Should You Quit Your Job?

Welcome to Syntax. We got a potluck episode for you today. You bring the questions. We bring the answers. We got questions on how does freelancing work in the age of AI? I used to bail based on time, and now I can do that so much faster.

So do I make less money? Should I move to value based billing? Got lots of thoughts on that. The future of CMSs. Right? We have all these websites that are block based, and now they're really complex. How do we tackle that, and how does AI and agents play into managing a CMS? What do job interviews look like nowadays? Should I be sandboxing my agents or just YOLO running them on on my computer? Right? And a really interesting question about, like, I can't find any blog posts that are good anymore. They're all slopped out. They're all clanked out, and I I can't Google it anymore. And we go into this little rant about, like, how trust and community is this huge new currency. So you got a Wes? Syntax.fm.

Click on the potluck button, submit your question, Wes we will answer it on a future episode.

Alright. First question for today is Sarah Chen Rodriguez asks, a long time listener, first time caller. Thank you, Sarah, for calling in. Web development has been paying the bills for about fifteen years for me. I do some freelance client work on the Node, and now that we have AI coding tools, some things just take less time.

I've leaned towards hourly rates. They have been working fine. I guess the core question or thought I have is, who should be benefiting from the increased speed, the dev or the client? I guess there are all sorts of ways to look at this. Devs have some additional Scott, and the client expects things to take less time. The budgets might stay the same overall, but you can expect more for the same money, etcetera, etcetera. Thanks for creating such high quality dev content all these years.

Sarah, I'm going to say this in the complete, most unbiased way possible. This is me being unbiased.

I think the developer should charge the same and get all of the rewards of the.

That's what I think.

Why do I think that? Because I'm the developer.

I mean, I I think, you know, I think if it's, like, taking you less time and effort the end of the day, though, I think value based pricing has always been a good thing. If if you make a website in a week and that website manages to make your your client a lot of money or, you know, it serves as their business foundation, the value is in the creation of it, not the amount of time it took to get there. I do think that this is tough now, because the amount of things that we can do faster has, you know, increased. So I I do I I I wonder about this. I because I haven't done any freelance client work since this AI boom happened,

I don't know, Wes. What do you think? Yeah. I I I thought for the longest time and when I did freelance, it's like like, hourly billing is usually not the way to go because I invested a ton of time in my tooling, in my process, and all of that stuff so that I could be quick, and and I shouldn't be penalized for for being quick. Right? Shouldn't be what? I you shouldn't be penalized for being quick. Right? Scott, come on. Come on. That's that's awful.

But now more than ever, as you can, like, write out your entire setup and a lot of the stuff, like, maybe, like, you get an email from a ESLint, they have feedback, you pipe that into your agent, it submits a pull request, and then you're the human in the loop there. Like, how do you how do you build based on that? Like and if you're spending more tokens than somebody else, you know, like, you're spending thousands of dollars on this type of stuff, it it gets really muddy. So I think now more than ever, it's very important to ideally move to

value based billing. Yeah. Do you put the token spend in the, in the invoice?

I I don't think so. I think that you should you should be charging much more than your actual token spend. Like, it should it should be negligible compared to how much you're you're actually charging the client. Right? If you are simply just, like, again, being a meat proxy and and taking what they want and sticking into the machine, you're not gonna have a job for for much longer because the client can so much more easily just do that themselves. Right. They could just type into the box themselves rather than email you and have them copy paste it. I think the TLDR about this type of stuff is that you just need to get better. What you're able to deliver for the ESLint. Right? Like, a 15 page website with a map integration and and link off to a Shopify store used to be the bread and butter of of a lot of these freelance devs where you could make $510.

Right? And then and the Squarespace came around and and, like, the like, Wix and whatever, and and a lot of that, like like, WordPress warp kinda dried up. Not to say not entirely, but a lot of it did Wes clients realized we could just do a lot of this myself. I have a little bit more flexibility in it. And I think that that's gonna go even further with a lot of this AI stuff is that you are going to be expected to be doing a lot more, not in terms of, like, the amount of work, but, like, you have to get better at your job. You have to get better JS as the stuff that we used to do is now simply a prompt away.

We are we have to move forward into harder stuff. You might have to get more into strategy. You might have to get more into design. You might have to get more into architecting your entire harness so that that these things are like, you can go on vacation and the whole thing will run itself. Right? Yeah. Yeah. Or be able to even get something unique and beautiful out of these agents rather than just another cookie cutter site. Yeah. Exactly. Yeah. So whatever that is, you should be, like, focusing and looking forward as to, like, what what's the next step? As this stuff is now table stakes for absolutely everybody, what is my next step? And and and this has happened forever. Right? Like, you look back of what we were doing fifteen years ago, that stuff is just an NPM install away a couple years ago, and now it's just a a prompt away. So it's it's constantly changing ever more now than ever. But, I would be looking at what the future of this thing looks like and move it to to value based pricing.

Yeah. I remember when you needed a team of engineers to build a chat functionality on your site.

Yes. Oh, yeah.

Or anything real time whatsoever.

Tim Smith says, I heard y'all say on Sedax that job interviews are no longer about coding Node they are about how you think, and more about how you prompt and and how you think about when you're prompted. Right? Like, basically, how you tackle a specific problem rather than the actual code that you're writing. Do you think there will be a time in the near future where actual writing code is relevant? And if so, will frameworks and libraries also become irrelevant? Is this is tricky because Yeah.

For for some aspects, yes. I I think, like, the the actual code will will become irrelevant for, like, just your table stakes stuff. But for for so much of the other stuff, it it becomes even even more important. Right? We look at, we had the PR computer guys on, and they were obsessive about the Node. Or I'm I'm looking at the, stuff that Mitchell Hashimoto is doing with Ghost2t. And what's this what's the thing he's working on? Super super super logical? Super logical, which is like a terminal multiplexer, and he's just, like, obsessed about their performance and, like, going deep and nitty gritty on it. And, like, it now it because we're building these crazy experiences, the the actual low level performance of the stuff matters so much more than than it it did before. So I think the code will matter differently based on the type of thing that you are are working on, but I do agree that it's going to be so much more important as to, like, how you attack things, how you how you solve problems.

Because, again, I know we said this in the last Wes, anybody can type into the box that they want. But when it comes to building a beautiful experience, into building a fast application, into building something that is actually going to be very good and be able to scale and provide an amazing experience, that's where the people that are able to do that, those people are gonna shine way above than just the meatbags that are able to type in a box.

The meatbags.

Yeah. I I mean, I I do think you know, I I've been the type of person that's been, like, hopeful that some of this stuff will still matter because I have so many skills wrapped up in this, and I have so many opinions. I have so many opinions. I'm wearing a Svelte T shirt here, so I obviously care about what I like to use. Yeah. I don't know. There are I I can't there is an app that I used React on lately that I didn't even look at the code, and it was fine.

I did not like did not like it, but, it was fine. I didn't look at it. So, do I think there's a time in the near future where actually writing code is irrelevant? And if so, I I mean, it depends on what the word near is. In in the near future, I do think that is a thing. I I I do think in the future, it will be that way. How far in the future? Depends on how how well the models and, process continue to evolve here. I mean, the day that we're recording this, Fable 5.1 is set to drop, and Fable one or Fable five was really good. So I don't think Fable five's gonna make it so we don't have to care about the Node, but I think each new model that drops gets a little bit closer there. So Mhmm. Yeah. It's a it's a it's a wild new world, folks.

Obviously, like, someone like us, we we've been doing this a long time. We've seen a lot of changes. And there were times where I cared nitty gritty about each individual function, and then there were times I stopped caring about that as much and install the library. I used to write my own auth from scratch somewhat regularly. And now guess what? I use BetterAuth, and I don't look and inspect every single file in BetterAuth. And to me Yeah.

Obviously, I think some people will disagree that this is the same thing, but it's just an a layer of abstraction.

Yeah. It it is. That goes back to the first one as well. Like, you you used to spend, like, weeks writing off and integrating absolutely everything, and now we have very good abstractions that we can simply just install, and and adding off is is now a couple hours, an afternoon to to get going Vercel versus the weeks and weeks that this type used to take. So it's it's just more abstractions on top of it, and and that is the way forward with so much. The I think the only difference that we have now is that it simply has just moved at this crazy pace in the last, what, seven months that everybody has a bit of whiplash.

Yes. Cool. Alright. Next question from Christina. Hi. From Czechia. Wow. Hi from Czechia. A big fan of yours. I have a question. I often find myself in a situation where I want to compare approaches, tools, and libraries.

For example, I'm trying to find alternatives to TRPC and figure out what to choose.

I'm honestly interested in other people's opinions on that. I'm a bit old fashioned, so I don't prompt everything. What a what a sentence that is.

I hate waiting for an answer when I could just find the right article on the web myself. But lately, when searching for something, I keep finding answers and articles generated by AI. Every website looks the same, and I hate it. I don't trust any site younger than a year old that uses, say, Tailwind because I also don't believe that so many people could actually love Tailwind. Christina, you're my kind of Vercel. JK.

So lately, I've started searching for the results that are a year old or older hoping to find an opinion on my question.

How do you find a good resources when web searching, or did you completely give up on that? Do you have any tool that could filter them out? I currently use DuckDuckGo as a search engine. I could really use a tag in the search results saying, don't click. It's AI slap. Well, I think the big problem is that, one, you are using DuckDuckGo as a search engine. Man, I love what DuckDuckGo stands for, but my god, I have a hard time getting anything useful out of DuckDuckGo. Yep. It's a bummer to me because I I do really admire what they're doing, and I've tried to Node DuckDuckGo for so long. And I just can't I just can't. There's too many times. I I think the thing that you're looking for is community here.

And Sanity, to me, is the answer to all of these things because I find a lot of value in some Discord channels and rooms on specific tech, and I get into these things where there's people that I trust.

You know, we have the benefit of having a large network. So if I'm asking for things like on Twitter, what do people I'm gonna get human responses. And the ones that are AI, I'm gonna block you. Right? So we have this, like, kind of ability to pull out from community opinions that most people don't have. But I still think that community and trust invalidation there in that information is the way, whether that is from trusted, I'm gonna say trusted posters on Reddit. I don't I mean, how how much can you say that? But, trusted posters in various chat rooms and various, communities that you become a member of and a part of, Slack rooms, those types of things. To me, that is where you find the value, and it's it's no different than going to your coworker and saying, hey, hey, Jeff. Is this a terrible idea? And then, Jeff, good story. Yes. You're you're done for thinking this, or, hey. You should check out this. Yeah. And I also think, like like, before all of his ASF, like, there were so many blog posts out there that were, like, clearly just

run up for the fact that they just wanted to SEO the crap out of it. Like like, Google results have been crappy for, like, four or five years now because people realize more traffic to your website. So they would write articles on absolutely everything, and they would do this very wishy washy high level comparison of three or four different libraries.

And, like like like, try it in Go, search for stuff. Yeah. Yeah. Like, the, state management library is JS is a really good example of one where, like, it it takes a lot of work to try them all out and to, like, get formed opinions on them all and write out a blog post, and so many of them are just these high level wishy washy things on it. Now I agree with Scott. Trust is the currency going forward. We we're already seeing this with, like, companies where literally anybody wants to launch something to get it out there.

It's impossible for these companies to get in front of of anyone. And that's why everything right now is is people are just being paid off right now.

Mhmm. And and you can tell. You can tell when when stuff is AI written. You can tell Wes people aren't genuinely into something, or you can tell when people are being paid off on something. And I think that that's why trust is going to be a huge currency Sanity, people that you will look at their word and say, yes. I I actually believe that. So you gotta get involved in a community. And and also from some same regard, like, start sharing that that content yourself.

Also, kinda related, nobody watches conference talks on on YouTube at all. Right? Like, I often will spend, like, two weeks writing a conference talk. That talk will be posted on YouTube or whatever or, like, if you go to, like, a conference that you missed and you wanna watch catch up on a couple of the actual talks, Nobody's watching them. They have, like, a 100 views, 150 views. It's unbelievable. Like, these are some of the the smartest people in that specific area, and and people are just, like, not diving into it. So I think, again, another aspect of this is just, like, fix your attention span.

Mhmm. And and don't hope to just be able to skim a blog post and and and absorb it all. It's like, spend some time and and dive deep on it. Listen to the experts. It might take a little bit more time. It might not be as exciting as watching, like, a a twenty second TikTok of some guys. Here are the five best state management libraries ever. You know? But you'll probably get a little bit more depth and and be a little smarter for it. Yeah. You know where,

people are actually watching conference talks? This should be not that surprising to you. The AI engineer conference.

1,400,000 views, 1,400,000 views, 1,000,000 views, 624,000 views.

Not all of the talks get that kind of views, but a number of the talks get major views, which I kinda shows you where the deep. Right? Yes. Kinda shows you where the conference,

scene is right now. One point. Yeah. But most of them, 5.6. That was that's pretty Node. 5,400,000 views. You know? Last thing I'll say here is being able to just test it yourself.

Like, AI can scaffold out the examples super quickly. You could come up with an example. Like, your example here is tRPC.

You could say, give me tRPC.

Give me better r RPC. Give me, like like, six or seven different possible RPC libraries, and I want to be able to update, pull data, refresh data. Like, give me an example of how to do this thing in all of them and have it scaffolded out, and then just go in there. Read the code, and and you can, you can form an opinion your yourself. That research used to be take so long to do, and now you can simply scaffold it out and make come to your own conclusions, by going in there, reading the code. Who read the code? Who who would have thought? Yeah. I mean, it all comes back to that.

Oscar Ericsson says, go Sweden. I'm a big fan. You're a big fan of of syntax or Sweden? Probably both.

So Yeah. Oscar links to this YouTube video, called the AI bubble survives on 1,600,000,000,000 in hidden debt. And it's kind of an interesting video. It Scott, like, goes into the economics of, like, people propping up and investing in a lot of these big AI companies and data centers and whatever. And, all there's just, like, so much money hinging on this AI stuff working out, and they're likening it to the two thousand eight, like, financial crisis where everybody, like, lost their homes.

And, we'll link up the the actual YouTube video because it's it's kind of a interesting watch. So he says, when I watch a video like that, I'm wondering how will this really work out in the end. I would love to hear your take on the financial situation regarding AI. Now I don't know if you necessarily want, like, economics or investing advice from from Scott and I, but, like, I I just Scott down a couple thoughts, and I'll let you know my thoughts here. First, Wes, I think this is scary because this has gone past, like, all of this investing and all of this money that is riding on there. It's gone past, like, oh, well, screw rich people. Screw AI. And it's like, no. Like, a lot of a lot of people's, like, retirements and Yes. Four zero one k's and a lot of the the stuff is is riding on this stuff Node. You know? And and if and if this pops, then then, unfortunately, so does a lot of people's, like, livelihood and and their investments and their retirement, all that type of stuff. So it's not just like a Wes, screw AI. It's it's become such a big pillar of the economy.

The other thing like, I was thinking about this the other day because, like, we keep going between, holy smokes, this stuff is so expensive, and, wow, these Chinese models are actually almost as good and are are extremely cheap. You know? Like, we just keep going between, wow, this is expensive.

Wow, this is cheap. And it seems like the the bridge between the cheap ones and the expensive one is is getting very is getting closer. Right? And I think that's why a lot of these AI companies are, like, doubling down on trying to get you into the the products. Right? They're they're they're not trying to just be the API that they can become a utility of anymore. They're trying to like like, how do we lock you into our system so that you have to pay us every single day? And and I just keep thinking about, like, how big Salesforce is. You know? Like, Salesforce is, like, hundreds of billions of dollars for I understand it's very complex, but it's just like it's not it's not that complex, but it's like there's so many people that are just have built their entire businesses on top of Salesforce.

There's so many people that are locked into it. Same thing with, like, Slack and all of the stuff. You wouldn't think that these companies are paying thousands of dollars per employee per year for a chat app. If you told me that, like, ten years ago, I would be absolutely shocked. Oh, especially with the how how

every other AI chat app that came before something like chat g p t functioned. Right? Yes.

And, oh, I'm talking about, like like, Slack alone.

I guess Slack is owned by Salesforce.

Yeah. Which is just like Is it? It's real yeah. It's really expensive

to have Slack for your entire Node AI editor as well.

Everything's converging into one thing. So

Wes what I'm trying to say here is that these companies are trying to, like, get that, like, like, corporate money. The enterprise money is where the big bucks are with all of this stuff, and that's that's also why Cloudflare is $5. You know Node Cloudflare JS $5 for me and you? It's because it's, like, $800,000 for, like, a medium company because it's that's where the the big money is. So I don't I don't see it having this massive, massive pop, and I don't know if people are are the higher ups are going to even let that thing pop, because of Yes. What I said earlier. So not too worried, but I do hope that a lot of this, like, warp stuff works out for these big companies that have so much money into it. Otherwise, there could be some, some big problems.

Yeah. I mean, it's I I tend to, like, just you Node, there was the whole, you know, financial crisis to the too big to fail stuff. To me, it just feels like we're actually in a world now where these companies can just do whatever the hell they want. So, like, are are we in a situation where they're actually going to be completely protected? Their house of cards is never going to fall down. It's kind of where I'm landing on it. JS that just that it's like the, cat. It's just always gonna land on its feet, and that could be an extremely ignorant perspective.

What I like, I just have a hard time with the folks that are constantly saying the bubble is about to burst, and they've been saying it for the past, like, two years now.

Like, I just don't know if that's gonna happen. There's too much money floating around back and forth. And, again, this this is very much a dumb guy perspective on this. You can replay this in a couple years when it crashes and whatever happens. But, like, I that's just kind of I don't know. Node I'm just in a, like maybe it's just like a cynical perspective where, like, yeah, you know, just everybody gets away with whatever they want these days. So, therefore, you know, whatever.

Yeah.

We'll see.

Yep. Yep. Yep. We we'll see. I certainly don't wanna be paying in ten years from now if we're still paying thousands of dollars per user to use a chat app. I I think I will say, like, where did we go wrong? Like, where's all the software that we said that this stuff would build? You know? I I think about that a lot as well. It's like, day to day, like, what software am I using that Yeah. That JS being built with AI? You know? There's something here with Wes AI. But there's just, like, so much stuff out there where I'm like, this sucks. Like, I had I went on this rant how dragging and dropping images on the Internet sucks. You try to drag and drop a hike. It doesn't know how to how to convert it to a JPEG. You try to upload something that's not the right pixel shape or or the the file size is too large. It it says, no. I can't do that. You drag and drop onto the wrong part of an app, and it downloads it. Like, what are we doing that we have this magical genie in a bottle you could literally just paste my rant into your genie, and it will fix it for you, but we still haven't. You know? Why is so much of the software still so crappy?

Yes.

Well, yeah, it it is.

Wes it JS. It is interesting. I mean, there there's a tweet that I always go back to where it's just like, okay. That's a pretty cool process you've got going on. Like but what have you actually built? And I'm not saying that no one's built anything, but, like, you know, you're not building Century with just AI. Right? All these people that will, like, say that, like, a SAS JS cooked because they can just copy any SAS. But I do think on a smaller scale, Wes, there's, like, some software. Like, there was there was, like, a very versatile image downloader that works on YouTube and, Vimeo and any site ever that has streaming video. Oh, you just click the extension and then click the video you want to download. It just downloads.

And to download videos of the highest quality from YouTube, it was like you pay $10 or something. And I said, you know what, Claude? Just make me this extension. And five minutes later, Claude's like, alright. Here's the extension. And now I have that without paying. So, obviously, there is some degree, right, where where things, are just getting completely replaced. Oh, yeah. Alright. So next question from Austin. What advice do you have for AI fatigue? Well, I guess this podcast episode Scott, doing good things for you, Austin. No. Before the days of agentic AI development, a typical workday involved a combination of authoring code and reviewing code. Nowadays, AI is authoring a lot of code, so I feel like all I'm doing is reviewing a lot of code, and it's exhausting.

Yes. It is. My output is higher, but so is my coworker's output. So I'm reviewing my own code, but also the number of PRs to review is greater.

I'm considering going back to writing code without using AI. I don't like this new era of coding, but I also don't want to bring the get off my lawn energy and get left behind.

I'm eager to hear the episode going over Scott's mental health survey. Austin, Wes, that video is out. And since that video, we have gotten man, we've gotten, like, an additional 2,000, responses to the mental health survey that we did. So go watch that video. It's called the true cost of AI coding.

And, yes, there is a massive mental toll from AI coding, not just from the amount of decisions we have to make. I mean, like, if you've used the grill me skill, grill me will ask you, what, 20 questions. And, honestly, the output is better when you use that stuff. But what is the mental health ramifications of answering 20 questions several times a day on features that get built and then having to review and look over the code and then go back and forth with it about it not being right or needing to be tweaked in these following ways or, all of the things. And there's people who say Node. Just use, get better, do all that, or stop being such a baby about it. But there is real, mental health effects that mimic addiction and all kinds of things from the process of AI coding. And, yes, I think we need to adjust ourselves. Now, Austin, I don't know if going back to coding by hand completely is is the move to solve these things. I think improving our process is the way to solve these things and improving the way that we work with these tools rather than just saying, you know what? Screw it. I'm going back to the Flintstones Yarn instead of using my actual car. As much as you want to, like, screw it, going back to only coding by hand, like, I don't think

I don't know. There's probably some people that are gonna be able to do that, have, like, this artisanal Etsy, coding thing. But I think if you wanna keep up in this industry, I don't think that that's the move. Hilariously enough, I think the solution to this AI fatigue part of it might just be more AI.

You know, like, the code review, the the PR checks, all of this stuff,

that will help quite a bit. Wes, isn't it our fault for for pushing this stuff on everyone, to always pushing it on everybody?

It's it's totally our fault.

But I I do agree. I think that we are trying to figure out how do we keep up with the sheer amount of stuff that is coming in and being able to check it over and keep things constrained and and stop things like like, obviously, security issues, but even just like UI drift, of like, oh, the UI just oh, you added in a couple font sizes there, and then six months later, your your app is just an absolute mess, from from everything that's been put in there. So I think that we are all in this spot right Node. We are all trying to figure out what does this look like, and I think that it's partially going to be just planning and teams being able to constrain themselves and and partially going to be new tooling that will help us have confidence in the code that we ship. Yes. Word.

Dave has a question about CMSs, not an AI question. Bring it on Dave.

Dave says, we have different editable blocks on our websites. Whenever we add a new block, editing the site becomes more complex.

We reach the point where having 20 plus blocks on the and the marketing team might ask us, how do we just add text to this page? Or we might use the wrong module for the wrong job. Do you have any insight on managing large flexible CMS systems? Initially, we thought we should limit the number of blocks to make development simpler, depreciation easier, style iterations less impactful, training less daunting, etcetera, etcetera. But since then, we've been looking at Apple and Stripe, and they clearly have no limit on modules and maintain coherence throughout.

Maybe the examples are indicator of manpower. Either way, have you ever mused over this conundrum? Yes. This is this is really interesting because back in the day of CMSs, what would happen is you would have, like you'd have a page, and then you'd have, like, like, a Bos. And you put all your text in there, and then and then people wanted a little bit more flexibility.

So you'd add, like, custom fields. You know? Okay. Well, maybe we want to have call to action. And then you could if you put text in there, then the call to action will show up. And then people said, no. I want, like, a I want a site builder. I wanna be able to have flexibility.

So we created this thing called blocks, and you could add all what you could just click them. It was it's like LEGO. You could just grab whatever block you want, click them together, and boom, you have this this website. And then the blocks didn't do what you wanted. Yep. So you made more blocks or you added you added settings to the blocks, and you had to toggle on. And then before you know it, you've got so many blocks, and then nobody knows where to go to edit absolutely anything. Right? Or or, Wes, or in Drupal world, you know, you you either get into the templates

or they have the, the dreaded PHP block Wes it's just, let me just throw some random PHP code inside this block instead of having it be editable just because I needed the PHP functionality, but I didn't want to build a whole module to do it. Right? Yeah. So then you have, the display components mixed with actual functionality components that people are supposed to be editing but are PHP. I mean, there's just yeah. It was always a mess. It was crazy.

Well, let me tell you. This is not a question about AI, but I think the answer to a lot of this stuff is is not limiting what people can do. To a certain point, you do wanna to limit people so that they don't go absolutely nuts. But and it's not just creating a million different possible blocks that that these people can use.

I think the answer to a lot of the stuff is is AI is actually really good at taking unstructured content and being able to to put it out there. So what a CMS looks like is is kinda kinda changing and and kinda up for grabs right now because the reason why we had such a rigid CMS was exactly for this reason Wes we wanted to be able to to scale up, but we still wanted to have some consistency across all of the pages. And and to answer your question about, like, Apple and Stripe, maybe they just have more manpower. They absolutely do have more manpower. They have entire agencies.

I've I received DMs. Whenever we talk about Apple websites, we receive DMs from people who work on it. They often have entire separate companies that work on just the iPad landing page. You know? Like, just big teams. So, yes, they they are able to do that with the manpower. But I think that what a CMS looks like in the future I know the, folks at Strapi are building this thing called Fimo at fimo.ai.

I think it kinda pokes at at this area, which is just like you want the structure of a CMS. You know? You have your your content that is somewhat structured, but you also want the, like, fluidity and flexibility of of an agent. And I think having an agent be able to manage your site and less so on the CMS is is probably going to be the future. I do it on my own website as well where I'm just like I I specifically use, MDX, so I can write markup and my content in the same file. And it's great because you can just type in, boop, boop, boop, what you want. And now granted I'm a developer, but I think that that we're getting closer and closer. We're just regular people are going to be able to to do to just use an agent to say, hey. I want to add x, y, and z on the web page, and it will pop it out. You don't have to fuss with all these stupid little rigid inputs and blocks and whatever because blocks were invented because humans need structure, and AI is so much better at taking messy inputs and and making them into a structured output. Not to say that it will look good. I think that you still need the guardrails and styles and whatever to make sure that it it looks consistent across all of your pages, but I think an agent is a really good use for that. So I messaged Pierre. He's the CEO of of founder of Strapi, what his thoughts are. I just messaged him, like, two minutes ago, and he didn't reply yet. So, I'm not sure if called the email. F I m o dot a I. This is making your website autonomous.

And and you have to think that this this is the the area because it's from Strappy, which is like a a CMS. Right? Yeah. Yeah. I I've been so out of the CMS game. When I worked at Ford,

every single market that Ford had had their own CMS for basically the same site just skin differently.

And so I was on a team of people that were designing and developing components and building out full on components. This JS, I mean, this is Angular one Node, full on components that we would then have to send to whatever markets existed, which you can imagine are a lot of markets, and then their internal teams would then have to then implement those components in their own CMS because each market had a different CMS.

The CMS game is is is kind of always been crazy, and I do think that, like you said, Wes, agents are perfect for this even, like, on small sites. You Node, I made a site for for Courtney's podcast, and I wanna think that, like, she could go into our AI, which uses Hermes, and say, update this on the website, and it would just do it. Like, she wouldn't have to go into a CMS. She would just go into her Hermes chat and say, hey. Update this. And And I also wonder with WebMCP, like, WebMCP is just gonna open this up even more Wes agents are gonna be able to talk directly to the functionality of a site, and it's just gonna do it. Yeah.

I I think, Sanity is in this area as well. I'm just on their their website. Sanity was the sponsor of the show for a long time, and they, like, every business, have moved into Wes do this with AI.

But they have a really good example on their website of, like, being able to just, like, click on the text you wanna change. Like, that was always the weird thing as well with, like, with CMS and WordPress world JS you would see the page, and you say, now where do I change that? And, like, often, it would be in the page, or sometimes it would be in the settings, or it'd be in, like, a like, a footer. You like, finding that piece of content. And, like, a lot of these, payload is another one of them as well. It's like, you should just be able to click on the text that is on your website, and then it should either you type it right there, like, content editable, or it should, like, pop up some sort of inline editor for you. Or you can click on that and then tell your agent, hey. Update this to x, y, and z, or add a section underneath this that will be a call to action or a product image.

Yeah.

CMS world, weird world. We I think we're due for another round up. A couple years ago, we did a, we did, like, a a whole roundup comparison of all the CMSs out there. It's probably one of our most popular episodes of the year, and,

that space has changed quite a bit. Maybe we'll do another roundup. Yeah. Hey. I wanna say this. Whoever is attempting to get into my x.myxtheeverythingx.com Me too. I've got, like, 20 password request reset things this morning. Stop it. Me too. I think There's nothing in there for you. Stop it. Don't don't, like, don't take don't don't do this. This is dumb.

I don't want these emails. Oh, I I'm I'm just searching it on on x. Absolutely, everybody is getting nailed password reset Wes. So I think Stupid. Some script kitty is out. Knock it off. Yeah. I'm my phone is just buzzing like crazy. I just got eight of them in the last That's what I'm saying. They're happening, like, all twenty four seven this morning. Gosh.

Alright. Well, next question for Bos was. Hey, CJ, Wes, and Scott love the pod and have been here since episode one. Yes.

Oh, yeah. Thanks for sticking around. My question for you is how you all would suggest or has seen risk Vercel devs take the plunge into leaving a stable job? I've been at my company for the better part of a decade. Wow.

They've treated me exceptionally well, and I'm now running out of people to learn from, and I'm at the upper end of what companies pay in my area. The problem I run into is that I'm the sole breadwinner for my family. So every time I think about jumping to a new company, how would you all think about this? Keep the growth on my own and stay comfortable in the position or take the leap of faith and start applying at bigger or remote jobs. The upsides of applying elsewhere are deeper, less local talent pool to learn from, and a larger salary, but that is a kind of secondary.

I'm specifically not unhappy, nor am I needing to chase that bigger bag. Any thoughts would be appreciated. Best Wes if you are happy, man, stay happy. If you are secure at your job, then that's, like, something that, you know, is, important because you'd see it in a lot of these bigger companies where you're just a number, man. They they've they, you know, snap that finger, and you're gone with, 10,000 other people or whatever. You know? So there is something I I've always been the person who does take the leap of faith, but I I haven't in a while. I guess the last leap of faith I did was when I quit my job at Ford to, do level up tutorials full time and then eventually start the Syntax podcast with Wes. But I had done job changes where I left a job I really liked for a bigger bag and then ended up hating that job and then left within, like, five months to take a different job because, the the talent pool at the new job was worse. I didn't like the work we were outputting. It was an agency work, and I was just, like, kind of unhappy. I was finishing faster than everybody else and just, like, not really, that excited to go to work anymore. So, yeah, there is something to be said about the comfort and stability of having a job and working with people that you know and trust. But then again, on the other side of it, you really have a lot of growth from moving on and experiencing new things and doing things differently.

It depends on on, like, if you were to take this new job and then get laid off in six months, like, what would you do? I guess you need to have that plan because, I I'm the type of guy who changes stuff up. I change things around. I I, you know, take new jobs, but then, you know, at 40 years old here, I think my position on this has changed to where I would, try to to to risk things. You know? I don't know. Yes. I

I think the risk here is the risk of not going to a bigger, better job is is, like, what's going to happen to your skills? If you're you're at the upper end of who you can learn from, if you can continue to skill up by bringing things into your job that are are going to be next level, learning new things that are coming, then I think that's gonna be fine. Or if you if you work on, like like, side projects or your own little things where you just put it out there to the Internet, often, that's how a lot of people find their, like, dream jobs is they simply just, like, working on stuff in public.

And and somebody that works at a company will be like, hey.

We need this exact skill set at our company. Do you wanna come work for us? So, like, that's that's maybe one, like, way to just, like, not actively go out and search for it, but to keep your skills up and to just, like, let yourself be semi available, just like putting yourself putting your projects out there is probably a very good move. So, like, I just don't want like, we look. We have this question all the time where people say, I've been at this job for ten years, and now my skills are, like, six years behind or four years behind.

And, like, I don't wanna get this message from you in three years from now when things are obviously changing very quickly right Node. And you say, like, now now my company is shutting down, and I'm not I'm not skilled up, and I'm not able to to jump ship to another one. So, like, I would probably just be very aware of where are your skills at in comparison to what it has so that if you were to have to jump ship for whatever reason, then you would be able to.

Otherwise, it sounds like you're happy. You don't want more money. You don't necessarily wanna change JS long as you can keep that that skill set up. That interest you're listening to this podcast, you're probably pretty dialed, so I probably wouldn't worry about it too much.

Yes. Probably a smart guy, Bos Wass.

Probably a smart guy, Best Wass.

Jurgen says something I don't hear much on podcast is how people are sandboxing their agents. I'm assuming you Yarn usually running them in YOLO mode. Are you locking them down in any way so they won't RM, RF your home folder? I have experimented slightly with containers, but I haven't figured out how that fits yet. There Yarn specialized tools like Node no dot s h, which I haven't tried. We appreciate your thoughts on it. But, yes, let me direct you to CJ just posted a video on the Syntax YouTube channel, and it is called CJ switches the names up every now and then. Thus is the the YouTube game. But, basically, he has a thirteen minute video explaining how and why you should be using different types of, like, dev containers, Docker, containerization.

Excellent, excellent breakdown of how these pieces work. And then he's also going to be working on a video he just told us yesterday. It's just basically instead of explaining how all the pieces work and what the actual options are, he's just gonna come up with, like, a full recipe that says just use this because I think that's what a lot of people want. They're like, I know I shouldn't be using YOLO mode, and it hasn't I haven't actually had any problems Wes. So maybe it's fine, but, it's probably especially when you're doing stuff for your employer or whatever Wes there's possible security issues, you probably should be running this in some sort of sandbox. So, certainly, subscribe to YouTube on Syntax on YouTube, and CJ has got you covered.

Yes. Perfect.

I love that answer. Yes. Because YOLO mode, sadly, is is, I don't I don't wanna give too much away how I operate a lot of the time. Yeah.

It's not just on on coding. It's just in life. So, Wes. I I I wanna see this because I need to get sandboxed, but it is such a pain in the butt sometimes. So I I need that. Alright. From dither Node, since I was stuck in bed for a short while using ox alpha test phase, I was prompting in bed for my phone. While a task was running, I asked the agent if it could determine my keyboard layout.

I switched to Colmack on mobile a year ago because QWERTY hurt my fingers to type fast. Wow. Colmack on mobile. One time, I switched to Dvorak on mobile, and, that was dumb because I didn't have a Dvorak on my laptop.

Which shockingly, it not only figured out, but it also correctly guessed my muscle memory as I use Qwerty on PC. Hey. What do you mean? Am I the only one who thinks anthropic, etcetera, will be using subtle fingerprinting technologies on their models to add target their user Bos. No. I don't think you're the only I think ChattGPT, like, announced that they were going to do that. I don't know if they ever did because I I haven't just straight up used ChattGPT outside of, like, a coding context.

But, no, I think they are gonna be doing that. And, get ready to, not only have the brands that, they think you'll love, shove down your face with the current for you pages, and now it goes that way, done even more because they have your every thought and everything you've ever sent to them. Yes.

Everybody Local models, folks. Everybody's fingerprinting you, which is is very unfortunate. Not to say, like, this ox alpha, whatever. I'm not sure which app he was actually using, but there are APIs that are able to tell you the keyboard layout, on mobile. So it might have just used those APIs to figure it out, but that's the besides your point is, like, they certainly are fingerprinting. Did you see that AliExpress was caught using audio APIs in the browser to they they weren't recording you, but they were playing a silent sound and measuring how a user specific device processed Wes God. In order to fingerprint them. So I Wes guess using the audio API, they played a silent sound, and and I'm I'm not entirely sure how they were able to turn that into a a unique fingerprint, but all of these APIs on the web are Scott. I don't know. Yeah. They're listening to it on the microphone. I don't know. Yeah. A user with Bluetooth headphones noticed that they couldn't play music via their phone because their headphones were playing AliExpress' silent sound from their PC. That probably has something to do with load or time to process the frames. You know? Something like an idea. Often, they will use several of these fingerprinting techniques, and they'll put them all together. And they say, ah, well, this person's computer was able to render this thing. It could play the sound, and it could process it at this number of hertz. And then they take all of these things together, and all of those together hopefully come up with some sort of unique fingerprint for that Vercel, and then you're able to track that person across all of their different websites that they visit, which is a bummer because so many of of the Node web APIs have been kiboshed because they could be possibly used for fingerprinting, which is such a bummer. Yeah. Yeah. Wes MIDI.

Yeah, I like wonder that when they're gonna start using, like, echolocation and map out your room. There's Sanity

talking. Oh, god. Oh, here we go. Brave. Brave says, there are slight variation on how each device plays the same audio file due to differences in CPU, sound card, browser, etcetera. When AliExpress played the silent sound, it measured these small variations to help build a fingerprint of a user's device.

Yikes. Yikes. Yikes. Yikes. Scott doing that. Stop doing that. Just let me buy my cheap electronics without,

tracking me across the Internet. Yeah. Do you really need to do that? Like, be cool about it. Do you need you already have a business. That business is selling stuff.

Yikes.

Okay. Podcast number Node, podcast number two. Would it be possible to have a visual cue on the videos that are on you that are YouTube exclusives? A specific emoji in the description or something into the thumbnail. Currently, end users that use a podcast app also subscribe to the YouTube channel have to play the Wes if this will show up on the podcast feed on every video on YouTube. At this point, although I'm subscribed, I pretty much ignored the your YouTube feed. Although, I would like to enjoy the content that isn't going to show up on the podcast feed. Thank you for your consideration. Here's yeah. You know what we should just do, Wes? We should have a YouTube exclusive playlist and just put everything there and then make that playlist the first thing on the landing page channel for it. This is tough because, like, we need to play some games with the YouTube algorithms.

Yeah. We used to put, like, the the show number. We used to put the show number in the title, but guess what? That that doesn't get get clicks.

You have to put it doesn't get it gets, like, deep de, deincentivized

or whatever. Yeah. Or or we used to, like even with, like, guests, we used to put the guest's name in the title, which it makes so much sense. You wanna know who's in that thing, but, like, people don't know who that guest is, so they don't click on it. So you have to say, like, this guy did x, y, and z. And it's it's such a frustrating thing to have to to play because we I understand. I just wanna be able to put, like, a hashtag in the thumbnail or be like like like I think only Joe Rogan can simply just say, like, j r e number one thousand eight hundred and put the name in it. But even him, I bet he could do he could do a lot better if he were to, like, put actual titles in it. We I will tell you, the Mondays episodes are on YouTube, they're gonna be live streams, and then they are released as a podcast the following Monday on the RSS feed. And then the Wednesday episodes are just exactly the same both on the podcast feed and on YouTube. And then anything else so if you could you could simply just look at the day that it was released, Anything else is going to be simply exclusive to YouTube.

Yes. So I I agree with that. Even do something in the thumbnail as Wes. You know? Like, because we do put the little syntax in there.

So maybe we could, like, change the color of it. We'll talk about it. I'll we'll bring it up. About it. Yeah. We have an on-site in October that we're gonna be chatting in the same room. So look out for that. We'll make some changes. We're always tweaking this bad boy, so we appreciate your feedback. Podcast one, podcast or YouTube number two. We appreciate that.

Cool. Well, I don't have anything, West. You got anything before we get out of here? Any sick picks? Any, thing you wanna share?

I've got a sick pick for you. Oh.

I, unfortunately, recently had to buy a new TV. So we had 75 inch Samsung frame TV. Beautiful. I know people don't like Samsung frame. They're like, oh, it's not a good TV. Shut up. It was sick. K? Kid hucked a toy at it. Whole thing cracked.

No longer good. Very, very upset. So what do we do? Bought a bigger one. 85 inch. Oh, you win 85.

Yeah, man. Oh, cool. The because, like, some of the feedback we got when I I've posted a photo of the 75. This is blowing my brain given that I, like, had a 13 inch growing up.

People said you could've gone bigger from the 75.

Yeah. So we didn't buy as nice of a one as the Samsung frame. It's not as flush or whatever.

It's not OLED because I'm I'm not gonna let a kid throw a toy at a $8,000 TV. You know? Yeah. But we got a Hisense, I think, u eight eighty five inch. Got it at Costco. It's about $2,000 Canadian, which JS, 1,400 US.

And it looks awesome. Sounds great. It's not like, I'm not, like, a TV snob, but, like, I do recognize that looks pretty good. You know? My wife's watching four k baseball on it, and, it looks sick. I'm I'm very happy with it. Aside from it's not being as thin as possible,

everything else, I couldn't couldn't be more pleased about it. So And how who cares about how thin a TV is these days? I mean, you put it on the wall, it's because we have Node. I know. But they're thin. They're oh, it's yours isn't your yours isn't as thin against the wall as the as the No. Because, like, the sides, like like, the sides bow out.

Oh, okay. And and, like like, when you walk by and you can kinda see behind the TV, our last one was, like, flush against the wall.

And it's, like, the Samsung frame was, like, less than an inch, and then this one was, like, three inches. But it's like, what are we doing complaining about this? You know? Like, unbelievable that you can get a TV that is larger than your children for for something that's not crazy. You Node? People Yeah. TVs have gotten so cheap. And, I wasn't ready to spend, like, a a ton on an OLED, so we went for the, like, a relatively cheap Hisense. And our our idea is, like, we'll have it for a couple years until the kids stop throwing stuff at it, and then we'll we'll move that to the basement. We'll get, like, a, like, an OLED once the prices on those come down, but we've been very happy with it.

Yeah. I I I've been the type of guy who's been, like, looking at getting and maxing out on the best OLED at 85 inches whenever. But, ultimately, I just don't need a TV.

The Samsung we bought a while ago is fine.

I mean, it's a Samsung, so it has a bloatware. The UI JS crappy. Yeah. Man, the new one has Google TV.

I love Google TV. People I don't people don't like it. No. I like it. Yeah.

I have portable Google TVs that I use for, like, my projector, or I have them in the basement that isn't a smart TV. The thing with me is that I buy a nice TV, and I hold on to it forever. So, I I haven't really looked at Hisense, though. We used to be like a Vizio. We'd get a Vizio from Costco when they were cheap. And,

there's something to be said about that TV that's just good enough. So I don't know. Check out Hisense. I Yeah. I it's it's kinda like a like a cheaper one. You know? It's the one of the cheaper ones at Costco, but it it looks awesome. And the other thing about it is that, like, you don't buy a TV for for the long run anymore because they just get better and cheaper and hotter every single day.

Yes. Yes. You stay at the same age.

Yes. Yes. Right. Oh my god.

Wes doing a Dazed and Confused. You never see that you never hear Wes quote anything, let alone a movie. Oh my god. It's one of the few movies I I did see when I was a kid. Yes. Wes Hot American Summer and Dazed and Confused. Oh, man. I love Wet Hot American Summer. Yeah. Have you seen the new one? It like, a new one came out a couple years ago.

I don't I might have. You know what? Courtney hates it. It's, like, not her type of humor. Like, her type of humor is Weekend at Bernie's and White Chicks and that type of I love those movies too. But Wes about American Summer? There's something about that movie that, I always just I I got it at, like, used at Blockbuster on DVD one time, and I watched it, like, a billion times. So no. I I don't know if I have seen the the new one, but I'll check it out. I need to know. It's hilarious. And the funniest part about it is that, like, I don't know who any

anyone is. I don't know who any movie stars are or any actors or anything. I just thought it was a funny video. And I just, like, I, like oh, I told my wife, we gotta watch this. And she's like, that's Amy Poehler. That's Yep. I I don't know who who all these people are. And all these people Wes that is, but they're hilarious. And she's, like, freaking out. She's like, how are all of these people

on this movie? Yeah. It's a very funny movie. Alright. My sick pick is going to be sanding pads. Wes, three d prints, apparently, look way better when you dang sand them. I I I started wet sanding three d prints lately. Oh. And I use these sanding pads. They're these little foam sanding pads, and they just start at, like, a three twenty grit and get all the way down to, like, a 3,000 or something. Oh, yeah. Either way, you just go one through nine here. You just start one. You do the next one. You do the next Node. And then the I I've been sanding all kinds of three d prints, and it's kind of a pain in the butt. But, like, for things that I'm gonna be keeping a long time and have on my desk or, like, the prompt boy that I made, like, the things that I want to be looking nice, my Node, it makes a huge difference. It, like, really takes that it's three d printed look and removes it. So, yes. It and yours looks awesome.

Thank you. It was a lot of work, and I almost sanded too much. I was getting a little transparent. So, but it is very smooth, and it feels great.

So check out these sanding pads. They're cheap. There's a whole bunch of them. I use them just, like, by running the thing underwater while I do it. Yeah. What's the name? Picking up dust dust. I I used to

in university, I used to flip Blackberries.

So I would buy I would buy, like, AT and T Blackberries, and then I would unlock them. I had this, like, sketchy website that would sell unlock codes for, like, $8, and I would lock them. And then I would take the the Blackberry apart, and then I would, like, rattle can it and then wet sand it to, like, look glossy.

And then I would put it on Craigslist, like, unlocked custom color phone, and they sold, like, hotcakes.

People love that.

Yes.

I love that. Well, also hey. I'm gonna actually throw something else out here too, Wes. I recently installed Jellyfin next to Plex because I wanted to to compare them. I know. And Randy shared this apps to Samsung, repo. So I'm gonna share this with y'all because apps to Samsung made getting the Jellyfin app on my TV actually possible for me, which I, like, couldn't do it before. And and I thought this, like, this little app was so slick, and it just made everything way easy. So now I have Jellyfin on my TV, and I can finally watch Miami Vice without desynced audio because, Plex was was giving me desynced audio on Miami Vice for some reason, and and who knows why? So, Jellyfin does not, and I'm gonna I'm gonna be watching,

the complete series of Miami Vice now. Yes. I didn't realize it. We our Samsung was not Tizen. It was Bos.

So, like, we could we could load the the Jellyfin app. No problem. But I we do have another TV down here that I had to, like, sign up for, like, a Tizen developer account and, like, side load it on, and that sucked. You had to sign like, get a certificate to sign the thing.

Node. Thanks. No. Thanks. This this is a perfect example. It's like, obviously, this is a Vibe coded app, and, like, somebody

fixed a problem. Made it work. Somebody fixed a problem. Yes. Took many steps and turned it into a couple steps, and now I got, some software on my TV that I flashed myself. Crazy.

Right on. Alright. Alrighty. Thanks, everybody, for tuning in. Catch you later. Peace.