Welcome to changelog and friends a weekly talk show about zsdd thanks as always to our partners at fly.io the public loud built for developers who ship We love fly you might too. We're more at fly than I owe. Okay, let's talk My friends agentic postgres is here and some of friends over at tiger data This is the very first database built for agents and is built to let you build faster You know a fun side note is 80% of clod was built with AI over a year ago 25% of Google's code was AI generate and say to say that now is probably close to 100% Most people I talk to most people talk to you right now almost all their code is being generated. That's a different world Here's the deal agents are the new developers. They don't click they don't scroll they call they retrieve they parallelize They plug in your infrastructure to place you to need to perform But your database is probably something you about humans only because that's kind of where postgres act Tired is philosophy is that when your agents need to spin up sandboxes or migrations, core use volumes of vector and text data
Well, normal postgres it might show and so they fix that here's what we're at right now Agentic Postgres delivers these three big leaps native search and retrieval instant zero-copy forks and MCP server plus your CLI plus a cool free tier now if this is intriguing at all head over to tiger data Dot com install the CLI just three commands spin up an agent postgres service and let your agents work at the speed They expect not the speed of the old way the new way Agentic Postgres is built for agents is designed to elevate your developer experience and build the next big thing again Go to tiger data dot com learn more So we are supposed to be in San Francisco, but we're not in San Francisco now we're not I would have sadness, you know Well, happy. I'm not stuck in an airport or trying to get to someplace But for those who don't know our government is shut down and therefore the FAA and TSA and all the acronyms a are not showing up Hey, and so you have a bad day Show up a you know what I'm trying to say let's get to San Francisco a therefore we cancel our friends
You know we told him hey don't come record with us this week We're gonna be in San Francisco. However, we are not in San Francisco So we are here in our home studios ready to friends and we're gonna do kind of a you know a grab bag The director's cut change our news director's cut deep behind the scene. I was a extended cut. Yeah, you know extended director's cut You know take a ten minutes make it, you know, my dear condition 90 Steel book edition you get the watch those do you watch those directors cuts when you're gonna? You know deep and you know he's gonna he's whispering here's where I should on this or I was trying to capture why I would say on certain movies You know certain movies that you want to go deep, but generally no, it's just not my thing I used to do it when I had odious free time and much more of a movie fan than I am now and I liked them Although they do get pretty boring. Yeah, I mean, it's not as good as the movie, right?
Right, it's like well watch I'm looking at the movie But I'm listening to a guy talk about the movie so yeah, sometimes you get nice Extras I would say about the plot or the twist or the storyline that you don't get in the movie It's in this case I do enjoy it, but that's not always the case Okay, so after this show let us know do you like the director's cut of change like news or do prefer just a ten minute? We shall see you know folks and get out of here, man get out of here, so There's some good stuff in this week's news. I thought I looked at the list and I was like dang dude I did well, but like there's lots of good stuff out there my favorite of which I put right in the top Which is the alive internet theory as you know, I've had I don't know if you call it an obsession or I've had a frequent References to the dead internet theory. I've been experiencing the dead internet of late. Do you feel it Adam?
Do you feel the dead internet or is it just a theory to you? You know what did it is essentially like bots commenting and like bot content bots commenting bots creating bot blog posts Hey, I slop just everywhere. I think I would experience that more sadly in the TikTok world I feel like it's on repeat. I feel like it's all about product placement. Yeah, I feel it's all about like selling stuff in the TikTok shop I feel like there's creators who just literally go into the shop and get incentivized to shill things in there They may like them But I feel like it's just like the only reason making content is because you can get some version of their products into you for free Yeah, then you can schlep it and make some money. There's not a real relationship there. So that's not a bot really
Yeah, it's a person, but they're driven so much by the algorithm that they're like the undead. That's right. Yeah, that's true I would say not really not really now I do think that the AI slop is coming to the world of video and audio I think it's behind where it is in pros because of course the text models were the first ones to get good enough to trick us a little bit Yeah, but you know with Sora to and that whole deal I'm not sure if you were I didn't join the social network for Sora But I saw all the cross posts elsewhere of what people are making with it and that one's actually okay I'm not a huge fan of open AI as an org But that was at least an attempt to be like transparent about like this is a social network for AI slop And that's what it is. And so you want to get sloppy. Let's go get sloppy together But let's not act as if we are real Let's just go ahead and brace the fact and there's some I mean there's funny interesting Cool stuff that you can do and so I'm not against the content necessarily Especially because you know what you're getting you're like Oh, here it is But when they're on the youtubes and they're on the tiktoks and the instagrams trying to act as if they're humans
That's where I get a little bit offended. So I definitely have been feeling it up late But I was uplifted by Spencer Chang. I did reach out to Spencer As in the come on the show has I haven't heard back yet It's just yesterday So hopefully come on and talk to us because he's doing all kinds of cool stuff and he created the alive internet theory of course a An answer a counter narrative and a very cool website to the dead internet theory Mm-hmm, which basically states his premise which I did quote in news That you know the internet will always be filled with real people looking for each other answering calls for help and sharing laughs Even the midst of the arguing that was lots of arguing for sure Mm-hmm and the cool thing about this website is basically has went down through the ages like it scrolls the timeline from 2004 and updates kind of the style of websites Based on where you are in that time and all the way up through present and you can click a time period and then start the experience And it's gonna now load up with audio which probably doesn't translate, but it's gonna load up. Oh, it's too loud actually
I'm blasting myself It loads audio very loud like a digital scrapbook of Randomness yeah that humans you know in all of our analog jagged weird ways have put onto the internet during that time frame And they're all pulled from internet archive of course the way back machine And just super rad and maybe happy I was like, you know what kind of reminded me of the nostalgic feeling We had going back to the Winamp skins. Yeah, that show we did last year looking so much fun Yeah, when it was the best so thanks Spencer for Restoring a little bit of my hope in the internet and the fact that yes over 50% of new articles according to a recent study Right at 50% of all blog posts and articles are now written by AI But and that's always a good increase from here The humans are there man. We're there and we can find each other and we can talk to each other and we can care Yeah
I don't know what the how to make of this because written by AI is is somewhat potentially a misnomer in the fact that sure written by AI But thought up and desire by a human, you know orchestrated directed by and I'm really sure I sit with that still Because it's it's really a conundrum that that I use a lot of AI to do things and it writes a lot of documentation for like little things I'm doing and if it wasn't for that cheese, I wouldn't even show up and do it It's just too much time. I don't have the time for that And so I'm kind of I'm a little torn with that So where do you sit with that in the fact that just the blanket statement of written by AI? But maybe not the sub context of directed by and influenced by and ideated by human the only reason is there is because the human was like Hey, this is cool. Yeah, I write it. Boom. There it is
I am totally okay with what I could just call perhaps utilitarian uses of AI Generated words such as documentation such as explanation and Especially when directed by humans who desire that content in order to better understand something. I'm also okay with it in the context of information such as Wikipedia and now the the the competing grokipedia which seems to be so far not as good as Wikipedia But I don't disagree with the premise of grokipedia Which is like if it lets let's remove as much as we can human bias now We're also the models are biased So that's why I say as much as we can but let's let Information be written by an unbiased as much as we can source Because we want just the information sometimes and that's fine And so I have no problem with it in the context of those things I don't know I don't even necessarily call that slop unless it's actually sloppy similar to the way that I Instructed a coding agent to write code for me and I direct it to write code as long as I'm As involved in that process and I approve of the code it
provides a use and a value Where it really feels sloppy to me and bad is when it's creative to entertain to persuade These things and it's or to SEO. I'm gonna try the other reasons why they do that when it's just content farms Yeah power guy I see a lot on the on YouTube where you'll search for something in particular about a product I do some product exploration, you know like an appliance or something that you know We're buying a an example of if we're buying a new air fryer. Let's say for Christmas And if you're doing this, you know, whatever Let's not you go on YouTube, you know, you find maybe the ninja version of it and you got this other version of it You know, maybe from cuisinart or whomever and you start Versus seeing them, you know, queasant versus ninja Well, you're gonna find some good people out there and I have no idea how they made this they're living to like compare air fryers Mm-hmm, and then you will eventually find well probably pretty quickly find some AI generated or certainly not human It's it's doesn't even sound very human when it's speaking to it something a computer voice
And it's clearly just an infomercial for the views Mm-hmm, and it ranks high and maybe it does get some views and though some of those views perpetuate more views in the algorithm I don't know but those things are gonna be crazy or just straight up like you can tell No human was a brightness or there was and it was just Just for the views. Yeah, not for the engagement or the true enjoyment And I'm honestly only really offended when I can tell right because it's kind of like that comparison like a toupee or A boob job like that's the two things. It's like gosh. You only know you then a person has one when it's a bad one, you know Oh, yeah, it was good Then you can't tell anyways, like you don't know that guy has to pay it's just you think it's hair It's when it's bad that you're like dude, we're gonna toupee kind of offensive silly sometimes you never know But you don't know you're fine with it when you do know now. It's bad That's the way I feel about I guess these things where it's like if you fooled me then like good on you I guess like it was good enough that I thought a person did it But when I find out it's acting like it's a person and then I realized that it's not a person
That's when I get mad. How about this maybe a civic a Honda Civic wrapped in a Ferrari shell. Okay, that's that what upset me Without the engine you open the hood on that sucker. Yeah, like what is this for cylinder DH whatever thing, you know I don't even know what the engine darn civics anymore But it was something like that overhead cams, you know, it's like oh Hc Or the real tall person and then you know in the trench coat and then you realize it's like three short people saying on shoulders Yeah, that's going on a lot But you know when they went to Trixia you don't you're not in the wiser So, you know, I think once we get beyond the young gany Valley where we can no longer detect and I'm not sure if and when that takes place Then perhaps the dead internet theory will be just an undead thing They don't they don't even know to pull it to pull an exact pull quote from I guess them the person who wrote this What was the thing again Spencer Spencer? Thank you Spencer and then I guess by by proxy you quoted his quotes I'm quoting you quoting him. Okay, this is Michael Scott moment. That's right. We're going deeper layers
You said the internet will always be filled with real people looking for each other answering calls for help and sharing laughs even in the midst of arguing I think that's when the internet will always be what we consider since the 90s or Whenever's when you and I began to do dial up or it was cool to email a friend kind of thing, you know That's when the internet is still the internet because I want to show up to a place that is globally connected and informational Shareable forever so long as it's sure with other human beings that have empathy have care have grace, you know mercy What all those things that the things you know embody that's When that's true while that's true. I will enjoy the internet when that becomes less true or not as true I'll probably you know going to walk about or move to the cabin the woods that I got reserved, you know Well friends, I'm here the good friend of mine again Kyle Galbraith co-founder and CEO of depot.dove Cal
We are in an era of destruction, right? I would also describe it as rethinking Well, we thought was true and I guess that's kind of the definition of destruction But from your perspective how are teams reliability teams CI CD pipeline teams? How are they all rethinking things and where does depot fit into that in the conversations that I have with customers a lot of DevOps teams platform teams site reliability teams They're really looking at this new era of software engineering that we're all living in and they're starting to question like the bottleneck is no Longer the active writing code the bottleneck is shifting the most time consuming part is integrating the code. It's everything that comes after It's the build. It's the pull request review. It's the deployment. It's the getting it into production once it's in productions
It's scaling up support teams to support it. It's adding documentation all of these downstream problems And so through the lens of depot what we're really starting to think about is there's a very realistic possibility that within the next two to three years Maybe even sooner that we're going to enter a world where an engineering team of three people could theoretically have the velocity of an engineering team of 300 people and what's the consequences of that what's the consequences of the code velocity Spiking up to that level with such a small team There's no way three engineers are going to be able to code review all of the code that's being created if there's three engineers and 297 agents also grading features and fixing bugs So that's just like a full request perspective But then you think about it through a build lens to of if your bills take 20 minutes with three humans and now you're gonna have three humans and 297 agents also running well Like you definitely don't want you to build a 20 minutes because now like the entire pinch point is the build pipeline And so we're starting to think a lot about how do we eliminate the bottlenecks that come downstream and what can we do with depot that stream So obviously friends we are in an era of disruption things are changing
You know it I know it that's how it is and the thing with production of what cows talking about here is how in the world You get your bills be faster. I get them to be more reliable faster more observability around those deployments You need it. It's required and depo is there to help you So a good first step is to go to depot.dev get faster try the trial is too easy again depot.dev is where to go It all begins at depot.dev Well, let me share this with you then which is a bit of a curveball because we didn't discuss discussing it But it is in news it was in the the lower three which get the most letter and they get mentioned so they don't get actually coverage Which is much tastic, okay, and it's in the same vein now of like what if there wasn't necessarily or there still was this global connected internet That maybe only robots use to talk to each other Maybe at our behest maybe not and there's no people there I mean because honestly if it continues to ramp up and the uncanny valley stays meaning it's just a I slop and I know it I'm not ignorant of it. I'm like I might check out. You know might I might be like yeah
No longer surfing the web to do other stuff. Okay, so that might happen and there's some social networks that they say's already happen Like a lot of people claim that X is now mostly robots, which if it's true and I don't know That's hilarious because it's one of the reasons they must bought it was to get rid of the robots anyways mesh tastic Yeah, this is cool in open source off grid decentralized mesh network. That's built around affordable low power devices So these this uses inexpensive low-run radios, which I don't know what Laura means. Do you know what Laura means? I'm just for the first time I think so hello RA So let's find out in real time long range. So lower. I just means long range. I already knew it was long range
Oh, yeah, I'm excited to know that it was Laura means long range I'm the low range off grades. Oh, that stands for no Laura just stands. It's from long range a physical proprietary radio Oh, it's an actual it's a proper noun. Okay, so that is different a proprietary radio communication technique based on spread spectrum modulation So it's not just an acronym or a shortening. It's actually like a technology that these things use what I'm seeing says Uh, Laura. Oh, RA is a software technique that adapts large language model by training only a small set of new weights
Is that what you're you're tracking? I think these are probably names based conflicts because this is all about communications Oh, it does say two different technologies here. I am way off. Okay wireless communication protocol or the thing I just injected to the right Definitely on topic for most of our coverage. I bad y'all. I'm just glad this isn't that
So this is long range radios and this mesh tastic software I think you flash it onto specific devices. They have a list of devices that you put it on And this that's like basically a mesh network that is not just at your house I mean the most much networks that I think of is like Bluetooth, you know, airdrop airplay kind of things But this is like you connect all of these Expensive radios and it can go long range and they're not getting they say long range 331 kilometers long. That's the record at least Wow using mesh tastic This is meant to be by use by humans to send messages. Yeah, you know, it's like the old Carrier pigeons man, you know, you have a message you're gonna send over mesh tastic So this would be like if you and I could probably do this with like Nick Nissi for instance He lives across town from me right probably 30 miles which is gonna fit inside this 331 kilometers And he could get one of these radios set up at his house a runny mesh tastic and I could get one at my house
In fact, we should do this be fun. Yeah, I was a test as an experiment And he would get one sent up at his house and my head went my house And then we connect our devices to the radios as if their Wi-Fi networks basically and so we are now meshed We have a basically a LAN by way of mesh tastic and so it's an off grid I mean like no one else can connect to that as decentralized We could probably set up, you know radios all around and they all would connect to each other and it creates a mesh network that you can you know Run your apps run your comms run run whatever you want all encrypted Excellent battery life Optional GPS features. I mean this is a cool project. This is meant to be radio based not a device I mean because they're making Raspberry Pies are they thinking the Pico?
I think in particular is it only meant to be radio based not like you know literally IP based through the internet I think that these things talk to each other via this lower-run technology Which I assume is a protocol and then they act as simply like routers I see that your local network that connects so these are like bridging local networks via this hardware and software so I'm sure can run on all kinds of software but the hardware is a set list of specific supported devices such as the Seed card tracker the helltech mesh node the nano GT ultra etc. Etc a bunch of specific devices They have antennas so you can go for long long range and I assume the three hundred one kilometers has a pretty stinking big antennas on them to get that far and The question is like I see stuff like this. I think that's cool I'm really impressed and and like would I go through all that like if I was sick of the internet?
But I still wanted to communicate with people and I assume that you could probably get like a Regional networks that up. It's not just like me and Nick Nieces houses, but you could probably have an entire mesh network It's like a city maybe multiple cities. They're gonna go in kilometers a long way And then you can just have your own little network. That's just like yeah, I can run my message against Now I can talk to those people over there and Lincoln on their message now I'm just might have to get out beyond your network and talk to apples like clouds over I don't know awesome thinking now like okay, the network is one thing But then what can you do on the network? I'm kind of protocols that support is it only messaging is it kind of an API layer in a way where you know the network gets created?
And then there's an API that you can send messages is there a length of messages? There's a certain amount of bytes you can send or that kind of thing. I think That's what would describe to me like a utility because I'm thinking there's unique businesses that might want a reliable Communication pattern whenever the internet is not always available. You might be in spot areas I was just on a retreat and I got T-Mobile and while there I basically had SOS like I didn't have I came in and out of like One bar half a bar and so I wasn't really trying to use my phone anyways It was a retreat that's the reason retreat to attack but nonetheless I still kind of wanted to have my tether to the world. You know like don't take my you just say your treat to attack Yeah, retreat to attack what's that mean in this context?
Why believe okay great. Let's go there. Okay. I believe to truly attack like we have to as men I'm a man like a call to for my family and my life and who I am sure I believe to attack, you know in the most kindest way Yeah, like my father whatever. Yeah, sure. Yeah, you're gonna have the the energy to do what I gotta do
Then I cannot just do it by just constantly attacking I've got to retreat to recharge to collect to reaffirm to examine so that I can better attack So I believe at the retreat to properly attack the world. Yeah, and I mean the positive way of course Yeah, I get you so when you're retreating you don't necessarily want your phone, but you still want to be you know Connect it just in case that's right. Yeah, I mean something can happen. You know, I need my I need my fix Without your phone recently honestly while I was the retreat I left my phone in the cabin all the important times I literally left it there hours at time. Oh like 12 days. Yeah, like the whole day nice and I
Did it as a force function But it was a little strange at first I kept tapping my pocket because like right habits I have as a an individual who doesn't lose this expensive ice my pocket is there yeah So I will occasionally you know, let's say once every five and a half minutes just get Once that some sort of intermittent measure. I'm tapping my leg around touching my right pocket Which is where I keep my phone so big pocketers now, you know, really you know most people It's right. You know, multi-layer I handed it's gonna be your right pocket me. You don't know where it's gonna be so you still know back I'm from pocket. No pocket in the shirt knows right could be your armpit That's sneaky people in the armpit man. Okay, like with a little case
Good I felt good. I was proud of myself honestly for doing that because I think it's so sad to say this but I think in today's world for a Normal first world person who has access to the things we have access to the privilege We have access to the things we have access to to leave this connected device that we pretty much rely upon to navigate Like if I got my truck, I know where most things are up But if I'm going into Austin to a particular building sure I'm gonna map it That's what's there for man. You know, that's your got ways you got ways. I don't wait I'm not talking about ways. Yeah, you talk about ways ways all day the thing about it's nice about a small town living You were a small big down is you know, really ways because the traffic's just fine I'm pretty sure where you go. There's times but most of the most part you're not gonna get a major reroute because of traffic
Yeah, that being said there there's so useful and yet Disconnecting from them is so free and I had the same experience with my watch my Apple watch Which finally died this was a I haven't bought one sense Which kind of tells the story because I was like addicted to like the notifications and the rings and the things you know on your wrist And when it died, I was like, I don't know 300 bucks I'm gonna go a week or so without it and just see and it was like you get a little bit with Rawls You know, it's not like quitting tobacco, which I've also done that was way harder But you definitely get withdrawals like oh, you know, where is it you look at your your look at your Risk and there's nothing on it like it's just weird like why don't you look at that there's nothing there and then beyond that It was it's just been pure freedom ever since like I just I just feel like that was a little Prison that I erected for myself and I lived in it and then I just happened to escape and I don't need it And it doesn't actually make my life enough better to be worth all of the baggage But the phone itself is way more useful than watches and so that one of a much more connected to and addicted to But I have been leaving on the charger, you know, that's my that's my recent move I called return to analog which is like let's just focus on the analog
Let's get away from the digital one more away from the digital Let's actually get away from the digital and leave the phone on the charger and This elongated this conversation a bit like there's this resurgence for a lot of folks to be against the the norms Like drinking has been a norm for a lot of folks. There's a lot of people like I'm now sober I haven't drank all year. I quit drinking in January I want to quit drinking haven't had a drink since and so I'm free of alcohol my life and never plan to go back And I've never felt better my whole entire life And I've got so many friends that are Either so precarious or going there and it's become a trend for young folks You're folks who are just like basically anti anything that is not health conscious Yeah, it's like my routine is to get up at 545 every day journal for 30 minutes work out for an hour and a half And then go slay work and then spend time with friends like that's the dream of a lot of young folks these days And that was not the case for me when I was like you better get you about their work, okay You better find your career and make some money you better build a family and have you know those things But a lot of things are like this returning analog this return to soberness this return to just putting pure things into your body and less toxins less
Poison's that really shape a lot of Americans lives. Yeah. Yeah, I saw a trend recently amongst teens where they will sit 15 minutes or whatever the time period is and they'll set their phone up and they'll start recording which is to keep themselves It's a task basically because they're gonna publish it so there's still like digital component to this And I'm sure it's a trend so they want to get all the kudos from their friends and stuff I put it up on the internet, but they just sit there so I put their phone it's almost it's not meditation What it's just like I'm not gonna listen to music. I'm not gonna have my phone. I'm not gonna do anything I'm just gonna sit here for 15 minutes and think and they find out it's really hard But it's like a challenge and they're treating like a challenge like can you do this and they all started doing it So yeah, these things are anytime. There's a move in a certain direction, you know eventually there's a counter move in the direction
And we're starting to see some of that but back to mesh tastic you mentioned you might use it out there and you know No man's land, right? That's actually the primary use case. I think for this which is like to do yeah in the boondocks It says right there. It's for areas without existing or reliable communications infrastructure So I'm sure there's there's lots of really good uses for this in areas where they don't have you know fiber runs go into their house so Really cool project I wonder if we should dig deeper and maybe get some extra folks on the show and hear about some of their cool use cases Because I'm here and Tia people are using this. I would love to explore how to like I don't have a big enough brain I guess to explore how to potentially Use this unique way besides the ones we've talked about like what are some really really interesting uses of it?
That's what I love to explore Yeah, really really interesting uses probably out in the boondocks where there's needs that we just don't even know about you know Well, like even I wonder if you can connect Imagine your a 10 location boutique Retreat or hotel right most your places are in the bones, but you're literally all over the nation is there a way to strategically Place nodes to enable this mesh network so that no matter what the connected cases for everyone else You've got some version of your own private network bit on top of this mesh plastic. Mm-hmm I mean, that's what you just said, but it's a bit more expanded, you know, right?
How can the case of like multiple states? Maybe you'd probably have some sort of a VPN connecting the two networks and have and then the mesh would be like an extender into the N So I assume if you're like we want one big network across multiple states And then when we're you know at each location we reach out two miles to this cabin or whatever it is Then your mesh tastic would be meshing from your cabin back to whatever you have at the headquarters there in that place Which is just on your local network and then that would bridge via VPN or some other solution to your other headquarters Which happens to be in Tennessee or whatever? Yeah, you know, I found myself exploring more of his specifications I never thought I'd be such a thanks nerd vacation or either creating them or exploring creates specifications Not myself and my little a buddy's of course I got you so you're still on some some snacks
I mean you call that I don't think I've got that I call a man's lot is another man's treasure So I sure what kind of specs you write in man? The most recent one was a way to version a go CLI Because I'd gotten a couple now and I'm like they're all versioned weird and I'm not really writing all the code I'm reviewing it I see the patterns But I was like can we just kind of create one unified way to version go see lies that are in this in this realm of what I'm doing And so I created a specification for how I want to Utilize versioning inside of a go see line and so now I just point the next thing I'm making to that specification and say it here To it and it does and life goes on it's it's amazing. The other one most recently was I swapped out 7z because it's actually not that good It's good. It's good, but it has some flaws that I was not personally aware of and so now I've become exposed to Z standard Which was created by meta? It's a compression algorithm and so now my new things. I'm combining tar which is
Has to the test of time right the tar archive protocol with Z standard Which is actually zstd and so now I'm looking at these specifications thinking okay great if I'm gonna implement the arkiving tool that that bridges the gap What I'm calling? Z standard from your mortals is what Zarch is about see Stanford mere morals. That's right Z Stand which is the compression algorithm that metamade now. It's just the compression I'm it's not an implementation It's a compression algorithm and so to archive with tar and use Z standard well, you can use that pretty easy with the command line But there's nothing there's no API. There's no sparse behind it There's no CLI and so I'm building a seal like I archive a lot of stuff that you may know And I want to do it in the best possible way so my most recent thing was going I don't really enjoy this as a nerd too Like I'm I'm going deep. I'm not that deep yet
But I'm trying to go deep into the tar format when did it begin? Where does it get utilized because what happened was is I had this really great tool but I was using it on arse Linux by the way and And it just wasn't working I kept having CRC errors when I was trying to like compress something and test and like what in the world's going on Here's a drop box is it the way I've used our clone is it is because I never really tested the memory because I'm trying to make Exeter memories it never really tested the memory properly like what is the root cause of? 7z not working properly on arch well, it turns out that 7z has some flaws that I was not aware of and I didn't experience this flaws on Mac OS they're non-existent there so on Mac OS toolers great on Linux does not and so I'm like why can't spend my time building this because it's That's lame. I want the good stuff. I'm gonna put myself a glass of compression I want the whole glass and I'm saying and so I went back to the board I'm like what's going on here and so obviously tar and I learned about Z standard and so to round about back to the specification I Have I'm working on and I think it's kind of there
It's at version 1.1 or you know 1.0 point one or something like that. What I'm simmering Jared Okay, I'm botchness a little bit But I've got this specification for how to implement tar in archiving tool And so there's a lot of stuff about the way you'd manifest a JSON file first and in what order you would actually create the archive to Properly support packs or POSIX and things like that in a tar format so that you don't have checked up archives And then you can press them on top of that was the standard and so I become like this little specification there I just like love creating specs because I can point at them when I'm working with agents and say that's the best version I like that version that's a version that represent represents the software I want to make and so I feel like these specs one I'm learning and two I'm defining what I think is the right way So that when I repeat myself I repeat it against that standard gotcha. I'm a little short. So you don't specter in development. Oh absolutely I'll be long. Yeah, literally I'll be like literally like when I'm doing it. Yes, that's the way
I'm it's a lot of planning a lot of thinking which I find is the The weirdest state of play I never thought I would find myself in when I first heard about vibes coding I was like, oh, that's kind of stupid. What are they doing? You're like, gosh, just he just learned it code already And now I'm like why why would you go through such depths when you can have your rust and your go and you can eat it too? Like let's do it, right? And I can just work with the AI and within these realms. I'm reading a ton of code
I'm loving it enjoying it. I'm not personally writing it But I'm the one who's visioning different things and so the this move from seven Z arch to Z arch is a has been fun While your code cup run over here all the time I thought Z STD was what German folks get when the two permits you know, that's right SDD It's the standard and thank you for making it cool. How is it better to compress is better. It's faster What's what ways do you measure better? I don't have the facts fully in front of you It's only go there their website and see if I can get it. Okay, so compressor name
You got Z STD right now. It's not the German version of it. You got Z lib You got quick quick LZ you got LZ for you got snappy and various other ones that they Compare against and so when Z STD which is short for Z standard compresses it can compress at a at a ratio of 2.896 I'm reading this chart here versus let's say LZ for which is 2.101 so those are very similar section pressing at a better ratio LZ for will compress at 675 megabits per second whereas the compression speed of Z standard is 510 megabits per second so it's pretty fast and it decompresses just as fast so it has a similar compression ratio But it's faster on the way you know way out and one of the biggest issues I've had with the thing I built was like my gosh It takes so long to compress and it takes so long to unpress like unbearably too long so long I was like why am I even doing this and so?
But I like that because I've now learned so much more making all those mistakes And I'm like man I I was kind of upset going down the wrong road and really happily going on the wrong road. This is awesome. It's so cool It's useful and it is useful and so all I've done now is basically swap out that to I made with a different compression I rhythm and then now I'm like Reimplementing against this tar spec how to best implement a tar archive and so I didn't lose a lot of runway in terms of what I built in time I just swapped out the format and Reimplemented the way of those tar and we're down the wiser so now we're back to square one But I've learned a lot going down the wrong roads and I now have a brand new version of empathy I never had for what I would call developers right software developers What I was saying myself a pseudo software developer You know this I've been an imposter for many years and if you didn't know this now you know Yeah, I'm like wow This is what it feels it to go down the wrong road and not be pissed and still be kind of happy because you learned so much along the way Yeah, it's kind of cool and then later on people combined say well why didn't you just use Z standard?
You're like well because I didn't know it existed. I didn't know existed I had to make those mistakes I had to try it on on on arch Linux see that and and be like why is this thing not working? I want to go back to root cause which is a great principle right go to back to root cause analysis Where's this problem at what's causing it and then you know what help me believe it or not stack overflow? Yeah, I was like comments on stack overflow I'm gonna spread it about issues with 7z on arch and if I didn't find those I'd have been like 70 works everywhere It's amazing. It's awesome Which I thought and it is kind of awesome, but it has some flaws that I just can't live with that's how progress works Yeah So in the real world now so we're looking at a compression speed versus ratio image that compares is he live is that's behind 7z or that Different thing that is a different thing. I think a 7z
Uses um, I think it's LZ ma4. No, what is it? Gosh your testimony memory here, man Hold on a second. I'll get you some information. Yeah, I was close LZ ma I was right LZ ma2 and so there's LZ ma the original Compression and then the fellow who wrote this This software his name is all over the actual stuff. Let me see what his name is I can get so quickly 7z the archive format
Was created by so it's both a format In an archive tool in one so it's not only the compression algorithm the format It's also the tool that does the job and so it's made by a fellow named Igor Pavlov And so he I think he rewrote LZ ma if I understand the picture correctly and LZ ma2 is his Compression version of it that 7z uses it's great. It does a great job. It's just notoriously slow and Do you like slow-dared slow music slow jams slow compression algorithms? No, I ate it. You know internet No, I ate the worst right slow is the worst. Yeah, so that's a slow jam. That's good several things are getting me
I was like why does it takes a long to the world win for you So that's like you're compressing one of our episodes and it's like 10 gigabytes Like do you know the time win for you like it used to be 20 minutes now? It's yeah, so it used to seriously tap my CPU like I use this this application called I think it's clean my Mac Z or X clean my Mac X Yeah, I use the same one and it hangs out in your system training right it's up there And if you go and tap it you will see CPU and it will tell you the temperature of your CPU And so whenever I was compressing with 7z or 7z arch Like it was always as as high as possible and it's like hey listen if you don't stop doing what you're doing your computer You know blow up basically having that was being very kind about it and I was like ah it's an app What's an M1 it's got it?
So I just never worried about it so it was really you know really hurt the CPU for quite a bit and it would take I Mean now that we have like 20 gig 30 gig projects Easily 20 minutes if not more to the point where I'm like I Forget about it, you know and it just became this as you know You know about our drop-off issues It became like I can't keep up because this tools jacked up and I'm trying to compress in archive in a way that gives me a little Compression but gives me an assurance of a single file that I can get back later So I want a good in and a good out and I thought 7z was that and it just turned out that it was great for Mac So it's great if you're only on Mac at least I might experience that could be wrong If you know more about 7z and I'm way off please call me up But that's that's been my experience well one thing about 7z has it been around forever I think when I very first are using Linux it was there and so it's old technology and There are reasons why progress, you know makes steady sometimes slow sometimes a little back stepping but eventually forward motion and improvement So What does that same 30 gigabytes? Are you on the Z standard now like you've completely replaced and you're running and you're
Working on it. This is super super early days So I can only be like my latest greatest but time wise so much faster like I was like what you're done already There's no way and I would check it flow like from 20 minutes to five Yeah, like I'm god I was like so fast like three minutes five minutes maybe ten minutes on like a so we had a couple archives that were like in the 50 Gig range and that was about eight minutes So I'll be in like the three-minute range for most of our stuff and we can if I do it right if I like if I do this This tar specification will enough then we'll have a really good Manifest not JSON at the very top of the archive like inside the the archive container the you know the thing that makes up the Tar archive and it will have all the cool stuff It will have one of the other things about archives I've learned is their attack vectors And I'm not a I'm not a pentester and I'm trying to like hack things But I'm obviously painting like you do but I never considered that wow I should probably do a better job with this because If I don't do it right then I mean I don't think archives it or a jeopardy But if I open source this when I open source this and others use it and they start getting jacked up because they're attack vectors
Why I gotta do a better job of security and so when I started to craft that specification It really exposed me to like a couple of different points around what was important and one of the more important points was being secure And I never thought about that so there's like certain things you could do with falling file paths and when to cap it When not to do it and again, I'm not super steep in it, but it's it's quite tantalizing once you have this I Would say AI is like this guide that can learn or knows a lot about most of things You want to know about and if it doesn't know it it can find a way to learn with you and help you learn too And so I find that that thing that ability into these age is is just so tantalizing the fact that like if I have a curiosity It sure it may hallucinate it may go down the wrong road for a little bit But for the most part I can explore some territories I almost would never have before not because I didn't have the ability because the guide didn't Was it resilient to my path like if I hit a blocker AI finds a way around it through new information or different path or an alternative Whereas a human might be like okay
I've hit the end of this book the appendix isn't helping me and Amazon's down for the day because the list is how to tap And I don't know whatever it might be like I can no longer explore over this hurdle But AI finds a way to help you through all these things Such a wild thing to have to be able to talk to our code base in ways we've never been able to literally ever in humanity That's that's me is what gets me excited about software today. Yeah, it's a entirely new experience and one that is quite Interesting at the lowest and exciting at the highest depends on the day and the concept and how else performing Oh, I mean I want to punch the screen today. I was not supporting now like listen listen I gave you clear instructions. I know you're smart. You're going way over there. Can you please stay here right here?
This is where you're safe at don't go over there. Don't do that. Don't touch that stop changing that file I'm just being for these ones. It's a version of that's like yeah, listen to me. Please we have a mission stick to the mission over there Looking at squirrels got here You know, so I mean sometimes it listens well sometimes like listen I'm gonna do whatever I want to do and you can't stop me because You don't have any control me. I do what I want. Yeah, you know
Yeah, one thing that I've been doing recently which is give me some success requires a little bit of patience in one direction But it helps me avoid a lot of the tail spins is just like you know, basically rebooting the AI and just like Keeping my conversation short and the context windows fresh, you know And like all the backlog of things that already knows about the project like I just may go read the files again So I will do like a couple of features and it's the first time I see it be like You can almost notice like oh build up a hallucination path or something I don't know what it is But you can kind of just feel it and I'm like I just control see that sucker and just not fresh And I just tell what I want to do without any of the sometimes like dang I already knows this part and already knows we're working on this and I was like now who cares I can go figure it out again That's the patient's part It's like your files and see what you've been up to you know, you know, you know, you just knew that until I control see you Now you don't know what I'm up to but I feel like it's this is Claude code specifically which is the tool I've probably been using solely for coding for the last two weeks. I have been using anything else It's so much better on its first feature than it is on its sixth. It's just better
And I'm not sure why I made it because it's just like it doesn't have any baggage That's the way I kind of mentalize that yeah, and so I Just control see that thing and like let's just have it's like it's like a brain white It's like I'm gonna get a new cloud code doesn't know my code base anymore And it'll figure it out I know the feature and it's gonna be better than if I have the one that knows the code base But also has like I don't know needs to be defragged or something have you ever tried that or you had experience with starting fresh? I know you but the code and you do that. Okay. Yeah, I have a process where I keep a context file like a context dot MD file Mm-hmm, and I keep a context dash guide Dot MD which is how I want this context file to be structured if it if it needs to be structured So if it needs to be deep so it's like multiple agents multiple features can keep a lot of things in check And so if I'm like hey, you're misbehaving just dump your context of this file and I'm gonna clear context It's like a threat and it's a very kind I'm being kind of just joking around I say that it's like you're gone too far You're not operating well. Just don't whatever context you have what you think is next to this file and I'm a clear context I do that happily and I have a command like an unfurl command
It says bring yourself back online read this file and measured against this guide So this is like you read it and you know, report back when you're ready kind of thing or confirm you're ready Okay, and it's what does it goes and reads that dumps its context I clear it comes back as you bring yourself a line say I'm in Westworld or something like that Which is where I get that line as bring yourself back online Dolores is is from Westworld and so they do that and it checks It's like okay, great. I see you're working on this. I'm gonna go that feature. We want to do next I'm like sweet now. You're smart again. Let's go
Gosh, it's a slightly different things. I don't let it remember anything I just tell it like I don't want to have any of that context at all I just want to be like a new human and so it's more like the men in black thing, you know, you where we have here Exactly. We don't I'm just wipe it memory. I'm just like no context. No nothing. You're done
I'm gonna open up a new clock code on the same directory I'm gonna give it what I'm gonna tell what I want it to do It could be the it could be copy paste when I told the last clock code, but this is a brand new clock code doesn't know I don't think it has a cloud that MD so it has some rules that can follow there and that's about it It's gonna figure it out from square one and so that to me is like a full reboot versus maybe you're just doing Like a sleep and wake back up thing. Neuralizer. That's where I can think of in my analyzer. You realize her I would review that context file So I don't like blindly just let it do that It's a way for me to know what it is thinking to over what it thinks is next and so if I look at that whatever it dumps It's way off Then I won't tell it to you know bring itself back online by reviewing the context I'll just either delete it or skip it But it's away from you also to like have a an awareness of what it thought was going on and how far off it was from the truth And it sounds like wow that was not true. That's a hallucination there We're not working on that feature or that's not what's happening here and and I won't do that But it's a way to save myself from having to type as make keystrokes or do something to like retail its context or even give it In direction I can just say review this if I agree with it of course and right and then go from there
Okay friends augment code. I love it. This is one of my daily driver AI agents to use super awesome CLI VS code jet brains anywhere you want to be augment code can bring better context better agents And of course better code to me augment code is my fault Most powerful AI software development platforms to use out there It's backed by the industry reading content engines the way they do things is so cool You get your agent you chat each next edit you completions. It's in slack You see a lie they literally have everything you want to drive the agent to drive their context to drive better code for your Next big thing for your big thing already working on or whatever you have in your brain You wanna dream up? So here's a prescription. This is what I want you to do
I want you to go to augment code calm right in the center You'll see install now and just go right to the command line There is a terminal CLI icon there click that and it's gonna take you to this page that says install via NPM copy that pop into your terminal Install augment code. It's called Augie instantiate it wherever you want to type in a ugge and let loose You now have all the power of augments in your terminal deep context custom slash commands MCP servers multi-modals prompt enhancers user and repo rules Task lists native tools everything you want right your fingertips again I'll be code comms for my favorites. You should check it out Well friends this episode is also brought to you by our friends over at Nord layer and you know scare them black Friday chaos Yeah, you guessed it leaving your business network wide open to ransomware fishing and whoever's bored enough to poke around your systems And that's a lot of people that's exactly where the little air comes in it is a toggle ready network security platform built from modern businesses It combines VPN access control and threat protection and it's all in one easy to use setup There's no hardware no complex configuration to secure connection and full control in less than 10 minutes
The cool thing is it is designed around zero trust meaning only the right people who have access can access the right things Here's the best part 28% off. Yeah 28% off the yearly plan and the next step is to go to Nord layer calm slash the change log Yes, Nord layer calm slash the change log use our code Change log hyphen 28. Yes, life is important again change log hyphen 28 use that code before December 10th 2025 for the awesome 28 discount and the video is you also try to risk free with their 14 day money back guarantee keep your team secure You're never clean and your mind at ease once again That's Nord layer your toggle ready shield for modern internet go to Nord layer calm slash the change log and use our code change log hyphen 28 Well, let's turn now to a new AI roll. That's exploding. We've been discussing our AI rolls That's right, but we are not officially employed by anybody. I guess we're officially employed
I are our own company, but we aren't certainly passing our resume around but if we were we will want to be have the acronym FDE On our resume. Yes, because that's the new hotness according to interview query calm Reporting on a report as the internet does gladly and I'm actually reporting on the report of the report So if you're listening to this and then you go tell your friends about FDEs you are four layers of Inception financial times report talks about monthly job postings and this new role FDE forward deployed Engineers are looking to get hired. There's been new job postings to the tune of 800% Which of course tells you it was really low before now. It's not as low and I mean it could also be from like nothing to some and the universe has much easier when you have small numbers but nonetheless it's moving and it's a new position led by The four runners as I call it in the a race and profit company I etc
People who are out there building these? The models but also the software around the models and trying to nab up as many customers as they possibly can before their money runs dry to have a new kind of engineer which is somewhat novel I would say mm-hmm I call it a I led because Well, you're kind of leading the AI, but the AI leading you I don't know who does Wag the dog or the tail wag the dog. Can you say that doesn't look at least who's holding a leash? Is it the humans of the dog? That's right. Does the dog wag the tail or is the tail wag the dog there?
I got it the question is that and the answer is we're not sure but If you want to be an FDE you're gonna have to wag the AI Because their whole point is to be deployed with customers is kind of how they're different than a typical software engineer They're like one-on-one with customers and they're tailoring AI models whether that means rag whether that means fine tuning whether that means Literally just setting up the right model to use and providing it context I don't know if that matters so much That's probably up to the FDE decide what that what is tailoring in the context of the customer and Working directly with the end users in order to give them what they need. Mm-hmm your thoughts on this new thing that Isn't it well? I have news Jared. Okay. I have met Talk to and experience FDE before recently We're working with Redis on engagements. They're one of our sponsors love Redis
I'm actually quite excited that Redis will respond to us like they've never heard is cool You know it's so cool, right? Yeah, and they obviously have a lot of stuff around their vector search You know Redis is really this you know in memory cache that's utilized in so many places But they're really solving some cool problems around the AI vector search and embeddings and like just I don't even get it Like I don't even play in that world very well So I don't really understand the full spectrum of it But I sat down I talked to Tyler Hutresson from Redis and he is a forward deployed engineer Oh, yes Yeah And his entire job is to obviously talk to people like me in some context but his other part is like the 99.9% of his job Which is not talking to me is being a customer like literally being there for sometimes multiple weeks or a month And so he'll kind of come in and come out and his whole entire job at Redis is to Not really at all sell Redis by any means it's not his job His job is super about what are they doing? How are they leveraging Redis and how is it working for them?
How can they make Redis better for them so kind of like that though kind of like either do integrations? They'll do migrations. They'll do like they work there like they work for that company But all they are to Redis is a customer. They're not paying Tyler any extra to my knowledge and Tyler actually works with the team of these folks Like that's the point of the forward avoid is like you are not just with the customer You're like kind of working for them. Yeah, and I mean you got to be fully embedded to have the empty factor Like you deepen the in the water with them you're waiting with them to kind of figure out, you know Where's the next drop offer? You know, where's the where's the land? I don't know. Oh, it's the point of waiting the water
I guess just when maybe who knows You're knee deep in something right? It's not a thing about what it is sometimes it's water That's right whole job is about figuring out how Redis is being utilized implementing features taking that stuff back to product Product iterating it out now it becomes in the next version of it and so it's like the rinse repeat cycle I think that's it's smart it really is smart because You know, you got that boots on the ground mentality You've got an engineer doing an engineer's job and their whole entire job is the success metric around it is just make the customer enjoy using it more And solve the problems run and hopefully if that is a product feature We should support bring that back to product Let's make it a true feature and you've got as best as you can truth Which is customer usage how do they fall down with it?
How do they try and with it? How do they work out in the in different deployments or on different architectures? And you take that back to the product and you make it real and now they're better and other customers and their problems better and obviously if It fits into the red is meant this is not a red is that but this is only context. I'm sorry I'm sorry. Good for you Redis you you got this one for free you contacted him if you if you if you Was I saying now if it fits into the red is way of product? That's great. I mean like yeah, that's that's really a
That'd be a fun job. I think yeah That doesn't cool. Let me bring this back to a company like anthropic here And out of a red is because that makes toll sense in the context of Redis where I think it's going with these AI model Situations is you have a company like anthropic who? once other companies to hire Claude right to use Claude as much as they possibly can and to get the most value out of Clap that they can well as we've seen lately there has been a difficult path towards Successful adoption of AI in the enterprise with you read whether you believe the 95% numbers or not our recent show It's on get a key does not believe those numbers you can go read as block was why but certainly there's plenty of Projects that have failed to adopt AI inside the enterprise like the promise is there you get excited you go out And you just like well, let's just slap it on something and see if it's just all unicorns and rainbows and it's not And so these FDEs these four deployed engineers that were for anthropic like they would perhaps go in and be deployed into a fortune 500
Or whoever it is that could eventually or maybe is a small Claude customer What could be using it 10x 100x more if they're having success? And so they are there to help them wield that model in a way that get a huge value out of it And I think that does not get a fun job. I mean, that would be fun. Yeah, I think so too. I mean, yeah Wow, I mean what time to be an FDE, right?
That's right. So if you're not an in brain, it's I will tell you that if you're not an FDE and you want to become an FDE How do you get there? Well according to this article? That's a summary of another article that I'm summarizing for you They say this is a clear signal of a shift from research to real to real world results And there's a clear message that AI fluency isn't enough. So if you're fluent with AI tools not enough customer facing AI skills are the real edge It says the most in demand tech workers in 2026 are not just coders But also communicators, problem-solvers and translators between AI systems and human needs So there's a short list of skills to be acquired If you have those skills, you know, throw FDE on your resume If you don't acquire away and I think that while this might not be the end I'll be all title because you know every couple years We need a new title. This is like the new data science. According to this, maybe it's a flash in the pan
But if it is if it does have staying power then those are the skills you're gonna need. It's not just the ability to direct your Agents in a successful way you have to actually help other people direct their agents in a successful way Yeah, that's interesting even the way it phrases this it says I'm gonna reset again because like you said it good But I'm gonna say good to say better. I may be all second. Who knows? Let me try. Okay, the most in demand
I'm gonna you're doing bad. I'm just laughing for the fun of it. Like it's it. I don't even want to try. I don't get it The most in demand actually this is so this is profound. That's all I read again. I'm most in demand tech workers in 2026
It's not this years. It's 25 are not just coders developers But also communicators problem-solvers and translators between AI systems and human needs and that is such a wild thing Because there are so many folks out there like that. So when we talk about you know disruption and job replacement How many folks were in product teams or the solution makers and problem-solvers spaces where they were sort of move around specs? They would say what do you do here? Right? Why I go get the spec from the customers and bring it to the engineers, right?
I'm doing right here, right? Right? And they're like well gosh everybody's taking my job I'm gonna take my job now Well, not anymore because you can now go into this role because so many of those people they may be displaced because of change or disruption I think that's a it's like a snapshot of a lot of those folks go to that go to that Mm-hmm, and if it's gonna be in demand next year and this episode is true This is November 11th 2025 when it ships. It's you know plus four days to the math. I can't I'm stupid Go and do this job being FDE for anthropic or the anthropic likes. Yeah, all right. Let's do one more
Let's do one more and we'll call it a day dead framework theory. You know, I like these theories, you know So many theories today, man. Gosh. Don't be like the live internet through. Okay. This is right. This is the day
We're back to the dead. This is different. Yeah, the freaks is the opposite of life This was written by Paul Kinland and he actually wrote a piece last year which I included in change I was called will developers care about frameworks in the future and He is thinking he was kind of wrong about that But he's wrong in the kind of a weird way or a timing way what he's found and what was news to me Which is why I definitely had it include this is basically now that more and more people are building with AI agents It has it has pushed react into like this foundational layer of web development Which he thinks is unassailable at this point not only do the models know a lot about react because they've been trained on vast amounts of react docs and examples open source code But it's also in the best interest of the tool creators if they want their customers like cursor for instance or V zero or Bolt these tools these vibes coding tools or whatever you want to call them They want their customer right the developer and user to have the best chance of success And so they're just picking react for that reason so it's like
It is not the winner of the game. I it is the game now and so far so much so that they're Hard coding it into their system prompts. So like if you're like hey make me a web app Like the system prompt like make it a react app. That's right And so that's why the then framework there is like maybe frameworks are over not for the reasons that Paul was thinking last year But now he's thinking because basically react is it substrate fully seated. Yeah, you know I find interesting about this is Is just that is that like if you're gonna code something else Let me rewind like think about the one trust here is maybe a layer above that is that what if future engineers 26 engineers, right?
Well, that's so close that is the future obviously but it's so close that's not true the future But it is the future but what if in 2026 engineers job shift from Innovating and handing it to humans, but they're innovating it to hand it to a LM So the LM can then be using a well-defined standard now in this case react has had 20 when not 20 years how many years? 14 Eight years 12 years what years it come out? I can't remember 2014 maybe yeah, so a lot you know Decade at least yeah a decade. So for a decade. Let's get the number. Let's get the 12 years ago. I drilled it 12 years
Okay smart man over there good for you good. Yeah, I guess good So 12 years all right, that's there but what if future engineers like I almost almost want this as a developer I would rather have a super good engineer who's way better than me who really knows the things teach the LM That really good thing make the framework make the standard define the specs even so that whenever I utilize the AI and those kind of things It's just kind of there It is the substrate like react has become this case in your example here. Mm-hmm like what if What if the LM already knows how to write the language? So I don't have to learn how to write the language I can learn through using it But it already knows because it was taught by an engineer It's almost the same concept like it was taught to use and deploy really good well-formed Some may not like react but really good well-formed react application because it is a substrate It is the winner and so why debate it and just hard code it into everything because that's the way it is the the standard format It's got the most battle tested tests against it. It's been utilized in all these places
Ensure it's got its blemishes and its flaws, but it's the standard So I fight it kind of the same thing But the opposite way which is like not the react way where it was born 12 years ago and it's become that through usage But the opposite way where it kind of comes in from the top It's like it's teaching a limb at the top and it's coming down to the LM through my desire to build out new things In this case, I'm like go and rust or pretty staples like I'm trying my best to never touch us again Because I'm just like that good. Okay. I'm just not that good go like an hang out there nothing against go It's just easier. It's easier to read it's easier work with it's kind of more fun for me at least and I feel Like rust is so cool, but I'm just not that cool to touch it. Okay. It is it is no doubt super cool
I'm just not worthy. Yeah, I'm tracking what you're saying. I'm wondering There's like a gap between here and there obviously like the world that you're painting the world that we live in and ironically perhaps the reason why at least Paul says that these tool creators and what as in model creators are front loading react is because Developers need to maintain the apps that are being created and developers no react And so you're giving them the best chance of actually maintaining the produced code Because of this deep-seated knowledge across the industry of the last 12 years of how to build react based websites All right And if you're gonna top load something else that we don't have to look at then I couldn't imagine react being the technology of choice There would be no reason to react because their browser already is a platform And the agents could speak directly to the browser via the DOM and wouldn't need something that's maintainable Because they're the ones maintaining it and they already speak the browser You know I'm saying so it's like there's because the humans are still in a loop
We are getting react apps, but if we took them out of the loop we wouldn't need to have react apps anymore Oh, but we aren't out of the loop and then we'll be anytime soon Yeah, I don't know if it will be there. Let me clarify it as you were describing that it it came to me more points of clarity Okay, so what if really good go developers? We're really interested in or even employed to make sure that cloud code knew exactly how to structure a go CLI So whenever I'm a Smithman UCLA I'm not having to start from scratch with context lost or what is the problem solving or oh because it has these these specifics like what if It came from top down like I don't have to think about structure I don't have to think about where the commands go where the internals go where the security goes like all these different things It just it just is because someone who's really good with all the go idioms is teaching the LM So that I can start off with all these sort of like not I would even call training wheels, right?
This is a dank tank, you know I don't have to even get a bike I know right to the fighter jet you know, which is totally different I got two wheels and two wings and massive and really high speeds right Tom Cruise all good stuff Now you know how your treats to attack right there and it's right your debt Let's go and even so far as like versioning like I had fun doing that versioning that goes the alive What if it's an yeah, somebody already knew that and top that's an It's just yeah, it's just an implementation you tell I don't have to do that right? That's that's a I don't know how that flexes out over time But that's me is cool like why should I have to learn all these things when the LM can be taught it from a human and even learn from the AI and this weird second goes on and on but it kind of comes down from the top Why do I have to sort of start off from scratch telling my LM like hey, you know, you totally botched the structure and now we have to You know a day or two hours refactoring like a day in AI times like one hour, right?
I was like we got two weeks of this like you're gonna be don't this like a minute I'm not even working about these things right? Yeah, so that's that to me is kind of a weird thing That's kind of a lot of thing to think about yeah, I see you're saying I think a slight trepidation from any sort of top-down design and the reason why it can be bad is when it doesn't fit Your circumstance but you're stuck with it because that's what the agent knows Yeah And so you now have to use you have to build a go CLI the way the agent says to build it because that's how it builds it You're not actually part of that conversation at all and if you tell it the change it is like sorry I can't do that Adam you know my ways are not those ways You have to go right everybody has their own opinions and their own taste and you and your taste clashes with the top down Well the tops of the top and you're the bottom So you know or even things in rust like memory safety like hey, don't ever do it that way. That's the wrong way That's always the wrong way I don't care if you're right a rusty Li or a rust back end or low-level tooling that taps into eBP F whatever like don't it's never wrong It's never right never do it that way. There aren't so many absolutely it's not for development
Why do you think it depends is one of our most True, okay fine. It depends I can see I can see Once I still have once I said one right. Yeah, well now you're a treat and I'm waiting for the attack waiting for him to attack He just retreated a little bit. Okay, so good thoughts interesting I can appreciate the depends I've heard about one about one million times in my life on this podcast and throughout my career So I can empathize with that I think my my dream is that helped me think a little less so I can be second thing about vision product and user experience Oh, I think that we could possibly get there without the top down approach I think that the more that we codify and Continuously improve on best practices and then you know feed them back into the systems Then the system will get better at working at increasingly higher levels I don't think if we take us out of the nitty gritty that we want our agents creating react apps any longer I don't see the purpose of that but in the meantime I get it I also think it's kind of sad because you just have this like accident of history now that react isn't acting in the history
But you have like this timing and this talk about codification a A freezing in time all of a sudden look at this chart you look on this chart So on this chart has it's crazy. So this is according to built with calm He says there were 13 million plus sites outside the top one million deployable react in the last 12 months And so there's like a straight-up chart if we look at react easily over time starting in and I mean it got It started gaining in like 2018 up through 2022, but in 2023 aka coding agents starting their coding LMS telling you how to write code now they're aging agent for us agent Agent Aquino to verb there is but it goes straight up I mean it's just insane how many new react sites are coming out today to the same comparison to WordPress in a different era Right like at one point by and large anybody developing a blog or building out a basic website for you know SMB's medium sized companies Even some larger ones. Oh for sure press right same curve. No LLM. Yeah, I mean when you are the de facto That's what happens. Yeah, so no good nothing is that but I think that's no
I'm here to blame reactor anything I think that I think it's different because there was human choice involved and there is human choice It's just like two degrees of separation away. Yeah, it's like this self self redirecting feedback loop Or it's just like reacts popular because reacts popular because reacts popular and we're not anywhere Like when do we step in and say yeah, but we don't have any we had a chance to say yeah, we're pressed is cool But what about what else people pull up? Ghost here about ghost what about us suppose ghosts? Yeah, sure. They sort of have more choice like Ruby on rails and different There's that yeah, the drone We're press. Yeah, we still build our own things
I think that what we need to do in the future is just build everything ourselves, you know, just like who needs libraries That's what I'm that's where I'm headed if the agents get us there. I think it is there Yeah, explore that a bit more How do you mean I mean I think testing frameworks just important protocols and known processes that work that are battle tested fast readable, you know digestible not against protocol. I'm just talking about libraries and frameworks right certain protocols and Common, you know bits like glue everything together Here's a quote. So I've been working on a blog post and not ready to say much about what I will say okay I'll say this much our friend Lewis via was writing his newsletter tonight read and he's been Agent to Cody now for about a month like in earnest like having fun build stuff just like you are and He wrote out some observations after a month of doing this Yeah, and he says eat here's one of his observations each of my new tools relies heavily on false libraries for you know
But I suspect many fewer than would have been the case in pre LLM code Hardest if it makes existing core libraries even more important valuable or what so he's finding what I'm finding Which is like there you still got your react, okay? But all the other stuff that used to pull in All the other and still every library is those utilities those helpers Especially the smaller ones It's like it's a burden So yeah, just tell your thing to write a thing that doesn't exactly what you need doesn't need to be generic doesn't need to be abstract You can be exactly what you need nothing more and you have way less dependencies over time And I think that we're at a point now where we can start to trim our dependencies around the edges and re implement They're in limited functionality just for a own use case and remove that and then over time Bill was less and less dependencies. I think that I think that that in certain ways is unfortunate But in other ways is kind of a good thing like left pad for example No one's for sure left in this future left pad will be less I'm probably never being a dependency again or I was never be specifically but you know the idea of left pad being a dependency like that in
This new world and LMM can easily replace that thing and it would it would never pull it in this dependency Like oh, I need to move I forget what a left pad even do like patty around it like that Like I added a string on the left side, you know To a string right the way you want to use it. Yeah, so they'll don't be like yo I just did that no left pad required almost every almost every developer in human history would also have done that one That's why that one's hilarious only in the jobs for unity Would we just you know pull in a package for that? But yes exactly that let me go slightly more? Complicated what about a markdown parser for instance? Okay, I know you're getting excited Well because I mean markdown parsing is a known quantity like this is not rocket science This is there's a markdown parser in every language people all written them etc Well, what if I need a feature where it's like I want to be able to add markdown links to this text blob?
You know so you can put your square brackets around your text and then you put your parentheses and put your link in there And we're gonna take that and we're gonna know that that's a thing and we're gonna either translate it into HTML or maybe into an I-Cal thing Or wherever we can output it. Okay. Well in the past perhaps you pull in a markdown parser I gotta get the markdown library of the day or I write a really gnarly redjacks, right? And the future why don't you write your own markdown parser because it's gonna take seven minutes for glad to do that for you Mm-hmm or just or 40 seconds and it might not be a full markdown part of the support It doesn't support all that you don't need all that surface area You're not worried about bold and italics and stuff like that You just I'll use a link right and like there's a place where it's like yeah regular expression could do that But also you're like well now I want the links but I also do want the bold and tell us okay my markdown parser Just does bolds italics and links and it's 500 line the code versus the official markdown parser for Ruby which is like I'm just making a step I've no idea what it is 17,000 lines. You have to compile the extension blah blah blah blah blah blah Yeah, it's true, you know, so like stuff like that. Yeah
Yeah, that's the things you start to eat in from the fringe and you start to like eat your way into your dependencies until they're all gone Yeah, I think that's the kind of open source. We won't truly miss I think the open source we would miss like if Like if you no longer need a Ruby on rails like that level of replacement that might be a different world to live in I think those are the last things to go I think they're this take around the longest because they're the biggest ones value I mean react six run along yeah I think as long as humans desire a need to read the code or Desire to have visibility to the code and it's never because I mean there could be a world I mean I'm living in a world now where I still tell people I tell people I'm a Technologist and software developer to podcast and all things I do and then I tell them like I'm doing things with AI never thought I would do I'm literally it's science fiction to me still yet. I still can't believe what you do It may be a couple more years where that may be true like I just invented framework and it's so good doing what it needs to do that we Do we stop checking it because it's just so good It's always been good and that might be five years or a few years on the road But it might get to a point where enough of the human race who's doing this kind of work and pushing this kind of boundaries like you know
Every time the thing does the thing it's good. So I don't really care about the code really ever anymore I I used to but now I don't I'm not saying I want that world necessarily But what if that's the next evolution like it's just become so good that they could talk to the Dom and in fact It is reinvented the browser You know Dom's gone to buy my Dom the things you thought were never gonna be replaced. Yeah are replaced, you know Yeah, I will see I'm still in the opinion even though I'm gone away from my I winter stances my plateauing because I've been wrong about the plateau on a couple occasions So you know I don't like to be wrong. I'd rather be right Yeah, I'm not gonna be calling plateaus anytime soon because I've been wrong enough But I don't think we get from where we are today with transformers to that Like that I think we need some other and some other thing most likely will come around So I'm not saying it's not gonna happen. I just think that we need another step change to get to that level I think we can very far where we are let's try to hypothesize what that might be then can we do that?
Oh my goodness, man, I'm not a computer scientist. I'm a Concaster well, I have some idea. I don't know if this is true But I wonder I mean because we are moving at such a pace now where the hard things got easier And I would love maybe this is where we should go either this year if we have one more opening for us I think we don't maybe it's a first thing you're gonna think because I try to do it last year is Dive into quantum computing I think that that's becoming the new for a ground because we've got such advancements in artificial intelligence LM supporting You know large swaths of generating code that we're moving a faster pace on the things we that were hard So it's obvious like something that that will be next we've talked a little bit about like bio computing recently That was kind of wild and even that was in the research phase It was cool. I had a conversation and you know no harm to file to the person that we interviewed she was amazing But it was still research so it's just it still wasn't quite here. It was in the works They've had some issues. They've had some bit some Billy issues
They got some clear windows of where they can go but I wonder if That next invention layer like you say is not here with transformers Is solved or is because you've got Microsoft doing like some cool stuff that I've heard about where they've got like a computer in like Negative 500 degree temperatures doing something and I'm not in that world. I know all that stuff I saw a lot of something doing something like this quantum stuff So I wonder if this quantum computer this this thing that I'm not even that familiar with is the next big thing that comes sooner because we're here now So quickly in building software at such a clip that we never had before Yeah, I can't answer that question like if you made me stick my life one way or the other I don't think I would say I don't think quantum computing takes us to a GI for instance I think it's like a different kind of thing But I could totally be wrong about that and I think there is other things that are coming beyond transformers I just don't know what they are and when they're coming and how far transformers can take us but even guys like y'all McCoon, which is like, you know, he is a computer scientist like capital C capital S and one of the smartest ones that I can you know
Fathom in the industry. He says that we're not getting to AGI with transformers and so I'll just take his word for that I don't know what the next thing would be and I'm not the guy to invent it Somebody else will invent it and then you know, I'll complain about it You think a jazz required to reinvent the DOM or the browser or to make pretty stinking smart rails obsolete I think it'd be pretty good like AGI AGI probably not but something way with way more Con conscientiousness, I don't know like it has to have actual smarts not just instant recall of the world's information You know, yeah, that's my take but love to hear other people's take on that because I've been wrong before I think it was 2017 Back in 2017. I was wrong and that was the last time so we were wrong about that I just mess it around a little bit You pull my legs. Yes, I got you. Yeah, yeah, all right for a bit there great conversation. This was fun. Yeah. Did you push record? Oh
You did it. I got you I got you. Oh, that's a throwback. Yeah, our last friends episode We did together we were sad if the first few minutes of that episode go back and check it out It's like we see a little gloom because we are we forgot to push record We talked for like 45 minutes total gold and you missed it because that's it was going to dev know and that's not cool Let's leave the folks with this go right an agent, right? Yes, go right an agent Listen to news go right an agent change log. News if you're going there this week
Well, that's gonna be there for you if it's next week well, then go back a week, but either way Thomas how you say last name? Well, I think it's pronounced Tatchik because it's Polish. I believe but it's spelled potato. Yeah Thomas Tatchik. There you go Co-founder of fly we love fly. We're hosting on fly. Thank you. I think he's one of the co-founders. Yeah, okay cool
I think he's CTO. I don't know. I don't know what he is. I don't know. He's been around a long time I know he's a co-founder though over fly below fly We use fly amazing, but he wrote this really awesome read is 13 minutes long you should read an agent and I read it I agree I don't have time to read an agent But I think I'd like to it'd be a fun exercise because then you got the get the full empty factor What it takes to make one of these things and maybe you could make your own little Archie or Alfor pick a name That's funded you and maybe it's Susie. I don't know pick it. Yeah, make yourself a little agent
You know so you go and then tell us what you named it It's my right name is around here. I tried to name something Archie recently man That's why I said that because I was bummed because I think you're having tools Yeah, I wanted to call it Archie and so I think it's yeah Archie core to spell around on purpose That's what everybody else does maybe so but I think it's Archie was something that was like that There's an Archie out there and I found it. I was like, okay I can't I can't do that because this is like AI native cloud infrastructure Is that the most buzzword ever but as they do and that's they are Archie Labs Archie core don't really are really but I found them I'm like I can't call this thing Archie so I call it Z arch Nowhere near as good. I mean, I think the artist cool. Why oh it's cuz it's an arch way that's in somewhere where there in Germany It's it's archiving. It's totally focused on Z standard not other formats. This is this is the standard from your morals man Z arch
Not Archie. I think Archie would have been cooler. I'm just trying to myself feel better, okay It's like when you don't have record and then you still record after that. I mean, yeah, I mean better Archie would have been cooler. Yeah, well, you know just Smell it on purpose man. Just know it on purpose. So I guess you could say a RC H Y. I try that. Yeah, I mean, it's not done
I didn't release it. So I mean it's this is a skunk works project at this point. It's my fun name Z arch sounds cool We'll see. All right Bye all bye friends. Go code agents. Have fun. Listen to news change all that news and we'll see you. We'll see you
Bye friends So I mentioned inviting Spencer Chang creator of the live internet theory website on the show to discuss good news He got back to me and he sat down with me yesterday to talk all about it and a bunch of interesting art pieces He's building that entangle with her everyday lives and each other. It was a fascinating conversation We'll bring it to you next Wednesday Oh, and let us know if you enjoy these extended news discussions with Adam and myself Maybe we'll do more of them. Maybe we'll do less But we can't divine whether or not you dig them So you'll have to tell us and if you want to discuss with us what we discussed with each other that happens in our communities Zulip chat which is free and joinable at change blog.com slash community Thanks again to fly.io to depot.dev to augment code.com to Nord layer.com slash the changelog to break master cylinder And to you for hanging with us. Let's do this again next week speaking of next week next week on the pod news on Monday Spencer Chang on Wednesday and Chris Benson talking drones and robotics swarms at the edge on Friday Have yourself a great weekend the hand of the diligent makes a rich and let's talk again real soon