don't share that mint, you can still pay with the e-cash to the device. And then the device will take that e-cash and withdraw with lightning to their mint. So anything, all the mints are still interoperable because we have lightning that glues everything together. and you can receive payments from basically any Mint out there. And we've built a cool feature into Numu, obviously realizing, recognizing that it's custodial.

There's a rug risk associated with the Mint that you're using. So you should only be using Mints that you trust already. But in Numu, there is a very simple way to say like if we ever reach 50,000 Satoshis, for example, you can enter a Lightning address there and it will just automatically withdraw

to that lightning address whenever you reach the threshold so that way you can build this pos in a way that the you know the maximum amount that you put in custodian risk can be very limited and so you can also withdraw every single payment if you like but you know for very small payments it might not make uh sense um so we've we've just released this yesterday you can find it on numopay.org and download it check it out and we'll be working on a btc pay integration uh soon we've already started working with that so it also works with existing btc pay server infrastructure but if you don't want that you can just download the numu app itself it doesn't need any back end there's nothing that you need to install or set up just install the app choose a mint and you're done

so took it out yeah i mean i completely agree i remember miles telling me that same story too i think the offline nature of the the ability to send cashew which should be e-cash via the cashew protocol offline is massive you think miles in that mountain in peru like his cell phone probably didn't have service but the secure enclave had his credentials for his card and he was just able to send it to the pos and make the payment probably offline i'm assuming And now we have feature parity via eCash now. And I guess what do you want to see in terms of like what or what do you what do we need to see for this to become more prominent as a wallet sort of integration of the NFC standards is a proliferation of just simply more eCash wallets while it's work with it now like macadamia. I'm pretty sure I could use Macadamia on iOS.

And just to clarify, for the iOS stuff, you can pay with iOS, but running the POS to receive the NFC is something that you can't do. That's right, exactly. To receive, you need an Android phone, but paying, you can do it with any phone. So yeah, as I said, Macadamia, but they're already, we've worked in the background

and we've added this NFC capability to around like six or seven different cashew wallets at this point. Most of the popular cash wallets already support this from day one. And Numo also supports bare Lightning, so you don't need to have a cash wallet in order to use Numo. Obviously, that would limit the number of people who can pay with that terminal.

So it supports Lightning just out of the box as well. And you can use any Lightning wallet to scan the QR code. And some Lightning wallets, particularly Phoenix, can also use TAP to pay for Lightning. So it will kind of fetch the Lightning invoice via NFC and then pay that over the internet. So it is compatible with all the wallet infrastructure out there. But generally, the thing that limits growth in Bitcoin today

is not the utility, but I want to say the penetration of Bitcoin itself. So I'm very well aware of the fact that our technology can only grow as fast as Bitcoin can grow. And when I say grow Bitcoin, I mean people using Bitcoin, not the price going up. So our ultimate goal with Numo is to further Bitcoin as money. That's our main North Star goal for everything that we do is to support the adoption of Bitcoin for everyday payments around the world. So the infrastructure for that is being built out, and there are great projects out there. For example, we've noted ETC Pay Server.

So anyone could already use some form of on-chain payment or Lightning payment with several existing solutions out there. But I think what's most needed today is kind of a penetration in real economic transactions. And we need more merchants to use Bitcoin. And I think that Cash App is doing an amazing job there. I should say Square, to be more precise, is Square is doing an amazing job there.

There's BTC Map that you can use to find merchants. There's obviously BTC Pay Server, which is the most popular kind of web checkout system that the Bitcoin ecosystem has produced. And so all these projects help towards the same goal. And we now support that effort as well using Numo and our NFC-capable payment systems that we built. If you want to support this in your Bitcoin wallet, But what you can do is, first of all, look at all the NFC specs that exist in Bitcoin. You can read liking invoices via NFC.

If your app doesn't support that yet, think about integrating that. If you want truly the tap-to-pay offline instant experience that Cashew can give, then you can also just plug in a Cashew library into your liking wallet and pay with eCash for that single step. it doesn't even need to be a full-fledged casual wallet you can just like that single one step that you want from bitcoin to e-cash you can also do that in order to support tap to pay so there are a couple of wallets working on that and generally speaking i think as a community as a bigger project than bitcoin we need to um help get bitcoin more onto the radar of merchants and and hope and drive the adoption of real brick-and-mortar stores so that's that's what

that's what we want to contribute to i completely agree and thank you for your focus on this it's funny too because uh when we last recorded i think i asked what your archetypal cashew user was and you said uh a woman in the jungle selling selling mangoes and i think uh i think with pneumo pay and projects like a btc pay server i think that's going to proliferate obviously what square is doing here in the united states for their emergence i mean merchant adoption because the squares roll out at the end of last year went literally parabolic i think there was something like uh 2500 merchants on btc map there's probably more than them um in the united states that that haven't put their um put their store on btcmap.org but um

i think the way square rolled it out and cash app in collaboration with them rolled it out they automatically added all these merchants and went from like 2,500 to over 6,000 over the course of a month which is great to see um shifting gears back to this intersection of ai in bitcoin and particularly Casio, that's a big topic of discussion right now, particularly in our world and in the fintech world as well, is agents using payments on the network, excuse me, using payments with each other to complete tasks on behalf of the humans sending them out to do things. And I think it's a wide open landscape right now. I think you have Visa, MasterCard working on their own protocols. you have Stripe, then obviously you have Stablecoins, Bitcoin over Lightning.

I saw that the guys at ARK Labs launched the ability to receive Stablecoins and automatically turn them into VTXOs via their ARK protocol implementation. And then obviously we have eCash and CashU specifically with the NUT402 protocol. And I think you've made the case that due to the sort of error nature of eCash and the fact that you can send the file, that it's actually the most efficient way for agents to make payments. Is that correct characterization?

Yeah, so that was just like from our experience what instantly worked. It was extremely simple. It was almost shockingly simple to teach an agent how to use Cashew because we basically told him, you know, here's your CLI and the CLI, it will produce Cashew tokens

and you take the token and do something with it depending on what you're doing. and instantly got that. And because you don't have this get an address here, check your balance there, sign and send, you have these packets of data that the agent can simply send in an HTTP request, for example. So what today, getting back to the original question

or the idea of 402 payments or ATIN payments, I think it's a very hot topic today. we're still yet to see how much of that sticks because we're in this like discovery mode of agents what should they do how much money would you give them how dependable are they with the money that you give them and so on there are stories out there of an agent like sending a random stranger 350k or something because he he backed him online and the agent made like a single mistake and the person who owns that agent lost a substantial amount of money due to that because they were kind of trying it all, I guess.

So how much of that really manifests, we will see. But it is clear that the big ones, and you mentioned them, the big fiat one, the big fiat monsters like MasterCard and Visa and also Stripe are looking into it. So what frustrated me a lot looking at this is that the 402 idea, the idea that we will have services online

that ask for money. so you want kind of your machine to be able to pay another machine is a very bitcoin driven topic has been a bitcoin driven topic in the last years you know for the last whatever 10 years or so the only people who talked about this were bitcoiners because we we always thought of ourselves and still think of ourselves as the native currency of the internet so So, obviously, the connection there is pretty direct. And there's been development from Lightning Labs on L402. We've been contributing to 402 with X402 and cash payments.

The Albi guys have been working on this as well. So, it was kind of only this niche topic until the agent came a couple months ago. And then you saw Coinbase working on it, Stripe working on it, the credit card providers working on it because they all seem to agree that payments of machines to machines will be a thing and the rails that they

market or want to see being used are usually stablecoins and stablecoins have the benefit of feeling more natural to people I think

the cognitive leap to use a stable coin is way smaller than the cognitive leap to use bitcoin because the stable coin is denominated in the same currency that you already use so it feels like you're still just doing the same thing but now in a kind of a digital way or something so the stable coin people have been pushing really hard to make sure that the agents of the future will use USDC or USDT or whatever it is. Obviously, I don't want to see that future happening because I think, you know,

just for this fiat system and stable coins are inherently trust-based and there's all sorts of problems with that. You need KYC to get in and out and all these problems with stable coins that Bitcoin doesn't have. Bitcoin just is born on the internet and lives on the internet

and it doesn't leave the internet. which stays inside the internet. And so that's why Bitcoin is still the native currency of the internet. And several projects in the Bitcoin space have been trying to also rival this new trend and make sure at least that we're up there with the others and can show that we show up

and that we are ready to compete with them. And one example is the ART protocol that I think is a fantastic idea to use. I've checked it out already, and it works really well. And another way of doing it for even smaller transactions and high-frequency transactions, Cashew is just amazing for it because it's basically instant, basically free, and with insane privacy. All of these factors, I think, are super important, especially in an agent-acquired world. So eCash is just something that works really, really well with agents. Yeah, and particularly in a world where the subject of ID verification on the Internet is becoming more prominent.

I think the ability to opt out and have these internet native options that don't require KYC at the edges like stable coins are imperative because it seems like the digital ID conversation is definitely beginning to pick up again. Every stable coin is permissioned. All of them can be turned off if they don't like your digital ID. Yeah. Yeah. I want to be respectful of your time because I know I pinged you in the afternoon yesterday to do this and you were gracious enough to do this last minute. But to end it, I love getting philosophical with you. Last year when we spoke, you really dove into how you view yourself as an artist in the digital world. And I think what you've done since then has proven that to be true.

but thinking about the proliferation of these ai tools and humans beginning to interact with these at an accelerating pace um what are your hopes for the future obviously the meme of the permanent underclasses here the meme of the singularity and skynet having been launched already and we'll only recognize in retrospect is beginning to permeate what what is your view for how these tools affect our humanity oh that's a that's a big one because um so first first i'll start with my craft and i hinted towards this already in the beginning of our conversation today but it's been it's been amazing to watch how things can change and in a way and in a pace that no one would have predicted that the penetration, the proliferation of AI in society

was always predicted to have the most negative influences on the old kind of manual job markets and the plumber and the robots will come and everyone will lose their jobs and so on, which I think is still a risk. But I wouldn't have predicted myself that programmers would be the ones who are affected first or even the most.

And I think probably will be the most. So we are at a time, you know, zooming out that software engineers had this incredible run in the last decade or two and make astronomical sums of money and still software engineering,

you know, the salaries associated with it kind of like still detached from the rest of society. So we have been seeing a decoupling of what you just maybe call the permanent underclass and the Silicon Valley upper class. But what I see is that

what's happening right now is changing that drastically. And I think that's maybe for the better because software engineering has never changed this fast before and has become just way more accessible to many, many more people. It's not this exclusive skill anymore

that you have to acquire yourself by decades and decades of working on your own and acquiring all these special skills that you can then use to find a job, for example. So every software engineer out there who still didn't get it, I just want to say it again, you will lose your job in five years.

You're done. If you're not using AI, like I'm, I'm sure of this and I cannot say this about like another job category just yet. So I think it's just a, let's say, let's see, it's ironic that today's, in today's world, the plumber and the carpenter and, you know, manual labor folks are, I think, more valuable than ever before. and have more security about their jobs than ever before relative to the rest of society. But that is something to think about. It's just, it's almost beautiful to see it happen this way. And at the same time, you know, getting a bit more philosophical on this is when I, when we last talked, I spoke of software engineering as something like an art form,

where if you have the energy to produce and you have the skill that you can really affect the world with your ideas, which is ultimately how I think what drives art itself is finding ways to materialize your thoughts that can be a book or a painting, but it can also be software. And although In this renaissance of software development, we're seeing that the brush really doesn't matter as much as we thought it would anymore.

So the manual part of it is just completely changing. But that means that the rest of software development is becoming more important. so now now that everyone can write software or those who can write software before can write even more software software just becomes kind of like a disposable artifact of your activity so what matters now is why why do you do what you do and that makes things a bit more complicated as a human to figure out why you do something or what you should be doing what's a good thing to do you know what makes sense to do those require a completely different set of skills to figure out good answers to these questions than just like the learning postgres and and and c-sharp and

whatever right so now you now it's about ideas and execution some people say taste and aesthetics but I also think about communication and marketing and trying to explain to people why you do certain things and why they should be interested in it so all of these like way more human things and way more soft skill things are suddenly at the center of software development And so I think that is just that's that's that's the reason why everything is going to change and it's changing right now. So we might see a new wave of of software. We might see a new wave of cyber culture emerge because because of these changes that we're experiencing right now. Yeah, we may find that all the fears of our humanity being stripped from us are actually completely wrong. And the technology forces us to explore our own humanity and bring it out, be our creative endeavors in the world of projectionist software.

There is this idea that, you know, it's popular in this universal basic income circles about what do you do once you have everything you need? What do you do if you don't have to work anymore? And it forces us as humans to really answer that question. I think that question is important, whether UBI has arrived or not, whether everything is fully automated or not. You should be asking yourself this question like 10 years ago and today you're forced to ask that question. If you're a software engineer, you're literally forced today. If you're not a software engineer, you'll be forced in like five years to answer that question, maybe in 10. And the good thing is that we're getting more prosperous as a society. Hopefully, at least we can all. I mean,

if we if this doesn't turn into a dystopia but the good news is that we only have more time in our hands more life quality in our hands we think about the more important questions like who do we want to be what is the goal of all our actions how do we want to live in a society what what should this world look like so there's room for there's room for new ideas and it's the most exciting times to be alive it really is it really is and i'm happy that you have that perspective somebody i respect in the world of software development it's great to see that you're optimistic about this uh thank you for all you've done i mean from bitchat to cashew protocol now numopay clawy.ai making this agentic world more accessible for other people

who i think should be experimenting with this it's uh it's insane to see what you've shipped and since we last recorded so keep it up brother thank you and i want to afford all of those great words that you just shared with me also to the team which i you know which did most of that work and we are a great developer team with incredible people who are working on cashew day and night and we all just want to see bitcoin uh succeed and i also want to you know use this dance to talk to the people who are listening out there thinking you know what should i do how can i help how can i help improve uh this world we live in is i you know this is a a uh bitcoin is a free and open source system that lives off the contributions of people like you and if people

don't choose to work on bitcoin bitcoin stops working it will die so i hope that you know there's at least one person out there with things like i've been a software engineer all my life and i'm fed up with this job and i want to do something else to consider you consider becoming a bitcoin developer and help with you know help us with driving these freedom technologies forward because no one else is doing it there is no one except this circle of people that we've talked about trying to counteract uh the the massive growth of surveillance technology and centralized control and uh so if you feel that you need or you can change something in your life and you you want to contribute your energy towards something good please please join us and become

a bitcoin developer become a cash developer look at bitch hat look at all the other projects that are out there and and contribute if not you then who ali it's been a pleasure sir likewise thank you marty peace love freaks okay thank you for listening to this episode of tftc if you've made it this far i imagine you got some value out of the episode if so please share it far and wide with your friends and family we're looking to get the word out there also wherever you're listening whether that's youtube apple spotify make sure you like and subscribe to the show and if you can leave a rating on the podcasting platforms that goes a long way last but not least if you want to get these episodes a day early and ad free make sure you download the fountain podcasting app

and go to fountain.fm to find that five dollars a month get you every episode a day early ad free helps the show gives you incredible value so please consider subscribing via fountain as well thank you for your time and until next time Thank you.