Subscribe
Share
Share
Embed
This week Aaron & Ian talk about Claude Opus 4.8, Pope Leo XIV's thoughts on AI, dynamic workflows in Claude, and the absolute hottest take you've ever heard on a podcast.
Sponsored by Bento.
Links:
Sponsored by Bento.
Links:
Creators and Guests
What is Token Town?
Token Town is a weekly show where Aaron Francis and Ian Landsman break down what actually mattered in AI software development this week—new tools, model updates, workflows, and shipping lessons. Fast, practical, and aimed at builders who want signal over hype.
Ian Landsman (00:00)
Welcome to Tokentown. Today is Thursday, May twenty eighth. I'm your host, Ian Landsman.
Aaron (00:05)
And I am your other host, Aaron Francis. And today, obviously, we're gonna be talking about Opus 4.8, dynamic workflows, burn, baby, burn those tokens. We got some mythos follow-ups, and we're gonna wrap by talking about the Pope and what he thinks about AI. But before we do that, we have to give a shout out.
Ian Landsman (00:26)
To our good friends over at Bento, BentoNow.com, the email marketing and CRM platform built for the AI era. Send your product marketing and transactional email with Bento. ⁓ check check them out, bentonow.com. We are both Bento users. Bento has fully embraced AI. You couldn't even more fully embrace AI, I don't think. So if you're into AI and you need to send email, ⁓ check out Bento. Okay.
Aaron (00:55)
We gotta talk about Opus four point eight. Neri, five minutes ago it was released. So ⁓ if you're if you're coming to us looking for ⁓ takes on how it works, brace yourself. But Ian, what what are we seeing from Opus four point eight?
Ian Landsman (01:01)
Mm-hmm.
Well, it seems very good. They gave us the little, they gave us the little Bezos chart of the AI version of the Bezos chart, which is like, look, all the numbers they go up on all the benchmarks, right? It beats all the other tools. ⁓ so seems good. I did have a chance to do one very small task with it before we came on air, which performed admirably.
Aaron (01:23)
Almost all. Yep.
Ian Landsman (01:36)
I'm still a Claude guy, as you know. So happy to see Opus moving forward here with new versions. a couple of the articles I read quickly seem to imply that they've had the people at early access had had really great results, better than GPT 5.5. so so far, so good, I think, on the 4.8. What's your quick take? This is this is breaking news, just came out like 50 minutes before we came on air here. So we haven't a big chance to play with it, but in your 50 minutes of prep.
Aaron (01:59)
This is breaking news.
So
yeah, I read the release. ⁓ I did I haven't done any coding with it yet, obviously. but I read the release and I found a couple of things interesting. One was how much they leaned on ⁓ like the honesty judgment, like how much they kind of leaned on the personality side of it. They're like, yes, it's it's better at everything, but it's also better at ⁓ making decisions and having judgment. One thing it said was like
Ian Landsman (02:06)
What do you got? Mm-hmm.
Aaron (02:34)
It won't ⁓ you know, things that it would have let slide in the past, it won't let slide anymore, which sounds very similar to, you know, the behavior of GPT-5.5, you know, that German engineering aesthetic, which I love. So if Opus 4.8 is that way, ⁓ I think I'll be happy. The other thing ⁓ that I think you called out was fast mode got a lot cheaper. So fast mode used to be like insanely expensive.
Ian Landsman (02:44)
Mm.
Mm.
Mm-hmm.
It was like six
X or something. I don't know, it was a yeah.
Aaron (03:02)
Six X and now
I think it's well it's like three times cheaper than it was before. ⁓
Ian Landsman (03:07)
I guess it's right. So maybe it's
two or two and a half X or something like that, yeah.
Aaron (03:11)
So
I don't know if this I don't know if this is thanks to, you know, Elon's GPUs or they're just it's just marketplace pressure. ⁓ but I use GPT five in extra high fast all the time. And so I know. Feels good, man. Feels good. I have every every interest in fast mode. You always tell me this is the land grab era, and so I'm trying to grab land by going as fast as freaking possible.
Ian Landsman (03:20)
⁓
Why? I can't believe people use fast mode. I d I have no interest in fast mode. No interest at all. ⁓ yeah. It is, that's true. Okay. I guess so.
Yeah. All right. Maybe I should try it. All right.
Aaron (03:41)
⁓ you should. So
I think it's interesting to me that fast mode is so much cheaper. I wonder if they heard complaints about that or or what. ⁓ I kind of think it's just market pressure ⁓ to make everything cheaper.
Ian Landsman (03:52)
Yeah. What's what's open
AI on fast mode?
Aaron (03:57)
I think it just consumes two X usage, I think. ⁓ yeah.
Ian Landsman (04:00)
It's two X ish. Okay. So they're probably
coming, presumably they're coming roughly in line with whatever OpenAI is doing. Yeah.
Aaron (04:08)
Yeah, presumably.
⁓ so there was one benchmark that Opus did not beat GPT five five on. I'm wary of most of these benchmarks, to be honest with you, especially the ones published by the people who just published the model. That always is like, ⁓ I'm sure. but there was one I think it was like Terminal Bench where it didn't beat ⁓ GPT five five. But I will be curious, ⁓
Ian Landsman (04:16)
they're all garbage. I don't think any of them are real. Yeah. Sorry. Yeah.
Aaron (04:33)
I'm sure it's a better coder. What I'm most curious about is this judgment thing that they keep calling out, that it like makes better decisions. And Tariq or Tariq, the you know, the dev rel guy there was like it just feels warm to interact with, which is the thing I've actually always loved about Claude is the feeling of interacting with it. GPT is ice cold.
Ian Landsman (04:38)
Mm-hmm.
Always the best part about Claude, yes. Best feeling, yes. it's terrible. I hate
it. Hate it.
Aaron (04:59)
So
that'll be the thing that I'm looking out for is how does it feel? And if it writes good enough code and feels ⁓ warm, then I think I'll be pretty happy.
Ian Landsman (05:02)
Yeah.
The other big thing in four point itself is ⁓ also apparently less like kind of checking in on it, less asking you questions, just superfluous questions. Obviously there's times you want to ask you questions, that's of a fine line for them to figure out, but seems like they've made some progress there in letting you kick off something and it runs for an hour and you don't come back and it's got a bunch of questions and only ran for two minutes and did did nothing.
Aaron (05:33)
The last thing on this
is I'm curious how long they're gonna let this four series go on. I feel like they're really setting themselves up for five has got to be pretty big. Because if they keep pushing like, well, four point eight is like whatever, so much better than four point seven. Are you are we gonna end up in a macOS whatever 10 thing where it's like, you know, it just stays forever? When are you gonna release five and how much better is it going to be? ⁓ I don't know if
Ian Landsman (05:38)
Yeah.
Right.
Right.
Aaron (05:59)
that ⁓ you know, if that portends something magical or if they're just like F it, this is the same one l letter rip.
Ian Landsman (06:06)
Multiple people said it's five. It should have been called five. That's how good they think it is. I saw multiple people with early access say that. So I guess we'll see. But yeah, the naming ⁓ is always interesting on these things. Why I mean we've always had it, we've all had this trouble with naming. Is it a point release? Is it a major version? We don't know. They're making it up just like we are. Yes.
Aaron (06:24)
Vibes only. In the new
Claude, we've got dynamic workflows. And interestingly, all right. So let's talk, let's talk briefly about what dynamic workflows are. ⁓ it allows Claude itself to spin up, and I'm quoting tens to hundreds of agents. Hundreds. It's almost as if they want you to use your tokens. Huh. It's almost almost as if they're incentivized.
Ian Landsman (06:29)
This one's cool.
Hundreds, hundreds of agents. Yes.
They want you to use your tokens. Yeah. Mm-hmm.
Aaron (06:53)
But the thing that stood out to me in the announcement was it specifically called out like, this gives you the ability to ⁓ port ⁓ libraries, frameworks, tools from one language to another, spanning many thousands of files. And I was like, where have I heard this before? And I went into the announcement and read a little deeper, and it was like, Jared Sumner, author of Bun, used dynamic workflows to port Bun from Zig to Rust. So what are we thinking about these dynamic workflows?
Ian Landsman (07:07)
Yeah. That sounds familiar.
I'm super excited about this. This to me is like, you know, it's not something I'm gonna reach for very often. ⁓ I don't want to spend my tokens at that speed, as we just talked about necessarily most of the time. But for certain types of tasks, which is seem specifically designed for, I'm hopeful that they come out and we see lots of people use it and they say, my gosh, it did a great job. Because I have my own needs for this, for things in my other product help spot where like we're doing a front end redesign at some point soon. It would be wonderful if it could get us, like, you know.
Okay, we spend three or four five thousand dollars in tokens, but boom, you know, we have a new version that's 85% done. And then we can tweak it from there. Just give it to me. Yeah, but if it saves you a month of work, right? That's all crappy work too, all those things, right? Like change your framework over, change your front end over. Like this is all you're sending some poor schlob in there for two months.
Aaron (07:59)
You are the ideal customer because you just said we spend three or four or five thousand dollars in tokens. That's insane.
Yeah.
Ian Landsman (08:18)
to go and look at every file and convert it. It's a terrible job. Even with just regular AI, like if you're doing it one at a time or small chunks, also a terrible job. The if I could just point it at this thing, you know, probably with a pretty robust spec and it goes off and does it, man, that's worth thousands of dollars for sure.
Aaron (08:35)
It it feels
like everyone, all the labs are moving up in the orchestration layers, right? So historically they had the model, then they add the harness, and now the harnesses are getting more and more powers to decide, let's spin up a hundred agents or let's use the computer or whatever. And so it does feel like a little bit of an arms race in the harness layer. Obviously, the models continue to get better, but I saw somebody recently say that the
Ian Landsman (08:41)
Yeah.
Aaron (09:04)
It's not just about the model anymore. It it is about the harness, which I think is is interesting for these third party harnesses, like for example, open code or whatever. Cause open code or Pi or AMP, they're not gonna get this dynamic workflow thing straight out of the box because they're just calling the model via API. And so it feels like, ⁓ you know, it feels like
These first party harnesses are becoming a new competitive frontier.
Ian Landsman (09:35)
Yeah. I d I I just think long term it's hard to imagine them leaving much air for anybody other than like the niche cases. Like it's hard to imagine like there's just gonna be some big harness that's a third party that does all the stuff that the prime you know the first party harnesses do. Like they're just not gonna let leave have those APIs, they're just not gonna allow that. Like, why would they allow that? So yeah.
Aaron (09:45)
Yeah.
You almost called primes like their defense contractors
Ian Landsman (10:04)
It's kind of like that. Yeah. So I mean, even you get
Aaron (10:04)
Which you know is directionally correct. Yeah. I I saw
Ian Landsman (10:08)
into that stuff, like security wise, like if we do have models that people feel like are too powerful, or we do have models that are regulated at some point, like that's just not gonna be something that's just open to the third party harness. You're gonna have to like authenticate in some way yourself and verify and they're gonna wanna be able to secure that on their end, since they're gonna be responsible for it if it's like a government regulated type thing.
But who knows if that's gonna happen, but
Aaron (10:33)
I saw
one of the open AI people tweeted recently that I think like five percent of their traffic comes from Pi, which is a third party harness. And I think maybe I don't know if it was five comes from Pi plus open code or five comes from Pi and five comes from open code. But open AI still seems very amenable to having these third-party harnesses. And it seems like obviously we've talked about Claude cutting everybody off in
Ian Landsman (10:42)
Mm-hmm.
Mm-hmm.
Aaron (10:59)
the next month or whatever. ⁓ this is this feels like part of the reason why. Like they continue to iterate on their harness layer and they want ⁓ everyone else to just pay market rates. So we'll see.
Ian Landsman (11:11)
Yeah. I mean then
they wanna be able to provide the best pro it's like Apple, right? They are the Apple mindset here of like we're not gonna let you do a bunch of stuff in the third party 'cause we wanna provide the best process. Now we'll see if they can do that or if people leave.
Aaron (11:18)
They are.
They want to own
the experience and they're trying to make the experience better. Okay, so Cloud Code, performance improvements, and reliability updates announced. They might have gotten out over their slop skis a little bit because Cloud Code has had a lot of flickering issues with their with their TUI, but it seems like they have ⁓ done a round of fixes. So give us a quick hit on what these ⁓ Cloud Code fixes are.
Ian Landsman (11:29)
Yes.
Yeah, for those of us who are in Claude Code every day, they seem good. They fixed ⁓ rendering issues, flickers, that kind of thing. So ⁓ supposedly. ⁓ improve responsiveness, ⁓ fewer mysterious errors, which I think is nice. Sometimes you get those like tool result doesn't match tool use type things, which you don't have any idea what to do with that. ⁓ compaction progress visibility, which I don't ever see compaction, I always clear out, but if you still are using compaction, this will be better. That was always the thing that used to annoy me about compaction.
⁓ it wouldn't let you compact. Like it's like too full and you cannot compact, which is crazy. Yeah. So it seems like they fixed that, which if you're using compaction, that's good. Improvements to MCP, self-healing sessions, which seems pretty cool. And easier feedback submission. This is another cool one where you used to have to find like the exact session or like thread essentially and to send them the feedback on that. But now you can just send feedback for like a day or a week at a time. So you can just dump it all back to them.
Aaron (12:21)
Yep. What's the point of compaction?
Yeah.
Ian Landsman (12:47)
And I guess they'll use AI to figure it out.
Aaron (12:47)
The one that caught my eye
⁓ that has frustrated me to no end in the past is you take a screenshot of your desktop and you're on a retina desktop. And so the screenshot ends up just ginormous, and you paste it in, and then it's your host. It's over. Like it's like, this image is too big, back up. And you try to back up, and it's like, the image is too big. And you're like, I know, I'm trying to back up. ⁓ so they claim that under the self-healing sessions. So
Ian Landsman (12:52)
Mm.
Yeah.
Just locks out.
Right.
Aaron (13:14)
Maybe with the new Claude Tui and the new Opus 4.8, we'll give that a shot. For now, I am still on GPT 5.5. But speaking of the primes, we've got some mythos news. So, first on Mythos, I have a follow up from a dear listener. We talked about mythos on a previous episode, and I got a follow up saying re mythos. It is not free.
Ian Landsman (13:14)
Mm.
Hanging in.
Mm.
Aaron (13:41)
They are charging insane rates. I don't know about the cost per token, but our security group won't rescan once fixes are in place due to having blown the entire year's budget. It finds a lot of false positives. Our security group has to trim it down a lot by performing ⁓ POC exploits to try to only pass on true positives. So that's a little bit of a follow-up from someone who has actually been able to use Mythos. Apparently it's terribly expensive. but what's the news item here?
Ian Landsman (13:48)
Mm.
That's a problem.
Yeah, I mean, apparently they're preparing for launch. ⁓ you know, it's leaked out on even some of the third party platforms like Google and AWS's version of of ⁓ Anthropics models. And so apparently we're gonna get it. Also, we're gonna get like a security more like I think OpenAI has something like this I've used a little bit, although it was a little clunky. but there's gonna be like a security zone where you can scan your repos. ⁓ hoping it
Not just a clone of OpenAIs, because I don't really like how OpenAI has done theirs, which is like based on your PRs, ⁓ as opposed to like or on your commits, as opposed to like just looking through the code base. Cause like if you have a really old product like mine, like there's stuff in that's not in a recent commit. So anyway, ⁓ hopefully it's coming. Yeah. I mean, the cost I figured that would be kind of crazy. So that's gonna be exciting. ⁓ the false positive feedback there is unfortunate, right? Because like,
Aaron (14:50)
Right.
Ian Landsman (15:05)
That to me is like one of the key things. Like you could just send the AI in there, they'll find a million things. Like you just send it to go wander around. Like you will literally have 500 reports, but like finding accurate ones is sort of sort of the key aspect of this whole thing. ⁓ so hopefully in the final release version, it will be super accurate and wonderful.
Aaron (15:10)
Sure.
Is this
going to be the security apocalypse that I think it's going to be? When Mythos is Mythos is open to the public and then everything is just a game over. Doors are unlocked. It's we're done. Is that what's gonna happen?
Ian Landsman (15:31)
Yeah, you I'm on board with the security apocalypse.
I'm pretty worried about it. Like, I think so. Like, I mean, just think about this. Let's think about even this, right? Like, okay, you get 500 results back, and a lot of them are false positives. So, like, you're just showing up for work every day, people are working on it, then they're going home. But the motivation of the people trying to hack you is different, right? They're gonna, they're spending all day, all night, like looking for every through every one of these to try to figure out how they're gonna get in. They don't always have access to your code, of course. But like even without that, like they are gonna be using this to really round the clock attempt to hack.
Aaron (15:45)
Yeah.
Yeah.
Ian Landsman (16:13)
everything. So yeah, I don't know. I I I don't know. Hopefully it's not just open open. I think there's a chance it's not open open right away, but we'll see.
Aaron (16:21)
Yeah,
I was doing some reverse engineering of some hardware here with GPT and I had to apply to their cyber program. So Sam Altman has my driver's license now. So I I have to imagine Mythos is gonna have some sort of know your customer type thing. But I mean
Ian Landsman (16:29)
Right. he's gotcha.
Right.
Aaron (16:40)
Mythos isn't even out yet and it's already the security apocalypse. I mean, every single day it's like, by the way, your entire machine is hosed because of Shy Halud or something. So I have I I do not have high hopes for ⁓ the near future of open source ⁓ code because I think everybody's just gonna get totally pwned.
Ian Landsman (16:44)
Yeah.
Yep. Yep.
Yeah, I mean, if you're not running in an isolated environment yet, you you really have to very soon, I think. Like, because it's just it it first phase one was like, well, the AI is gonna prompt inject you, which is still a a problem. But yeah, now with all these hacks on like NPM and stuff, it's like, I just ran NPM update and my whole machine is taken over. Like, there's that's just not a good space. Right. Yeah. And at least like, I mean, that's as terrible as that is, even if you're isolated, but at least like it's not getting at your SSH keys and your.
Aaron (17:21)
I'm now a part of a botnet, yeah.
Ian Landsman (17:30)
passwords and installing a key tracker on your actual machine with actual data. So yeah, I don't know. It's gonna be super ugly. ⁓ I don't have any words of comfort on on the old AI hacking. I think we're kinda toast for a while. I don't know.
Aaron (17:44)
Hmm. Dark dark ending. ⁓ we it's over, my friends. Pack it all up. ⁓ yeah, I'll be
I'll be curious to see once Mythos gets out there how they gate control ⁓ and how much ⁓ how much of the hype lives up to or how much of the reality lives up to the hype. Because again, they've really set themselves up here. This is the same, like this is the same 4.0 series that we're talking about. Five must be huge. They've
Ian Landsman (18:08)
Yeah.
Aaron (18:13)
Really hyped up Mythos, and so if it doesn't deliver, I think it's gonna be egg on their face. But ⁓ I guess we'll find out. so anything else on Mythos?
Ian Landsman (18:24)
⁓ nothing else right now. I just hope they allow it to to the small guys, you know. Like I think the a lot of the articles implied it's staying in the enterprise plans, so hopefully that's not forever. But we'll see. Yeah.
Aaron (18:33)
Yeah. Yeah, y no shot. No shot.
Although if you show up with, you know, fifty thousand dollars, they might give you a shot. How are you working this week, Mr. Landsman?
Ian Landsman (18:40)
Right.
⁓
I don't have any changes really this week. I'm still Clawed. I'm ⁓ I do I added a few more custom skills to my ⁓ toolbox to expand from just I'm a little bit of a skills guy for my codex cross checking, which is phase one was just like see if there's bugs in this code that was produced. I broke out an architecture one.
Aaron (18:52)
Becoming a skills guy.
Ian Landsman (19:05)
As we think we talked a little bit about last week and I broke out a like sort of polish one. you know, I don't know how much I'm gonna use the polished one. But anyway, now these are like three different things I can use separately. ⁓ definitely the bug fix one. I'll I use often the architecture one for certain things. I'll probably pop that in there and see. But yeah, otherwise, I'm still cruising along. I'm excited to try four point eight. I love my superpowers. I don't do anything without superpowers. I use superpowers for every single thing I do. So good. I love the superpowers. Yeah.
Aaron (19:24)
You still hot on superpowers?
Crazy. It's crazy.
Ian Landsman (19:34)
You gotta try the superpowers. You gotta.
Aaron (19:34)
I have got I haven't tried the superpowers. I'll tell you what. I've gone the other way. I can't remember the last time I used a skill. I've got I've just I've just gone the other way. I will say I have like a React Best Practices one loaded that GPT picks up whenever I need something ⁓ somewhat complicated or even mildly complicated touched on the front end. It picks up the React.
Ian Landsman (19:43)
Yeah. I mean I don't use many skills either, I have to say, but yeah.
Mm-hmm.
Aaron (19:59)
but I don't do I don't do manual invocation of skills very often. ⁓ part of that is because I want to exercise the vanilla paths of solo, solo term.com, it's very good. ⁓ but part of it is like I just talk to the agent. That's my new, that's my new catchphrase. Just talk to the agent. You got a question? Just talk to the agent. How's it work? You just talk to the agent. ⁓ and so I do a just a lot of
Ian Landsman (20:12)
Mm-hmm.
I mean that's pretty much what I do too. No.
Aaron (20:26)
planning, not not even in plan mode anymore. Just in like, hey, GPT, let's talk about this feature. Ask me questions. Okay, well I think this. What do you think? Just talk to the agent. So my new my new catchphrase is just talk to the agent. You heard it here first. not not a big skills guy right now.
Ian Landsman (20:46)
I mean me either. I I hard the only skill the only other skills besides superpower are these codex ones. And those are just to like save the typing. They're literally just to save the typing of like I can tell it spin up solo MCP codexes and cross-check about like I so I don't have to write that every time is like the only thing I use it for. And then superpowers, which is basically just doing what you just said, which is like it's going to ask me a bunch of questions. ⁓ yeah. So that's that's primarily what superpowers does. But
Aaron (21:07)
Mm-hmm.
Ian Landsman (21:13)
But so I don't know. I like the superpowers. I'm still using it, but it's pretty much just prompting. I don't have any other the only other skills at all is like larval boost. If you're a larval person, you might be using boost. so it will occasionally on its own pull those in. I never tell it to, but ⁓ that seems fine. Yeah.
Aaron (21:28)
I will say what you just
said about ⁓ the like orchestration skill, my mindset has changed a little bit on skills, away from like I want them to be auto-invoked, to just these are prompt templates. They're just saved prompts. That is how I think about skills right now. And I'm building saved prompts into solo, not skills just like, here's my library of prompts. Let me invoke it with you know a hotkey. Because that, like that that you just described.
Ian Landsman (21:33)
Mm-hmm. Mm-hmm.
Right. That's what I do.
Mm-hmm. Hmm. Okay.
Aaron (21:57)
is very helpful, saving all the typing. But the whole I'm gonna load up these, you know, copywriting UI, beautify, I'm gonna load all these skills up and hope that it picks it up at some point. I I think I'm I'm just past that. I don't think that's super viable.
Ian Landsman (21:59)
Right. Yeah.
Yeah.
They also just seem it's I think it's a very one of these very temporary type things, right? Like they're optimizing these models for these development tasks. We're all doing the same development tasks, and like it's already gotten better. In a year from now, it's gonna be a better designer, it's gonna be a better architect. Like you're not gonna need a custom skill to teach it like basic stuff. So it's really just gonna be like your super custom stuff and to save you time. ⁓ and that's pretty much it. Like, and so yeah, so prompts and solo, that will be interesting.
Aaron (22:18)
Yes.
Prompts and solo.
Solochurm.com. It's very good. Do we have producer Dave on the horn for any questions from the live audience? We are live at two p.m. Eastern, one PM Central every week. Producer Dave, what do we got?
Ian Landsman (22:42)
There you go, you heard it here first. Yeah.
Producer Dave (22:54)
All right, we do have a couple questions here. we're gonna start with ⁓ you know, a very nice acknowledgement from from Jetpack Joe here, where he says he acknowledges he's been overly negative towards Aaron in the past and he will strive to be more positive going forward. It's just I wanted to start QA on a on a really nice note.
Aaron (23:08)
I appreciate that.
Ian Landsman (23:09)
There we go. Just a
Aaron (23:11)
You know, I
Ian Landsman (23:12)
beautiful
Aaron (23:12)
I appreciate I appreciate that, Joe.
Ian Landsman (23:12)
place. Yeah. That's great.
Producer Dave (23:15)
Yeah. All right. Let's
get to our first question. ⁓ from LaCroix, Steven, which is a a flavor of sparkling water I'm not familiar with. Since frameworks like Laravel have public source code, does that make them riskier to use or actually let me get it back up, or actually more secure because they receive regular security updates compared to a custom no framework?
Ian Landsman (23:35)
I mean, my take on this right now is the framework. I'm down with the open source frameworks because I think they're big enough that like they're gonna get like we just saw Symphony got early access to Mythos, right? Like, so they're gonna get some lead time. I am very down on individual open source packages because they do not have the pull to get that early access and they're just leaving. You just have these packages that are unmaintained often and things of that nature. So, and even maintained ones again will be behind the curve. So yeah, I'm a little bit down on those.
Producer Dave (23:36)
Yeah.
Ian Landsman (24:05)
And I'm up on frameworks. What do you think, Aaron? Yeah. You don't like it.
Aaron (24:06)
I hate to agree with Ian. You know how much I hate to agree with Ian.
use a framework. Use Ruby on Rails, use Laravel, use Django, use whatever Cloudflare's fork of Next.js is. Use one of these big frameworks backed by either a big company or a big community, ⁓ because they those are going to be battle-hardened. Do not roll your own framework, roll your own business logic. 100% it's better to use a framework.
Ian Landsman (24:35)
Yeah.
Producer Dave (24:36)
I'm just gonna clip that for the next ⁓ larval.com. Thank you so much. All right, question. do you feel like you think less coding-wise as you use AI more?
Ian Landsman (24:39)
Nice. There we go.
Aaron (24:39)
Perfect. You got it.
I got this one. I got this one. No, I'm thinking so much more. ⁓ I'm typing way less. I'll give you that. I'm typing way, way less. ⁓ especially with voice to text, I'm barely typing at all. but I think, I think so much more. And I think at such a ⁓ higher on the ladder of abstraction, I'm thinking much, much higher. And so I'm always thinking about.
Ian Landsman (24:47)
There you go.
Aaron (25:10)
I have this like mental memory palace in my head of how this are this application is architected. And I'm talking to Codex, like, what if we pulled this part up and then we made this an adapter and then we had this as an ingress point and then that solidifies and unifies all of this stuff so such that that's shared and codex is like, That's a great idea, you're a genius. And so I am thinking so much and so hard, and I'm thinking more than I ever have before, and I'm typing literally zero code, and I could not be happier.
Ian Landsman (25:40)
I mean, I think that pretty much says it. I think if they meant by code-wise, like in terms of actual code, I definitely think way less than before in terms of like what is the literal function name of this thing and all that stuff. But I also agree with Aaron that like on higher, it's I this is the place I've always preferred to think. for all 25 years I've been a software developer, I like to be up in the product. What's the product doing? How's the product doing it? So I this is my glory era of like I get to just think up there at the high level.
Producer Dave (25:50)
Right.
Aaron (25:57)
Mm-hmm.
Ian Landsman (26:07)
And then let the minions do the kind of ones and zeros.
Aaron (26:11)
So I'm I'm I'm even one step below the product. I love thinking about the products, but I love thinking about the architecture. That's the thing that I love thinking about. I've always wanted to have infinite time to ⁓ just create beautiful architecture for no reason. And now I have infinite time. It's great. I can, I can say, let's, let's refactor everything to be cleaner. And it's like, yes, sir, right away, sir. So I I am loving it. I'm having a blast.
Ian Landsman (26:15)
Yeah. Architecture there, yeah. Mm. Mm.
Right. Do you? I don't know.
Mm-hmm.
Producer Dave (26:37)
You you
you just need white whiteboard, some sort of, you know, vision recognition for the agent. As you whiteboard it it's cooking. You could just Yeah.
Aaron (26:44)
I know.
Listen, I got a chalkboard and cameras out there. We could get that set up today.
Producer Dave (26:49)
Aaron, an important question for you. ETA, solo for Windows, what's the story?
Aaron (26:49)
Alright, we got any more?
Solo, SoloTerm.com, it's very good. For Windows, ⁓ I have a tab open right now that is a GitHub Actions tab that says release V point ⁓ zero point eight point zero has built and passed, and it includes Windows. So it is imminent. Solo for Windows, check it out soon. Yeah.
Producer Dave (27:10)
Hey
Ian Landsman (27:10)
Go.
Gotta set aside some time for the ⁓ the
bug fix round there.
Producer Dave (27:17)
All right. Here's an interesting one. With models getting better, do you think a new sort of developer's comp package will include tokens? Or could you see that as a reason maybe to shop around when job with when you're job hunting? One job, sure, yeah, we're paying this much, but you just get unlimited whatever spend or, you know, however that might work.
Ian Landsman (27:29)
Mm-hmm.
I don't know. I I I feel like I mean this is like a thing you see a lot, right? Where they're like, I'm gonna move jobs because they offer me more tokens. Like I don't think that's how people think about their jobs really. I mean, I guess you might have a paw if you're into the using AI and you somewhere's like, We don't use AI, like you might that might give you pause in taking a job there. ⁓
Producer Dave (27:36)
Yeah.
Ian Landsman (27:57)
But yeah, and I just think that that's not how any of this is gonna work. Like in the end, like the companies are gonna have X resources and maybe they're on you to not use too many resources at times or whatever. But like I don't see you taking a job or not and being advertised as you get a billion tokens a month or whatever. I don't know. What do you think, Aaron? Yeah.
Aaron (28:14)
Nope, I don't think so. ⁓ I think
companies are cutting back on their token budgets because they're just getting absolutely hammered. So you take a job that gives you a bunch of tokens and then it's like, by the way, we're cutting your token budget by 80% because we're going broke. ⁓ it's also like, hey, if you join this company, we'll give you two thousand dollars a month to spend on servers for this company. And you're like, I don't care. I want money in my pocket. I don't I don't care if I can spend your money on servers. I want to get rich. ⁓
Ian Landsman (28:38)
Right. Right. Yeah. Yeah.
Aaron (28:44)
So no, I I don't think so. If a company said you're not allowed to use AI, psh, that's a hard pass for me, but yeah.
Ian Landsman (28:48)
Right. That's a different thing.
I just don't think that token like your token usage even works that way in reality. Like there's very few companies that are like, we want to make sure you use this many tokens a month. It's, you know, at the end of the day, it's like we have features we want to build, we have produ bugs we need to fix. Like that's what we're focused on. And like some months you're gonna use a lot of tokens, and some months you're gonna lose less tokens. Like there's very few companies out there, despite what Twitter will tell you, that are just like, use all the tokens, just spin up a million workers, do random shit, like see what happens. Like that's not nobody's doing that in the real world.
Producer Dave (29:18)
All right, we got time for one more.
Aaron (29:19)
I don't think
so. sure, we'll do one more.
Ian Landsman (29:21)
Yeah.
One more.
Producer Dave (29:22)
All right.
If money's not a problem, what model or harness, Aaron, would you recommend as an orchestrator and which one's for a sub agent using solo?
Aaron (29:30)
if money is not a problem, ⁓ I would probably still use GPT five five extra high fast for my for my orchestrator. If money was not a problem, I would use AMP way, way, way more for implementation. ⁓ I think GPT five is very good at tool calling and orchestration and keeping everybody on the rails. AMP has been undeniable at how good it is at
One, reviewing other other agents' code, but two, implementing things well in the first place. But AMP is ⁓ pay as you go. ⁓ they now have included, I haven't turned it on yet, but I think they have experimental ability to attach your GPT sub to it. ⁓ but it doesn't always use GPT. They they decide what model it uses. So if I were a rich man, one, I would not be on this podcast, but two, I would probably use AMP all the time.
Producer Dave (30:23)
Whoa, whoa, whoa.
Ian Landsman (30:24)
Boo.
Aaron (30:26)
Guys, I'm gonna walk into the woods never to be seen again.
Producer Dave (30:30)
Leaving Token you're gonna leave Tokentown.
Ian Landsman (30:30)
Come on, you can't give the
No, he's not leaving Tokentown. He's never leaving Tokentown. I don't care if you had a hundred million dollars. You're not leaving Tokentown ever.
Aaron (30:33)
I'm gonna leave I'm gonna leave Token Town. When the day that I don't show up
If the day I don't show up, it's because I'm in my underground garage tinkering with th three D printers and hobby electronics, okay?
Ian Landsman (30:48)
Like the end of the town where you're just off in the swamps of Florida. Say a movie you know. Let's go. All right.
Aaron (30:50)
⁓ great movie. All right. We got to keep going. I've seen that movie. We got to keep going. Thanks, Producer Dave.
Thanks, chat. All right. Speaking of burn, baby, burn token budgets, Claude Code adds a usage command to track token consumption by component. My guess is that they are continually inundated with people complaining that their context is used up, their limits are running out.
Ian Landsman (31:14)
That seems right. Yeah.
Aaron (31:16)
And so now
they've surfaced some visibility. And so you can see a breakdown. This is from Boris Cherney, who is the guy. And this is it shows you a breakdown of skills, sub agents, MCP servers, and what is using up your context. So why do you think we did this, Ian?
Ian Landsman (31:34)
Yeah, I mean, I think that everybody's complaining about this all day, every day, right? And like this way, there's just more evidence for them to be like, hey, this crazy skill you wrote that does all kinds of stupid stuff, like it's using thirty percent of your tokens. Superpower skills using all the tokens, just so you know. ⁓ so yeah, I'm sure it's like literally a thousand complaints a minute about token usage, right? So I think this will be useful. I mean, I've like as we talked about before, like when I switched to the team claude plan, like I definitely feel like the
Aaron (31:42)
Mm-hmm. This superpower skill.
Ian Landsman (32:01)
Tokens are a little bit weird. So I think it'll be cool to be able to really at least really see where it thinks they're all going and have that feedback. Maybe there's stuff running you don't even know is running. Maybe it's calling skills, you don't even know it's calling, right? Cause it doesn't always show you everything there. Or and I'm certainly not reading everything. So yeah, I think I think it's a good good little tool. Good enhancement.
Aaron (32:10)
Yes.
Yeah, I think this'll I
think I think this is ⁓ useful for the users, but also useful for PR for ⁓ anthropic. And I think at some point, I don't know when, I have to imagine one of these scaling laws applies. And at some point we're not gonna be thinking about this token usage stuff. Maybe that's optimistic, but I think maybe in a couple of years it's just gonna be you just use as much as you want and you know, intelligence is too cheap to meter, but you know, a guy a guy can dream.
Ian Landsman (32:26)
Mm-hmm.
Aaron (32:49)
I don't have much else on that one. Do you have anything on that?
Ian Landsman (32:52)
No, I think yeah. I think it's good good addition.
Aaron (32:54)
All right, hit us with some United States AI adoption shows steady growth. What do you think?
Ian Landsman (33:01)
So, what was cool about this, so you know, not unexpected that adoption is growing of AI in the United States, but they have this kind of chart inside the article, and we'll link to this, ⁓ where Microsoft has done this research and AI usage in metropolitan areas is at 33%, at micropolitan areas, which I'd never heard of that before, ⁓ 22%. And rural is 16%. So, just like another interesting little data point, like I feel like
Aaron (33:21)
Not a thing. Yep.
Ian Landsman (33:31)
No, our world's in a weird spot. And here we go again with like certain groups embracing the technology, other groups maybe not embracing the technology. What divides does that create? ⁓ there's no reason for this divide to exist. Like, there's not a reason why people in rural areas have 20 bucks for Chat GPT. I it's free, there's a free version, right? Like, so it's not a cost thing for normal people. Like it's either free or very cheap for everybody right now. So there's something else here. Yeah.
Aaron (33:36)
Mm.
And you know, you know what's interesting about this is ⁓
it's almost double in metro counties. So it's 16% in rural areas. And you know where they're putting data centers? In rural areas. Yep. And so I think that is I think that's some of the ⁓ sentiment split is they're putting up these ⁓ these megaliths in these areas where people on the majority don't use it and probably don't like it.
Ian Landsman (34:09)
That is true. Yeah, so maybe there's something there.
Aaron (34:25)
So maybe we're in for a rock rocky couple of years.
Ian Landsman (34:25)
Yeah. So that's
Hopefully ⁓ Ben Thompson has this idea where they should just pay off people in the towns that get data centers. And I totally agree with that. I think it's that's a great PR move. And you're gonna have companies out there like SpaceX that are trying to inflame this in a negative way because they want you to send your data center to space, right? So, like there's gonna be a lot of pressure. No, that's a great take. That's a perfect take. That that's a perfect, of course they are.
Aaron (34:36)
A hundred percent.
Yes.
Nonsense. Bad take. Bad take. Boo. That's a bad take. That's terrible. They're not fomenting insurrection so that we'll put
nooo.
Ian Landsman (34:58)
Well it's not insurrection.
They're forming people. You don't want this in your town. Use all your water. Blah, right? Like, of course they are. That's what's happening. Space impact. what's the motivation? Why do they want more terrestrial data centers? They don't. They're building space data centers, Aaron. That's what the whole they just had they just had a whole IPO thing released that was full of ridiculous stuff. And one of them was everything's gonna be in space. Yes.
Aaron (35:04)
This is not based in fact. Dear listener, strike this from the record. This is not based in fact. This is nothing.
Yeah, they want they want space data centers, but they're not fomenting an uprising.
No, the dear listener
needs to understand that one time Elon stole Ian's girlfriend and he's been mad ever since. And so he hates Elon. So don't listen to that. Don't
Ian Landsman (35:32)
⁓ man. You'll see. You'll see. But note
note. Token town episode three, Ian called it. We will check back six months from now.
Aaron (35:43)
Don't don't listen to that. The SpaceX
IPO mentioned AI more than Jesus, but you know who loves Jesus? That is the Pope. And the Pope. The Pope has written a document. He has written an encyclical, which is quite long. And I have not read it, but I've read many takes on it. And I've read ⁓ yeah, I've read some. So my Twitter feed, I'll just have you know.
Ian Landsman (35:48)
That's true. He does.
Mm.
Yeah.
Yeah. Me either. Let's let's do some takes.
Mm.
Aaron (36:08)
Is ⁓ full of Bible-believing Christians commenting on the Pope's takes. I also see ⁓ people in actual Token Town who are commenting on the Pope's takes. And so I'm seeing a lot of I'm seeing a lot of interesting takes. But this article, ⁓ this or this topic that you ⁓ added says that the anthropic co-founder, Chris Olah, remarks on Pope Leo the 14th encyclical.
Ian Landsman (36:13)
Okay.
Mm.
Aaron (36:33)
Magnificata mag magnifica humanitas. ⁓ so forgive my Latin. Wha that was pretty good. What what do we think what do we think about what ⁓ the Pope has to say here? I've got a few takes, but I'll let you I'll let you warm us up.
Ian Landsman (36:38)
Come on, work on your Latin. Let's go. Yeah.
Have
a heavy take. I think it's cool. The Pope is like this Pope seems into like putting his mark on what's actually happening right now and not being kind of focused on like just all the same stuff that's been people been fighting about for thousands of years. So I think it's cool. Yeah, American Pope. So I think that's cool. but I don't really have a hot take other than that. So what's what's your take?
Aaron (36:59)
Hell yeah, American Pope. Yeah.
All right, listen, the Pope is not my guy. I respect the Pope. I am not a Catholic. So ⁓ I am I I am a I'm another flavor. I protested in in my roots somewhere against that. so the Pope wrote this document that broadly compares the ⁓ the AI revolution to two stories in the Bible. One is the construction of the Tower of Babel.
Ian Landsman (37:10)
Right. Try. Right. You're the other flavor.
Mm.
Aaron (37:35)
Which we all know, hopefully, ends in the scattering of the people and God intervening according to the biblical tradition, God intervening and scattering the people out because they were trying to become God through this technology, right? They were trying to build a tower to reach the sky and be like God. That's the one side. Go to space. Gotta bring it back to Elon. Good point. The other the other construction story in the Bible that he compares it to is.
Ian Landsman (37:51)
Mmm. Go to space. Putting stuff in space. Yeah. Right. Mmm. Yeah. ⁓
Aaron (38:02)
⁓ is the Nehemiah constructing the wall around Jerusalem. And these two stories stand apart from each other because one is trying to be God and trying to reach God and trying to ⁓ inflate one's ego. And the construction of the wall around around Jerusalem by Nehemiah is bringing the community together, ⁓
Ian Landsman (38:08)
Okay.
Mm.
Aaron (38:24)
To work for a shared purpose and serve God, like protect the city of God. And so that those are the that's the kind of the dichotomy that he lays out. And he's, yeah, it was I found it quite insightful actually. And what he is saying is technology is morally agnostic, but represented by these two stories, the way that we handle it, the way that we build it, ⁓ does have moral like ⁓ ramifications. And he's suggesting that AI.
Ian Landsman (38:29)
Yeah.
Hm, I like this analogy here. Yeah.
Mm-hmm.
Aaron (38:53)
should benefit the people and not replace their ⁓ dignity in work. So the Bible says that work has dignity. It is good to work. ⁓ And he's saying we shouldn't take that away from people. Instead we should use AI to empower the people. And so I haven't read the whole thing, ⁓ but I have read that section and I found that to be pretty compelling and honestly very informed and level-headed. It wasn't like I I didn't I didn't think like my
Ian Landsman (38:58)
Mm.
Yeah, let's go.
Aaron (39:22)
God, he's making us all look bad. I was like, yeah, I'm like, no, no, no, he's not my guy. No, I don't trust him. But no, I I was like, hey, this is pretty good. Way to go. Way to go, Pope America. ⁓ so that was that was my that was my take on it. I think one of the interesting things was the co-founder of ⁓ Anthropic describes that these models are being grown and not engineered. That that gives me the ick a little bit. ⁓ you know.
Ian Landsman (39:23)
Right. You're cringing the whole way.
Yeah.
Did notice that. That's interesting, yes.
Yeah, that was a little
that part stood out to me. I was like, nah, I don't know if I'd go with that analogy there.
Aaron (39:53)
I'm I
I don't know that I'm a hundred percent in disagreement, but I'm about ninety-eight percent in disagreement. I do think there is some ⁓ some weird substrate that we don't fully understand, but I'm I am not like these things are alive or conscious, but I do think there's something here we don't understand, which does not mean it's conscious.
Ian Landsman (40:00)
Mm.
Yeah.
No.
Yeah.
Right. Yes. I it's amazing how anthropic has the ability to like say the absolute worst angle on these things. It's like, how about we how about we say it's like a human brain and we're growing it? Like, no, how about you don't say that? It's just a bunch of servers in a rack somewhere. Like, so I don't know. Yeah. Right. Yeah. Right.
Aaron (40:28)
Incredible. Incredible.
Yep. Yeah. You got the big boss saying there are no more jobs, and then they send this emissary to the Vatican, and he's like, By the way, we're growing these things as if they're
entities. And you're like, Guys, guys, talk to literally anyone. Talk to someone in a micropolitan area. Talk to someone in a rural area. Let's like ground this in a little bit of reasonableness. ⁓ but I do want to read the whole thing. It is incredibly long, ⁓ but I think I I think it
Ian Landsman (40:48)
Yeah. What are we doing here? Yeah.
There you go.
⁓ okay, I didn't realize that.
Aaron (41:06)
I think it would be informative for me because it is my faith tradition and I am very interested in AI. So I would I would be curious to read the whole thing. But I did like I did like the historical biblical analogies that he set up of the two types of construction and the way that one can go really poorly and the way that one can bring the community together for the greater good. I found that to be a pretty ⁓ compelling argument.
Ian Landsman (41:29)
I don't think I'm the guy for this, but there is a world where you have like an AI podcast that's like religion focused that I think would be rather fascinating. Like this little yeah. No, I'm not the guy for that. But it's still it would be I think an interesting podcast. Like there's gonna be a lot of things that are conflicting in these worlds and and aligned possibly, and stuff's going on with it. So wow, okay.
Aaron (41:39)
I hate to agree with you, but you are not the guy for that.
It would be. yeah.
All right, you wanna end? You wanna end on my hottest take of all time? Red,
red hot, flaming take that I I don't think I don't think anyone that shares my faith tradition would agree with me. That which I'm out, I am out to lunch here, okay? I think I think this is crazy. This is crazy. I think the biblical antichrist is real.
Ian Landsman (42:01)
Okay. Let's do it.
⁓ okay. Yeah. Hmm. ⁓ boy. All right.
Okay.
Okay.
Aaron (42:20)
And is
going to be an artificial intelligence.
Ian Landsman (42:23)
⁓ then you gotta start the podcast. End times.
Aaron (42:27)
I actually,
I actually believe that. I think all all everything no, I think everything in the book of Revelation will come to pass, and I think it is not a human entity, rather, it is an artificial intelligence that will give us unlimited abundance and peace for three and a half years, and then it's over, my friends. That's my red hot take. Isn't that crazy?
Ian Landsman (42:30)
Okay. You're you're blaspheming here, I think. I think you're okay. Okay.
Hmm. That is
that is a crazy take, I have to say. I I don't know. I mean, you it's not like a crazy take in that like could AI wreck everything. Seems like a pretty reasonable take that even the creators of the AI have. So ⁓ was this foretold in the Bible sort of interesting? ⁓ I don't know if there's a lot of clues. I I would like to see some more clues. I'd like you to really I would really like you to dig into this and come up with some like actual.
Aaron (43:06)
It's the lawless one. What is more lawless? What is more lawless than an infinite AI?
Ian Landsman (43:17)
⁓ the connections there of like this passage implies tokens or computers or whatever. Give me some of that stuff. But ⁓ I like where you're going with this. This is a hot take. This is how you're gonna get a lot of views. You post this up all on the solo term blog. You're gonna get some link action. You're gonna get people talking about it. yeah, you know, it's like a Fermi paradox kind of thing, too, right? Like, you know, all that stuff.
Aaron (43:28)
It's good.
Mm-hmm.
Uh-huh.
Ian Landsman (43:43)
But
okay, well I like the hot take. We'll be on the lookout for the Antichrist ⁓ on next week's episode.
Aaron (43:47)
Look out for the Antichrist. And
until then, listen weekly to Token Town, where we talk about AI until it turns. Yeah, in the remaining time that we have on this earth before the Lord Jesus Christ returns and vanquishes the AI Antichrist. You can find more religion every week on this podcast. Two 2 p.m. Eastern, 1 p.m. Central, live streamed every week. And you can find it in your RSS podcatcher of choice. Go to tokentown.com.
Ian Landsman (43:51)
Yeah. In the weeks left.
Mm boy Wow ⁓
Aaron (44:17)
To find more, we're also available on YouTube. Thank you to Bento for sponsoring this and every show that we ever do. Check them out at bentonow.com if you need an email service provider built for the AI era until next week. I will see you later, Mr. Landsman.
Ian Landsman (44:35)
Later, everybody. See you next week.
That was a hot one. He's coming in hot.
Aaron (44:38)
There's no chance,
no chance we're ever gonna get to thirty minutes. No chance.
Ian Landsman (44:42)
He's coming in hot.
What did we get out here? Forty ish or so? Not too bad.
Producer Dave (44:46)
The
I'm clipping that so fat like I just the that whole my god.
Aaron (44:49)
That's a good one, right?
Ian Landsman (44:50)
That's a real time.
Real time clipping. Man, interesting.
Aaron (44:52)
That is a sincerely held belief of mine. I truly and like my best friends, probably my pastor, my wife, none of them are like, Erin, that's a really good take. They're all like, dude, you're insane.
Ian Landsman (45:02)
I buy it.
Producer Dave (45:05)
Yeah. ⁓
Ian Landsman (45:07)
Man, I don't know. yeah, I don't I don't know. I I need to I need to brush up on my whole antichrist sort of ⁓ knowledge to be able to have a hot take back. Yeah, I know. I gotta look in the new all right, I'll check out the last book. I mean, I have some New Testament stuff going on too, but I'm not that knowledgeable in it. So I will have to research. I'm probably not gonna do that. But if I do, I'm going right to the last book. Yeah.
Aaron (45:15)
No, you are historically an old testament guy by tradition. You gotta look into the New Testament, even even just the last book. Just the last book is the scary one.
You're not gonna do that, but if you did, it would be interesting.
Ian Landsman (45:35)
Interesting. Maybe I'll ask the AI.
Aaron (45:35)
How about this? Take ⁓ go to 4.8 and say, I want you to review the book of Revelation, specifically the description of the Antichrist as the lawless one who cannot be contra constrained. And just as a thought experiment, could that be a future AI? Not you, 4.8. God forbid, not you, but could that be a future AI that could not be constrained, that could bring us world peace? Huh.
Ian Landsman (45:39)
Hmm. Yeah. Yeah.
Mm. Yeah. N obviously. Yeah.
Aaron (46:01)
Maybe. Sounds interesting.
Ian Landsman (46:01)
I like it. Yeah.
I think we're a long way from it bringing world peace. So if it's gotta do that first, I think we'll be okay.
Aaron (46:12)
I need to brush up on my antichrist knowledge. Yeah.
Ian Landsman (46:14)
⁓ that's true. ⁓
Producer Dave (46:20)
So
Aaron (46:24)
I know.
I know.
Producer Dave (46:25)
I didn't
I didn't see us pivoting towards the the religion and AI ⁓ angle, but you know.
Ian Landsman (46:31)
Next week we're starting
first segment. We're going, we're going Middle East peace. Yeah, we got we got Tower of Babel. I knew about the Tower of Babel. I didn't know about the wall. But ⁓ you know, maybe we start off every every show with a little Aaron's church corner and he just gives us a little preaching.
Aaron (46:32)
Mm-hmm. Yeah, Tower Babel and Nehemiah, that's Old Testament. That's right up your alley, yeah.
That is the only
I mean, I've got plenty of church corner. That is the only raging take that I have that n almost nobody agrees. I'm sure there's a whole pocket of Reddit that agrees with me. I'm not sure we share many other beliefs in similar on Reddit. ⁓ but I bet there's a whole group of people out there that are like, Yeah, obviously.
Ian Landsman (46:50)
Ha ha ha.
⁓
Right. I'm sure. Yeah, yeah. I'm sure. I don't think you invented this one, but yeah, I like it. Yeah. ⁓
You're gonna get a tattoo? I feel like this needs a tattoo.
Producer Dave (47:11)
I
Aaron (47:11)
⁓ should I get it on my head or my hand? Because that is the mark that is the mark of the beast, yeah. So maybe maybe I'll just get like open AI on my forehead just to, you know, get ready. ⁓ I would love a sweet hang with the Pope. Man. He does. Yeah, you guys could talk baseball.
Ian Landsman (47:14)
I don't know. Yes, neck, maybe a neck. Okay. ⁓ sweet hang with the Pope. That's a great idea. How great would that be? I would love it. He even speaks English now. We got an English-speaking guy from Chicago. He could just come on and talk. How great is that?
Yes. Let's do it.
Aaron (47:36)
Let's try it. Anybody have any connections?
Ian Landsman (47:38)
Yeah, if you got a
connection to the Pope. Aaron's useless for this. We need a legit connection to the Pope. Yeah. Yeah.
Producer Dave (47:40)
Yeah.
Aaron (47:44)
I know, I've already claimed he's not my guy, and you're an old testament guy, so we're out
there too. So we need we need a Catholic. Glauber Costa, CEO of Turso, is ⁓ very Catholic, so we'll see if he has an N, you know? We need a Catholic.
Ian Landsman (47:53)
There you go. We need a Catholic guy. Let's
do it. I like it. There we go. This is why we do it live. There's no editing. That's it. It's out there. I love it. all right. I don't know. Anything else? I think we're good. Let's wrap it up. I gotta fix. That's that's what we'll do. ⁓ Mother in law went to school with him. Who?
Aaron (48:01)
Well, that took a turn.
Producer Dave (48:02)
Wow,
wow, really did. ⁓
Aaron (48:05)
That's why we do it live.
Y'all go brush up on your antichrist.
Producer Dave (48:14)
Think that's good?
Hold on, hold on, hold on. We might have do we have a connection here?
Ian Landsman (48:22)
The Pope?
Aaron (48:22)
What's
Producer Dave (48:22)
Pope?
Aaron (48:22)
the PO?
Ian Landsman (48:23)
Well, let's make this happen. There's a there's a podcast called Token Town. They're interested in having an interview with you about your manifesto thing. Let's go.
Aaron (48:24)
USA USA ⁓
There's one Protestant, one culturally Jewish. You wanna come on and you can you can talk about you can talk about Catholicism and baseball in Boston. It's gonna be great.
Ian Landsman (48:36)
Yeah. There you go. ⁓ Yeah.
Producer Dave (48:44)
I think it's from Chicago. Yeah.
Ian Landsman (48:44)
I think he's a wa I think he's a White Sox guy though, not a Cubs guy. Yeah.
Aaron (48:47)
yeah, I don't know. All those places
are the same to me.
Producer Dave (48:50)
So when you go to Laracon US, don't go to Chicago. You want to go to Boston. Just throwing that there. She's got she's got your fli Yeah, yeah, yeah.
Ian Landsman (48:54)
Thank you, Boston.
Aaron (48:54)
Well, Jin Z Abbe's in charge of all that. I just I just show
up at the airport and get on whatever plane she says to get on.
Ian Landsman (48:58)
Just show up.
Are you speaking or not? Hey Kent.
Aaron (49:03)
I am Kent C. Dodds
is here. Kent, we just had religion corner. You missed it. You gotta listen back. I just claimed ⁓ I just claimed with scant evidence. I I claimed that ⁓ I believe the biblical antichrist is going to be ⁓ is going to be a super intelligent AI. So we also did a little bit of Tower Babble, Wall of Nehemiah. ⁓ so you'll have to listen back to that one. Yep.
Ian Landsman (49:07)
Yeah, you missed it.
Yeah.
⁓
We're gonna have church corner every week. Every week.
Producer Dave (49:29)
Update from
Max. Max says hi they'll see what they can do. So I mean
Ian Landsman (49:32)
Let's go, Max Kelly
AI. Let's go.
Aaron (49:36)
Listen, if I know anything about mother in laws, she's gonna think you're insane when you come to her and you're like, Hey, can I get in touch with the Pope? There's a tiny podcast I listen to. Yeah.
Ian Landsman (49:40)
No, she's gonna love this idea. Yeah. No. It's an
up-and-coming podcast. Yeah, yeah.
Aaron (49:49)
Yes, for sure. ⁓ what
was the question? My speaking at Laracon. By the way, if you're just joining, the show's over. ⁓ I am not speaking at Laracon save for my MC duties. You know who is speaking at Laracon? Is Mr. Kent C Dodds. Yeah. We got a lot of we got a lot of big names outside of the Laravel community this year. We've got Thorstenball from AMP, which is the tool that I love. We've got Kitsy from the internet, we've got Kent Cods.
Ian Landsman (49:56)
Mm.
wow.
Producer Dave (50:02)
See Dodds.
Ian Landsman (50:14)
Mm.
Got Fideloper.
Aaron (50:18)
Trying
well, he's inside the Laravel community. I'm trying to think of all the all the big timers that are outside. ⁓
Ian Landsman (50:21)
Yeah, but still You're still listed
as a speaker, you know, I just want to say. False advertising. False advertising.
Aaron (50:27)
it's fine. It's yeah. I will
speak from stage. So technically true is the best kind of true. ⁓ so yes, I am a speaker, although I I don't have a talk to prepare, fortunately.
Ian Landsman (50:31)
Right.
There we go.
Yeah, looks there's quite a crew. The the Kitsy guy, did you say him? He's on there. Yeah, we got a looks like a good crew this year.
Aaron (50:47)
I'm so excited to have Kent there. It's gonna be so fun. We're gonna convert him to to Laravel. ⁓ he's he's just gonna fall in love with the kindness of the community. He's gonna love it. It's right up his alley. It's gonna be perfect.
Ian Landsman (50:52)
Yes, let's do that.
I think you should have
a whole in the back, in the back of the event space, right? You set up a little Laravel baptism thing where you baptize people into Laravel. How freaking great is this? How freaking great an idea is this? No, come on. No.
Aaron (51:08)
⁓
Producer Dave (51:11)
all right, hold on, hold on. People already think that people I I already jokingly
Aaron (51:12)
That is so blasphemous. Yeah, I love it. Yeah.
Producer Dave (51:18)
refer to and this is my employer, I already jokingly refer to my to to Larville sometimes as a bit of a cult. I I don't wanna go all the way to the f
Aaron (51:24)
Mm.
Ian Landsman (51:25)
No, we're going all the way.
Aaron's baptizing people into the church of Laravel.
Aaron (51:30)
Yeah, that I would get a few I would get an email from my pastor if I yeah, I might I might get struck down by the Lord himself. Yeah. He might come down and scatter my language like he did at the Tower of Babel. ⁓ but it's worth it, you know, there are no bad bad ideas in brainstorming. I will think about performing extra biblical baptisms at a conference.
Ian Landsman (51:32)
that. The lightning. The lightning might be a problem. We gotta not stand next to you. Yeah. ⁓ man. Okay. All right.
That's right. Yeah, maybe you could come up with an angle. Yeah. Yeah. I'm
not talking about literal holy water. I'm just saying like it's you know, ceremonial.
Aaron (51:58)
Yeah, no, I get it. So like take this thing that is near and dear to our ⁓ central tenant of our faith and do it as like a bit. Yeah. No, no, okay, yeah, okay. I no, I misunderstood. I I misunderstood. I misunderstood. No, that's a good point.
Ian Landsman (52:00)
Yeah.
Yeah. Sure. Yeah, it's a bit you're doing it for the show. We'll have the cameras going. Yeah, come on. I've seen a lot of very
religious people do bits way worse than this, first of all. This is not even that high up on the list. They claim themselves.
Aaron (52:14)
Boy, we do not claim them. Listen, if you are flying, if you're flying
in on a zipline to the front of the church, I don't claim you. Okay. I've seen a lot of bad bits. I've seen camels brought on stage. I've seen pastors in twelve hundred dollar sneakers. I just want to be clear, those are not my guys.
Ian Landsman (52:21)
Yeah. Hmm. Yeah, there's a lot of bits out there, man.
Producer Dave (52:32)
yeah.
Ian Landsman (52:34)
Okay. All right. You're excommunicated, a lot of people already. I see I think you need to run. You need to have your own check. You have a lot of stronger takes on who's in and who's out. I think you should set up your own little world, right? You can have you're allowed to break off. You guys broke off before. You're gonna break off again. Yeah. Exactly. Exactly. You're gonna this is gonna be the AI t version, right? You're gonna be for the modern era.
Aaron (52:37)
Yes.
We broke off before and and we will do it again. I've got some theses ready to nail to some doors. Yeah.
Producer Dave (52:55)
You're he's gonna be nailing them
to the doors of data centers, is what he's gonna be doing. He's gonna go walk right
Aaron (53:00)
That's right. Yes.
Ian Landsman (53:01)
there we go.
I can't believe your take on
Aaron (53:03)
I'm gonna be nailing Magnifica human humanity to ⁓ humanitis to the front door.
Ian Landsman (53:07)
Your space X take
is so bad. I can't even believe you were you're countering me on this. They're like, we're gonna send all the data centers to space. That just inherently means they don't want them on Earth, right? You want to make it hard for ones on Earth so you can send them to space. If you have lots on Earth, you don't need the ones in space. Do you see the problem? That's very obvious. Yes.
Aaron (53:24)
No, no. They're they're making an economic argument. They're making an economic argument.
It's going to be better in space, not hey, let's let's go and rile up the rural people to no.
Ian Landsman (53:35)
You couldn't be more
wrong. You I you I I can't even believe you don't agree with this take, to be honest with me. They just riled up a bunch of people for a very similar but sillier even purpose, you know. And this one is actual money behind it. What do you say?
Aaron (53:45)
Put put Kent's put Kent's back on screen here. So Kent and I, Kent and
I share ⁓ different faith traditions, but he says, my guy walks up on stage in a suit and tie. I can't imagine my 90-year-old church leaders coming in on a zipline. You know what? I can't imagine any church leader coming in on a zipline. What the hell are we doing? so anyway. You know what? We should probably clip this as like a post-show and put it somewhere. Because this is gold. This is
Ian Landsman (53:55)
Ha ha ha.
Mm.
Like a zip lines for everybody.
Mm. This was this was a little more
than just a post show random stuff. Yeah. All right. A little bonus in the feed. Or it could just go in the pod. Could just go in there.
Aaron (54:15)
This is gold, baby. Yeah. Alright, I gotta run. I gotta two o'clock.
Producer Dave (54:15)
Mm.
Aaron (54:23)
I'm I I'm a I'm a businessman. I gotta call. I know. All right. See ya next week. Bye.
Ian Landsman (54:24)
⁓ man. All right. We'll see you later. Thanks everybody. Later. Thanks, Dave.
Producer Dave (54:27)
All right. Thanks everybody.