Isaac Askew: Welcome to Never Rewrite. I'm Isaac Askew. Jeffrey Sherman: And I'm Jeffrey Sherman. And today we're gonna riff on the future of SaaS development and SaaS software in general. so to to set this up, Isaac was gifted some credits from Anthropic, Isaac Askew: Okay. Jeffrey Sherman: and he said we've been riffing on the question of like what does software look software development, SaaS software look like going into the future? And you know, the idea that anyone can just build their own version of your SaaS. And the idea of there's permanence, you know, one of the big things we talk about is you could build you can write your own mail server all day long, but no one's gonna deliver your mail because there's a whole lot of trust and permanence issues. And so Isaac Askew: Yeah. Jeffrey Sherman: Claude can build you a a mail server, but nobody will deliver the email because otherwise it would just get overwhelmed with spam. and so Isaac, I'll and I'll let you tell a story. Ba basically he's like, All right, I'm gonna see if I can Isaac Askew: Mm-hmm. Jeffrey Sherman: one shot with my free credits a company that we used to work for or he used to work for, and see how much of it it could recreate in one shot with two hundred fifty dollars in credits. So I said take it away. Isaac Askew: Sure. Yeah, so it's an email marketing company. I don't know if you want to name it or not. He intentionally didn't, so I was like, okay. But either way, people know, people can look up on our our LinkedIn's and find out. But I was curious, because I I was really just inspired by the last week where somebody reverse engineered Adobe's suite of software and Microsoft's suite as well. And you know, some of the haters are going like, well, you know, this particular thing as part of Premiere or Photoshop or something doesn't work and this buggy here, a buggy there. And it's really not the point. The point is like in a weekend somebody got it doing enough. Yeah, right. Jeffrey Sherman: Eighty percent. Right. Isaac Askew: So it's not Perfect and yes, maybe professionals will still use the other tooling, but I'm not even really talking professionals here. I'm just talking about like their average Joe who just wants to like quickly crop a photo or something and they don't need all the bells and whistles. It does that for them. And also if you've ever used Adobe, it is super frustrating trying to cancel. There's like an early cancellation fee. You have to like commit to a certain like six month cycle or twelve month or whatever. it should be illegal. Jeffrey Sherman: Yeah, they're legendary. Isaac Askew: What's new one? Jeffrey Sherman: Adobe is legendary legendarily user hostile when it comes to cancellisation. Like they are Isaac Askew: It's very frustrating. So there's I think there's a little bit of community banding together going ha ha at at Adobe, like look what look what we can do because they're so you know, like like me, a lot of us are just annoyed by you know, back in the day you just paid a hundred bucks for like Adobe C S two or something. then they moved the subscription model and it's just like a a pain in the butt, you know, like so either way. Jeffrey Sherman: Right. Well i it's the we you know, we live in a in a in an extractive age, it's where the term insidification comes from, where the idea that these monopolies have become so entrenched that they instead of competing, they're just doing things to extract more and more money out of their customers. And Adobe is often listed as one of those companies because their terms of service and their model is so very against their users that if their users had any alternative like it's clear that there is no competition because if there was competition they wouldn't do this. and Isaac Askew: Yeah. Jeffrey Sherman: the traditional competition has always been open source software, which you know if you've ever used GIMP, that's a pleasant experience. Isaac Askew: Mm, I have. Jeffrey Sherman: But like if you're going online in the cloud and all those features, it becomes much harder. And so there's this cloud lock in. Isaac Askew: Yeah. So that being said, you know, somebody reverse engineered a lot of the tooling and people were criticizing it. But if it's open source now, then people can just do pull requests against it and fix the features and eventually, you know, I imagine over time the features will be there and it runs well. and maybe even better. Who knows? in you know, in in a more modern architecture. Or like I know some Jeffrey Sherman: Mm-hmm. Isaac Askew: people rewrote things in in Rust, but maybe that was it. and Rust is known for being super efficient. So maybe it's like You get a faster version of something. So either way, I I had from Claude I had this two hundred and fifty dollar cloud credit thing and I had I didn't know what that was yet. I just I knew it expired. and then I read up on it. I'm like, okay, cool. So I have two hundred and fifty bucks to use before November fifth. Well, in the spirit of of reverse engineering, and the GitHub package is called reverse engineer anything, R E A. so I guess you could use that. To do it as well, but I just used Claude. I was like, okay, well let me rewrite or generate reverse engineer this email marketing company I worked at and just see how well it does. And so I know the main fable model is for like deep investigation and planning work and whatnot. So I decided okay, I'm gonna put Fable as the main model and then I'm gonna do ultra code, just max effort. And and I gave it a prompt that was essentially, you know, you're going to reverse engineer the software, investigate competitors, investigate open source software, and then also look into user sentiment to see which features are beloved and used the most Jeffrey Sherman: Mm. Isaac Askew: versus ones that people don't really use, because I don't want to spend any extra credits building some features I know that are not like no one ever goes to. Jeffrey Sherman: Right. It's some super useful if you need it, but most people don't need it. Isaac Askew: Yeah, I was like, can can it get the bulk of what this thing is supposed to do? Which is like sending campaigns and then managing contact lists. and then you set up your own mail servers through their configurations. It kinda walks you through setting that up for for DNS and all that. Jeffrey Sherman: So I I think that's a a big piece right there. So in the discussions Isaac Askew: Mm-hmm. Jeffrey Sherman: we've had, right, permanence and literally the ability to send email was the example we did. So in this one shot example that you're using, it Isaac Askew: Mm-hmm. Jeffrey Sherman: does not actually send emails. It has integration, so you can provide your own email provider and then you would have to pay for that. Isaac Askew: Right. So I was just thinking, i if we can't if if the moat for for email marketing is reputation and mail sending servers, then I just won't worry about that. Instead I'll use something like SendGrid or another API that I can plug in and have that send for me, 'cause they manage all that. Jeffrey Sherman: Mm-hmm. Isaac Askew: So that was the idea. So I mean I think part of the prom mentioned that too. Like we may have to use stuff like for even for SMS sending too. And I didn't know if that was a big deal with things getting marked spam with text messaging. I don't I'm not as familiar with it. But I figured, okay, we could just use Twilio or something like that for the SMS sending as well. So what I did was I hit enter. and it thought Jeffrey Sherman: Ha ha ha. Isaac Askew: for a good few hours, probably about three hours. of Jeffrey Sherman: Well. Isaac Askew: investigating, planning, doing all the user sentiment analysis. and then I went to see Mastodon in concert. I'm like, okay, I'm gonna Jeffrey Sherman: Nice. Isaac Askew: tie it I'm gonna turn on remote control so I can kind of look at the session while I'm at the concert. And I go there and then in between shows I see that it's done with the with the plan. and then so I go, okay, well plan's built. And usually what you do is after you you don't need to use the most expensive model for everything. Gotta be a little economical about it. So I swapped Jeffrey Sherman: Do you know how much money it cost to use the fable to to figure out what a reverse engineer should be? Isaac Askew: Yes. I think it was around seventy three dollars in in in cloud credit. and I didn't see the the the big file that it looked that it made. I I actually was just you know, I'm at a concert. I really I'm not gonna like an open an MD file. Jeffrey Sherman: I mean it's an amazing concert. They're like, I'm gonna check this concert's amazing. I can't wait to check my experiment. Isaac Askew: Yeah. Well I was able to do it like like between the Jeffrey Sherman: Separate. Isaac Askew: had a couple of openers. between one of the bands it said it was done. So I swapped the model to Opus five point five, and then put it on medium effort. So that's I think that's like second lowest. Jeffrey Sherman: Mm. Isaac Askew: Right. and I just said build it. Hit enter, you know, or tapped it on my phone. and then I got back to my show. And then when I got back home it was done. So Jeffrey Sherman: Yeah. Isaac Askew: I I didn't know what dumb was and I was at a concert. I'm tired. I'm going to bed. So I check it out the next day. We're gonna you know like see what it spun up. and there was a couple things I had to configure to get it working locally. Like it was there were some commands that said to to run but they weren't installed on my machine because it had done it in cloud environment. So I had to kind of finagle with it just a few to to get it booted up on my local computer. But it looked pretty good. And I was like, okay, well I need to start and see if this thing actually sends a campaign. it provided a ton of different options for connectors for sending. So there's like the Amazon what was it called? S S E is it S E S for that one, I think. Jeffrey Sherman: Yeah, I think it used to be simple email service. Isaac Askew: Yeah, so yeah, I could use that. I could use Syngrid. you could also use just an SMTP, like a Google one, like if it would give you Google's SMTP and you log in with like a one an app password for your account. so there's a couple of different options and I just picked the easiest one for me, and at the time I actually just used the Gmail one. Jeffrey Sherman: Okay. Isaac Askew: and then I went back through and created a list and added my contacts to it and made a like a it made a WYSIWEEG campaign editor too, and I was just I didn't really tinker with it too much. I just wanted to see what it sent and Added you to the list. Jeffrey Sherman: Yeah, I saw apparently there's a WYSI there's an email format now, which I didn't there's an email markup language. Isaac Askew: No. I didn't see that. Jeffrey Sherman: Which is what it was talking about. Isaac Askew: cool. Well, either way, I added myself and you to a list and sent the campaign and it seems to have arrived for both of us. Jeffrey Sherman: Mm-hmm. Isaac Askew: so I'm like, okay, well cool. You know, that's interesting that I I think I it it actually ended up using almost all of the all of the cloud credit. So I I think I had like I don't know, ten dollars left. So it it still spent it's it must have spun up a ton of sub agents with Opus V, even with medium effort and to spend the rest of it. Jeffrey Sherman: Right, so it's like seventy five dollars to figure out what to build and then hundred fifty ish building it. Okay. Isaac Askew: The rest is just actually building it, adding tests and all that kind of stuff. Yeah. So let's say in the end, let's just say I exhausted the credit, 250 bucks. And as I'm clicking around and playing around with it, like it it does what I need it to do. It's sending it, people are receiving it in their inbox. and I was thinking, I could even use this for our podcast mailing list because there was other services that were trying to charge me. There's like a free tier, but you can't use RSS feeds in the free tier. So you had to upgrade and pay five dollars a month for this other thing. I'm like, screw all that. I Jeffrey Sherman: Mm-hmm. Isaac Askew: can do whatever I want. And if the feature's not there, I can just say, Hey, build the thing. I don't like the way this other email marketing company does RSS. It's tricky and weird and it's not Working right, so I can fix the bug. Right. and and for me it was free. They gave me credit. And like I don't know if you've played around with it locally, but every now and then both ChatGPT and Claude give you just like a free reset. It's like you you had your weekly limits. Click this button and we're gonna reset and you you're good to go from scratch, like all for the whole week, which is great. Jeffrey Sherman: No, I've I'm on an enterprise thing and the company gets I don't know. The it has never mentioned to me it mentions how many tokens I'm using but it never tells me about costs or Isaac Askew: Well there's like a weekly limit and a daily, like a five hour limit, right? for Claude at least. And every now and then they'll be like, You're out of credits. Now we're going to you you're you're out of your like for f if you use the the monthly plan, which I use, it'll tell you when you're done with your limits and say, well, we're gonna stop now unless you want to exhaust your actual money that you have. Like you might have a hundred dollars in credits. Jeffrey Sherman: Mm. Isaac Askew: on top of that, then it goes into that reserve essentially. And every now and then I mean if you turn it off and No, I don't want you to do that, it won't do that. But every now and then it'll just get say you have two free resets this month and you can reset and like wipe your whole use of distance out from scratch. Which is great. That's just another another whole another free week of coding. I don't know why they do that, but I enjoy it and So anyway, whenever that happens to me, I I'll use that for some creative stuff or like play around because the like my resets are usually like once on a week on Wednesday or something. And if I hit that limit before then, I'm like, bummer, I don't wanna pay for anything. I'll just wait for let me check and see if I got a free reset. Anyway, all that to say it did a pretty good job at building it for two hundred and fifty dollars. And it the design was more modern and looked fresh. the UI was nice and it wasn't clunky. and it built the whole thing in TypeScript. So it was like super zippy and fast locally. I thought it did a pretty good job. I didn't test every Jeffrey Sherman: Okay. Now did it Isaac Askew: feature exhaustively. Go ahead. Jeffrey Sherman: did it use open source libraries or did it basically clone or reverse engineer those as well? 'Cause I I wasn't sure 'cause I I don't do TypeScript, so I didn't go into the code particularly, but it mentioned like atomic CRM as a CRM thing. Like so do you know if it pulled in atomic CRM or did it just basically clone it? Isaac Askew: What I saw it pull in, it inv it investigated all of the competitors that were open source and and not not just the open source ones. In this investigated all competitors and their features. It investigated open source libraries that did email marketing. it also pulled it looked at Active Campaign's public API contract and public SDKs to understand how the API is built. And so I think it was trying to reverse engineer not just the functionality of it, but maybe it was trying to understand from how the contract is described what it's intended to do. So I happen to know the the company is built on PHP, the original company's built on PHP. and I didn't use any intimate knowledge of the inner workings of how the company, like I you know, used to work, at least I haven't been there for six years. So this was all in TypeScript. All you know, I even saved the session just as proof that I didn't tell it. I know some secrets about how it works. I was like, you just use public information and go. Jeffrey Sherman: Yeah, that's Isaac Askew: So I had the whole prompt and session saved as like a kind of a proof on the side, just if anybody was curious. so yeah, everything it it pulled all that. I don't know if it did a one-to-one clone of the public things that existed. I would have to go back through and see which ones are in TypeScript or not. Or maybe maybe it did, and then they just moved it to TypeScript for whatever reason. And it even named itself Turn T-E-R-N. Don't know why it chose that name. I didn't give it any direction other than investigate and then when it was done, build it. Except for the the few Jeffrey Sherman: Interesting. Isaac Askew: things I I taught to to make it boot up locally for me because there were some commands that were broken, like Docker stuff that needed to be installed. But other than that, gave it zero direction. and it I think it did a pretty good job. Jeffrey Sherman: Okay. Now I'm gonna play devil's advocate for a moment. Right. Isaac Askew: Mm-hmm. Jeffrey Sherman: So the the there's a SAS and you have now in in theory, like a let's assume it works ninety ish percent correctly, and that is good enough to save you the or it it's worthwhile, maybe. Right, 'cause now you so you're paying to send emails in theory, like you're gonna Isaac Askew: Mm-hmm. Jeffrey Sherman: use Sengrid or some low cost provider. So Isaac Askew: Mm-hmm. Jeffrey Sherman: you're using Sengrid, so you have to pay for the emails, and now You know, the the the goal is okay, you're gonna save I don't know what the monthly cost is, maybe fifteen dollars on a minimum account, right? So the Isaac Askew: Mm-hmm. Jeffrey Sherman: to save fifteen dollars. But now it's running off your local machine, which Isaac Askew: Mm-hmm. Jeffrey Sherman: maybe is okay in this instance, right? You've got it maybe, maybe Isaac Askew: Yeah. Jeffrey Sherman: not. But if you had to pay, you know, get this thing hosted somewhere, and so now, you know, t for permanents as well, like to back up the database, so that you can keep all your things, because right now you are one computer crash away unless you have a backup system set up for Isaac Askew: iCloud. The folder sinks to iCloud, so we're good. Yeah. It's in the documents Jeffrey Sherman: Does it really? I mean I guess it could. Isaac Askew: folder. Jeffrey Sherman: Okay. So right, you don't really need hosting. But if you needed hosting, right, 'cause Isaac Askew: Yeah. Jeffrey Sherman: it right, if you wanted to use something more advanced and just send a campaign at a certain time, if you wanted to have it do automations, which is one of the big Isaac Askew: Mm-hmm. Jeffrey Sherman: things that the company does, you'd need to have some kind of permanent instance running somewhere that's not gonna periodically just power down because your computer went to sleep. Isaac Askew: Mm-hmm. Jeffrey Sherman: How much would host like 'cause I I have not set up hosting myself in decades. How much is hosting cost these days? Isaac Askew: good question. it depends on what you use. Versail is slightly more expensive. I think the tier I'm on for Versail is twenty a month, but for for a separate project, but I I don't think you actually have to pay twenty a month. I think there's like a free tier up to a certain amount. There's a lot of free tiers out there. I know Light Sale Jeffrey Sherman: Okay. Isaac Askew: is dumb cheap for AWS Light Sale. I used that to host WordPress a w long time ago, and that was like four dollars a month. and that's It's a heavier thing 'cause you're also running a database and everything on top of that. so Yeah, yeah, yeah. I'm not I wasn't saying they they weren't mm-hmm. Jeffrey Sherman: Mm-hmm. Well this is the database. It said it's using Postgres. But I mean it's all running in a Docker container, but it it seemed like it had Postgres and Redis. Isaac Askew: Yep. So I don't know, not not very much, but I don't also know why you'd want to host it there. I if you could do the sending from your machine in a safe way that doesn't expose any information, then yeah. Jeffrey Sherman: Right, which you could because the actual sender is a third party that you're paying. Isaac Askew: Right. I might I don't have any exposed ports or anything, so I I would just let I mean me personally, I would just let a Quran run locally and then use SMTP Google servers to send my stuff. Jeffrey Sherman: Right. And in this case, security's not much of an issue because it's running in a Docker container on your computer. It's not on the cloud. It's not a there's nothing you can poke at it. Isaac Askew: Right. And I don't know, I mean y you could argue how long would I have to have ho you know if I had to pay the actual two hundred and fifty dollars to build this thing, then Jeffrey Sherman: Mm-hmm. Isaac Askew: no, probably not worth my time, but it would be over the course of years, but Jeffrey Sherman: That's years. Isaac Askew: it's years and also I can fix things that aren't working. I don't g like last time I actually used the company, try I tried to for a separate project. And something was broken and it just wasn't working. And I was trying to contact support and it was just awful. I'm like, what is the deal with this? It Jeffrey Sherman: Yeah. Isaac Askew: was super frustrating. I don't have to deal with that. I can just like because I'm an engineer, I can just make it work. And that's the whole point of the open source thing. Like any average Joe might not know how to go navigate GitHub and get it installing and working locally. They could tell Claude Jeffrey Sherman: Yeah. Isaac Askew: too, you know, but like The idea is more like and and if I keep getting free resets too, it's essentially cost me nothing, right? I don't I don't there's not a reason to to use it, to use the real software if if mine can also do it better and more depend dependably. And if something breaks, I don't have to actually contact support. I can just say Claude, wind it in his end, and it goes through the logs and figures it out for me. Like this is all it it w it's it solves my problem. Jeffrey Sherman: Right, you're it's locally running Isaac Askew: Yeah. Jeffrey Sherman: right, it basically i it's open source software that runs locally and solves the same problem. and it's not as full featured, but you don't need th I mean that's one of the problems with the SAS model is no one needs all the features, but you have to pay for Isaac Askew: Mm-hmm. Jeffrey Sherman: them all. Isaac Askew: Yeah. And also if, and I've seen this happen before, if some big customer is on the same database instance as the smaller customers and clo customers and something bad happens and the whole cluster goes down, you know, your reliability is kind of whack a bit. Like you need to make sure, you know, you're still dependent on the fact they're trying to host everybody and support the whole like the whole service is live. And so if you don't have to worry about, man, is Twilio down or is this down or whatever down, then great. Jeffrey Sherman: Right. It it removes some of the problems. I mean, your computer could go down too though. Isaac Askew: Yeah, and if it did, big whoop. I just reboot it and send it when it booted back up. It's to me it gives me ultimate control over everything. Again, I am Jeffrey Sherman: Hmm. Isaac Askew: coming at it from an engineer mindset though. If you're a small business owner trying to use, you know, email marketing software for some other thing and you the learning curve is too hard to figure this stuff out your own, totally get it. Stick with what you know. But if you have a company that Jeffrey Sherman: Right, well there's also the Isaac Askew: starts jacking up costs and you're like, okay, this is getting a bit expensive for what I need it to do, then absolutely luckily. Jeffrey Sherman: Right, there's also just the you know, from a pure price standpoint, it sounds like the break even time would be about two years on this kind of thing. Where if you use this software Isaac Askew: Yeah. Jeffrey Sherman: instead of the SAS for two full years, with your time being not worth anything, then just any pure cost, you'd break even, and so then it's how much is your time worth Isaac Askew: Mm-hmm. Jeffrey Sherman: to to putz around with this stuff. And so Your time's actually very valuable. But, you know, it still it would be like a three year break even 'cause you'd still have to invest time in using Isaac Askew: Yeah. Jeffrey Sherman: some other product that doesn't work exactly the way you want it to. Isaac Askew: And depending on how valuable your time is, if you have to wait a day or two for a customer success person to to resolve your issue, Jeffrey Sherman: Mm. Isaac Askew: and you could fix you, call I could tell you what happened in less than 10 seconds and fix it for you. That's a big selling point to me, the turnaround time. Especially dealing with support these days where everything's going to be like automated support, you know, AI responses, and you have to escalate to real person. I had to do that with Comcast Jeffrey Sherman: Right. Hmm. Isaac Askew: recently, and that's like really frustrating. Jeffrey Sherman: The Isaac Askew: Right. So and after a while I'm like, I'm done. I'm just anything I need here, I'm just gonna be what's the easiest way I can do it? And that 'cause I'm not gonna support a company that like makes me like wanna pull my hair out just talking for basic support. Yeah. Jeffrey Sherman: Yeah, that's an interesting we should do another episode on that. The idea of one thing AI lets you you running it locally means you have access to the logs, which no SaaS company is going to do. Isaac Askew: Yeah. Jeffrey Sherman: But it all but that that makes you able to debug your own issues. Isaac Askew: Yeah, and also we we talked about it before the on-prem stuff. Like there's a certain security of having it hosted on your own machine if you're if you have the know-how to keep your own data secure. You don't have to worry about the company getting hacked that you had the all your information on. It's kinda nice. Jeffrey Sherman: Yeah, that happened to my accountant. His cloud provider the accounting software provider got hacked. And so his customers, because it was on the cloud, which Isaac Askew: Yeah. Jeffrey Sherman: included myself, got hacked. not due to his office or anything he did. Isaac Askew: Yeah. So that's kind of a nice selling point for me. Jeffrey Sherman: Mm. Isaac Askew: yeah, so I it but it really got me thinking like, okay, that was just I mean also like I I'm actually curious to know how many emails personally I could send in an hour. I wonder, you know, with a with a proper queuing system. Jeffrey Sherman: I I would imagine the the not enterprise version of Gmail, the the personal and free stuff, will have a fairly low limit, but Isaac Askew: that one will, but I'm talking about if I swap to SES or something. Yeah. Probably I'm I'm just I'm just Jeffrey Sherman: Yeah, I don't know how much it Isaac Askew: remembering prior limitations of seven hundred and fifty thousand an hour for the largest customer from back in the day, using the database as a queue. Jeffrey Sherman: Th those are PHP issues. Those are Isaac Askew: Yeah, exactly, which we don't have in TypeScript land. So I'm like I'm actually just a little bit curious how what the throughput would be with this locally hosted thing. Jeffrey Sherman: Yeah, that's that's a good question. I don't know. Isaac Askew: But what is but yeah, what's really got me thinking about though is just like if if if software like Adobe Photoshop Suite has been like features added to it over the course of you know decades, it's you know full-fledged, awesome software to use. The company sucks, but the software's great. so like if if somebody can reverse engineer good enough, and now all the features Jeffrey Sherman: I mean. Isaac Askew: like the you know, again, like basic editing, but not like any kind of Certain blur that you need to affect or whatever, I don't know. and get does it good enough for people. It's not even about like me or you generating it. It's just like somebody somewhere is going to create a verse reverse engineer of some software. So it doesn't even matter that I spent 250 cloud credits on it. Somebody else might already have done this and it's open source and I can just look and see. You have like a you're you're gonna have a whole globe of people creating all enterprise software and putting it. I I bet you GIMP will be a lot better. I'll put it to you that way. Jeffrey Sherman: I bet you GIMP will not get better. But there will you know, Isaac Askew: Mm. Jeffrey Sherman: this being open source software there'll be what would GIMP be backwards? P I G P Mig Isaac Askew: Sure. Well, either way, I I think that it's just an i it's an exciting time because it like i so I was thinking about like my my most recent company, Papaya Pay. They Jeffrey Sherman: Okay. Isaac Askew: paid but we talked about this before, I think on one other episode, but they but they paid you would attach a payment method and pay a bill, submit a bill through their app. And back in the day that was cool. And today I was like, Muse can do that too if you trust Muse, which I don't. Jeffrey Sherman: Yeah. Isaac Askew: You could tell it, hey, here's my credit card, go pay this bill for me, here's a picture of the bill, and it would find a way to pay it too. And I'm like, ooh, wow. Okay, so a whole ten year product, not everything they did, but that particular part of it is is getting replaced by just Claude Do It For Me. Claude Do It For Me is replacing a lot of a lot of bits of software, whole companies that, you know, before made it their whole mission to take away that friction for you. And now they can do it themselves. So you have this empowerment, but I'm also like nervous about like the actual economical impact of this when people start to realize one they can do more things with less people because the engineers are very expensive. So that means they're their people can make more money because they have more profit because they had to pay less engineers. That's gonna make them excited. but two, like if and you started seeing this less recently like HubSpot laid off six percent of their company, Netflix a couple days ago laid off seven percent of their company. you know, like layoffs each year, it's not just normal layoffs. I'm pretty sure these are like AI specific layoffs, or that's my hunch at least. Then where do these people go? What happens to software in general? My idea was like maybe okay, we you end up getting to more towards contract work where you gotta bring in a developer to fix this issue and then they fix it and they're gone instead of it being like a W-2 employee, unless you have that one W two employee. tinkering with the thing the original contractor set up for a lower salary or something. Because what's the need if if Claude can just generate it for you? To to have Jeffrey Sherman: Well I think there's the Isaac Askew: go ahead. Jeffrey Sherman: like you know, in the old days it was and and by the old days I'm gonna go with the eighties, with the whole GNU and Unix philosophy where you Isaac Askew: Mm-hmm. Jeffrey Sherman: wanted to have a lot of very specific programs that did one thing. Grep Isaac Askew: Mm-hmm. Jeffrey Sherman: does one thing. And if you wanted to you know, do any kind of control over the output, then you would grep it and then you'd pipe it to less and then you'd be able to to control the scroll. And so you'd have all these very specific, very thing, very tight programs. And then, you know, with the many cycles later, you know, programs got bigger, then they got smaller, then they got bigger and smaller. But you had, you know, like Office, Microsoft Office was this not only is it a word processor, but it's also a spreadsheet and it's you know, Isaac Askew: Yep. Jeffrey Sherman: it's it's sort of the pinnacle. And then it it went down again. and then it w you know with the cloud and the subscriptions, there was this pressure to always to make everything a big I've got a s cloud pe I've got a piece of cloud software and I want you to come to my website and use my software and I want you to stick around as long as possible. And so you can have more and more pieces sticking around longer and longer. And Isaac Askew: Right. Jeffrey Sherman: I think this is gonna push us back to the other cycle of, hey, you need a product and it needs to do one thing and then just have multiple products. Don't have one product with a giant subscription like Adobe. Instead have Isaac Askew: Yeah. Jeffrey Sherman: many products that are cheap. And so you aren't continu you the there's no reason to continually work on these things. I Isaac Askew: Yeah. Jeffrey Sherman: mean it becomes a cycle of well it it's big because it's expensive and it's expensive because it's big and so I need a whole you know, I need to have all these developers and so on, then I need to justify what all these developers are doing. It's like, you know, maybe you don't. Maybe you just decide, hey, I'm gonna build this email system and I'm gonna amortize it over two years and I'm just not gonna you know, I'll do bug fixes, but I'm not gonna keep pushing new features all the time because that's expensive and my customers don't want that. Isaac Askew: Yeah. Well, this too is one I thought Microsoft has failed a lot lately to me. But I thought one brilliant thing they did recently was like they were announcing like this partnership with NVIDIA and this new you know series of desktop computers that they're gonna build. They're gonna be kinda like AI optimized desktops. And I think that's what we're going to come to expect from AI instead of just being, you know, you talking to cloud or you you it'll be some Claude, it'll be like you talking to your own local computer, running and building Jeffrey Sherman: I mean Isaac Askew: things locally, quickly. and solving those problems without reliance on a third party. Like just just having to collect all the data on your own computer and then like Microsoft owns that data and sees how people were using it or just makes it more intuitive process for that for their operating system. I think that will be I think the on-prem thing and just like making sure each computer out there can do this without being dependent on the company doing it will be a huge deal once it gets faster and cheaper. Jeffrey Sherman: Yeah, that makes complete sense. Where we'll go back to essentially shrink wrap software where Anthropic is going to sell sell you, here's weights, here's a harness, install it Isaac Askew: Mm-hmm. Jeffrey Sherman: on your machine, and it will work as fast as it your machine works, and there's no credits because it's your machine and your electricity and go knock yourself out. And you know, and if you want the the monthly weight update, that's a subscription that we will charge you for. And that's what they'll start making money on. Isaac Askew: That and and maybe collecting the kind of queries that you're giving it to or something like that, so they can have some a bit more data around you know, what people are searching for and how to optimize the the work that's being requested from machines. Jeffrey Sherman: Right. I mean, I'm sure that there'll be a tier where they spy on you and a tier where they don't spy on you. Isaac Askew: Exactly. Yeah. So yeah, I think the future is going to be very weird. Jeffrey Sherman: Should if the future's so bright you're gonna need shades? Or if the future's so weird you're gonna need Isaac Askew: Yeah, we'll put it you might go blind. I like that better. The future's so bla bright you might go blind. Jeffrey Sherman: The future's so bright it you should really make sure that you're got eyeglass protec eye eyewear Isaac Askew: Yeah. Jeffrey Sherman: protection. All right, we've mangled this and we're over time. So thank you all for listening. I'm Jeffrey Sherman. Isaac Askew: Ex Excellent. And I'm Isaac Askew, and this is Never Rewrite.