s03 e05 The Future of AI: Google vs. OpenAI

Sean Hise (00:01.23) in five, four.

Sean Hise (00:06.436) All right, we're back. This is the Big G's AI podcast. Today we are remote because Crafted decided to move to a new office. So our studio is down right now. But here I am always with my boys, Brandon and Jacob. What's up dudes?

Jacob (00:22.83) What's up, man?

Brandon Corbin (00:23.743) Howdy! Howdy!

Sean Hise (00:25.68) I'm getting kind of used to this podcast set up because I need to look at the camera a little bit, right? But I'm like, you guys are down here. Like, what do I do with my hands? I'm not sure here. So.

Jacob (00:34.382) I'm just gonna readjust while we're on air here. Okay.

Brandon Corbin (00:35.176) You

Sean Hise (00:37.84) so no, we, we, we're, I'm actually in the new office. Jacob's in the office next to me. We moved down Mass Ave a little bit. So we're, so we moved to a new office. It's no longer in a yogurt shop anymore. and you know, we have some more space and we're, actually have a dedicated podcast studio room that we're building in the new office, which Brandon doesn't know about yet. So we're excited.

Brandon Corbin (01:00.795) No, so what's the plan there?

Sean Hise (01:05.436) You'll see man, you'll see next week. It's gonna be different. I think it'll be a little better and it'll be permanent so we don't have to spend so much time getting set up every week.

Brandon Corbin (01:18.718) How's the move been for you guys?

Sean Hise (01:22.684) It's been good. it's been kind of, we moved all our stuff over and now we're getting painting done, But we've found that we can reuse most of the stuff from our old office, so. It's not a total loss.

Brandon Corbin (01:35.496) That's nice. you find the. Are you finding that real quick, Jacob, are you guys are you finding that it's screwing up any of the work that you guys like? Is it screwing up any of the mojo as you guys are trying to transition from the old space to the new space?

Jacob (01:35.918) It can't, I think it feels like a.

Sean Hise (01:51.804) Yeah, I'm completely thrown off. I'm the type of person that needs to have everything in its right place at all times. So yeah, I'm very thrown off right now, but it's getting better.

Jacob (02:04.013) You

Sean Hise (02:05.378) Anyways, hey, check it out. So we've got quite a few topics today. It's been kind of a weird week for AI maybe, like the slowest week in AI in some ways. In the last three or four weeks.

Brandon Corbin (02:17.63) With only like a hundred news articles coming out. It's like the slowest week of it. It literally is all like just this just absolute it's almost impossible to kind of keep track of everything. We'll go through this week and we'll send each other all these links. They know we'll post them up on to Slack or whatever. And then at the end of the week we're like, my God, there's just there's there's 15 other things that we didn't even cover. I mean, it's crazy, dude.

Sean Hise (02:22.99) Right.

Sean Hise (02:43.482) Yeah, it is. it's either like, you know, there's a $1 trillion economic impact event with DeepSeek or something in the last couple weeks. But now it's more like the aftermath and people are trying to catch up. People are trying to do some things. But we're going to kick it off today with Jacob's favorite company, Google, and the Gemini 2.0.

release that no one seemed to even care about. Google also released their earnings and didn't seem like the market cared about it either. But let's just talk a little bit about it. And Brandon, why don't you kick it off, just get it started with, because it is quite a big announcement that happened this week in terms of these big drops these companies have been doing.

Brandon Corbin (03:27.677) Yeah. Yeah.

Brandon Corbin (03:34.107) Right, so it is, it's funny because like, so Gemini 2.0 Flash is coming out and the whole model for this one's supposed to be again, huge context, super fast, super cheap. And so it can literally do a, it could do all seven Harry Potter books in one prompt, right? So it could take a ginormous PDF.

and it can process that and it can then do like the, you know, the needle in the haystack and be able to find content, be able to actually understand it and kind of interact with it. So it is actually huge that, that it can process this much information. Now I don't have access to it. So I just went once I saw, you know, once this kind of was announced, I went to look at my Gemini and I don't have access to the 2.0 yet.

So it will be interesting to see, it seems like they might have just been like, we got to announce this. We got to say, here's all the things it does. Here's what the price point is basically because DeepSeek pretty much just sucked up all of the attention over the last three weeks in the LLM space. But Google's whole downplay of the DeepSeek was basically like, well, it's obvious that models are going to get cheaper.

Of course, you know, so here's ours now, you know, like, duh. So it's interesting. I haven't played with it. But I do think that, again, we're just going to start seeing that we're going to get models that are going get faster, cheaper and to do more. And intelligence seems to be going to zero is kind of what it seems like.

Sean Hise (05:03.984) Yeah, I don't know. Jacob, do you have any thoughts on the Gemini 2.0 release?

Jacob (05:07.918) I I've had it for a little while now. You losers need to pay the premium. It's great. mean, yeah, all these models are...

Brandon Corbin (05:12.374) Duh!

Brandon Corbin (05:16.956) What do you use it for?

Jacob (05:19.69) you know, I, I, well, first of all, I use the deep research quite a bit for like, I want to explore this topic and get me started. And if I'm trying to write something or, learn a in-depth topic, but the 2.0 flash is like, it is fast as hell. And, I oftentimes just use it honestly, it was like a benchmark of like against other models. So I'll do the same task and chat GBT as I do in flash and it's always faster, but you know, there's also some new models with, with the.

open AI that are available too so we can talk about those if you want.

Brandon Corbin (05:52.712) Mm-hmm. Mm-hmm.

Brandon Corbin (05:57.852) So when you're doing the comparisons with flash and some of the other models, what were you seeing? Were you seeing like, is it a better reasoning? Is there?

Sean Hise (05:59.868) You know?

Brandon Corbin (06:10.268) but we're good.

Jacob (06:10.286) What's up, Sean?

Sean Hise (06:11.854) Joe was just saying hi from the other office.

Jacob (06:13.71) I thought you were like raising your hand to speak because I was like, okay. No, you're good. Yeah, I don't know, man. I feel like Googles are always faster for sure. And I think like it's better at structured data. But honestly, they all kind of they're all like getting pretty damn good, right?

Brandon Corbin (06:17.38) Well.

Sean Hise (06:17.466) No, wasn't. Sorry for interrupting.

Jacob (06:41.346) They still have like the same problem as always. If you get a, give it a general problem, it'll give you a really generic response. That's kind of shitty. so with all of them, I'm finding the more context, the better, which I think we've all we've been talking about that for a while, but, I guess I'm just mostly impressed with how fast it is.

Brandon Corbin (07:00.091) Are you guys using it in or are you using it in any other capacity other than directly through either the Gemini website or through God like are using in cursor or are you using it in any other capacity?

Jacob (07:13.666) Now I'm still rocking a Claude and cursor, Sonnet 3.5, yep.

Brandon Corbin (07:16.858) Claude, I'd be curious to see how it would work with how well it would do.

Jacob (07:22.542) Is it even an option? Let me pull up cursor real quick since we've got our computers.

Brandon Corbin (07:27.643) There, I thought, I thought 1.5 is, I thought. Oh no, no, no, they do, Gemini 2.0, thinking experiences there. And cursor, sorry Sean.

Jacob (07:31.371) Is it?

Sean Hise (07:33.54) Hey, have you guys, is there dedicated, no.

Sean Hise (07:46.416) There is a, is there a dedicated app? It seems like there's a dedicated iPhone app for Gemini.

Jacob (07:52.876) Yeah. Yeah. I just got it yesterday. Actually, for some reason I didn't, I never had it.

Sean Hise (07:56.932) Here's an interesting stat and I'd love for anyone to fact check this but. to demand sage.com, Google Gemini gets 275 monthly visits. 275 million monthly visits. And each barred search query. Costs up to 0.006 cents to point to three cents for per query.

Do that math.

Brandon Corbin (08:23.493) So are they saying that that number, is that number the, every time that I do a search that Gemini gives me a search, an AI response, does that count?

Sean Hise (08:34.192) Yeah, I mean think about it if it costs three.

Brandon Corbin (08:36.09) then that wouldn't be that impressive. Like that number would, that number is not impressive at all, if that's the case.

Sean Hise (08:39.055) No.

Sean Hise (08:43.292) No, and it said under a million people have downloaded the they have not hit a million downloads a month on this app, but they're getting about 600,000 500 or 600,000. But it's just it's just interesting to talk about Gemini because you know, you'd think that like like first, for example, Apple had their earnings this week and chat GPT actually drove a lot of their services revenue because check the one of the smart things that Apple did was they're like build AI all you want.

We're gonna get 30%. They didn't give them a cut. They're getting 30 % of chat GPT's app revenue. you just don't hear a lot about Gemini, the user growth and the actual adoption of Gemini. It seems like this thing that Google's trying to force feed down people's throats. But it also seems like it's a competitive product. So it's one of those things where it's like everyone uses Google every day. You would think that people would be saying Gemini, Gemini. Is it?

Brandon Corbin (09:34.209) Right.

Sean Hise (09:41.286) To question, do you think it's just literally brand awareness and brand power at its finest with chat GPT? Or do you think there's something, you know, behind the curtains that we're not looking at? Like does Google.

Brandon Corbin (09:53.786) I think it's, it's both, right? So there's definitely the brand awareness that you have with chat GBT that people just assume the AI is chat GBT most the majority of the normies have no idea who anthropic or Claude is right that are out there. chat GBT is AI and Google like Gemini doesn't even exist other than they might they might have some sort of recognition that AI happens now when I do a search and I get this weird

loading animation thing that goes, and it dumps out some stuff sometimes. So I think there's definitely a brand issue, but then there's just also the that I wouldn't be surprised if people are necessarily not the most trustworthy of Google in terms of of being their AI provider.

Sean Hise (10:39.558) Well, like for example, so I use cursor every day. That's probably the way I interact with AI the most. So that's with just coding and that's embedded into the code, the IDE I use. mean, none of Gemini's models are even a default option on that for coding.

Brandon Corbin (10:56.973) Yeah.

Jacob (10:59.448) Let me ask you a question, Brandon. Do you have the most up to date cursor? Okay. I need to update mine because I, I don't see, I don't see Gemini's models even available to me.

Sean Hise (10:59.684) It's quad, EBT, A1.

Brandon Corbin (11:00.185) Ride it!

Sean Hise (11:10.104) No, they're not. They can be selected, but they're not in the default. Here.

Brandon Corbin (11:10.785) Well, so if you go and they're not in the default, so it's very possible that well, so but what's weird though is I never went in and enabled Gemini 2.0 Flash thinking, right? Like I know that for a fact because that's brand new. And so I just went to my option. I've got 03 mini now. I have Gemini 2.0 Flash thinking. I have Gemini Experience 1206. Who's, look, we can share screens.

Sean Hise (11:36.132) Yeah, so here are the models that you can install with cursor so you have Claude you have which is this is I did not edit this except for I did select Claude 3.5 sonnet the one with lots of numbers on it. That's the one I use. But Gemini, DeepSeq is right there. But Gemini is there, they don't even I'm wondering why would that not be a default option? I don't know, but it just seems like you look at.

Brandon Corbin (11:38.626) Hell yeah!

Brandon Corbin (11:51.427) Right. Right.

Brandon Corbin (12:03.681) It's gotta be a cost thing from cursor standpoint, right? They're obviously eating all of the costs of this inference. So I'm paying them 22 bucks a month, you're paying them 22 bucks a month. If we all go and start using 03mini, right? Maybe 03mini not, because I think 03mini is actually very affordable. But you know, anytime that you have any model that would be a higher price, they're not gonna necessarily wanna have everybody just sitting there selecting that. So, you my guess is it's

They probably are continually trying to adjust which models people are using, trying to perfect their own models to eventually not have to basically be able to pay the tax of using these different models. I don't know, it's fascinating. I will say though, dude, Claude 3.5 and OpenAI for the last three, four days have just been stupid as hell. In cursor, it's been ridiculous. The amount of arguments that I've had with Claude over the last three days.

is elevated more so than just like me losing my marbles. Like it seems like it's really bad right now. And you saw it, Sean, when we were doing that GPT I generated to generate like the show notes for our things, like there's misspellings throughout the whole friggin' thing. It's just crazy. Like I don't know what the hell is going on.

Jacob (13:20.718) Dude, yesterday I tried to build what I thought was a pretty straightforward project. like, all I wanted to do was I said, go to, I want you to use JavaScript and node to go to this webpage. Here's my login credentials, log in, and then I want to scrape this page. can't, that's not that complicated. And I spent like 30 minutes arguing with it cause I kept getting some puppeteer error and like, I know we've struggled with puppeteer in the past, like with different, you know, I've got a.

I've got an M2 or whatever and there's, there's definitely some issues with that, but it wouldn't try anything different after I threw it the air. I was like, here's the problem, fix it. And it just kept telling me the same thing. And I was like, okay, this is useless. Might as well go watch a tutorial and actually do it myself.

Brandon Corbin (14:01.163) Yeah.

Brandon Corbin (14:07.608) I'll write this myself.

Sean Hise (14:08.636) Well, so the Gemini mystery continues. They obviously have what they call reasoning. I'm still not using it. The stock market's not responding to anything they're doing, but I do think that for those that use Google products, it's still a really good option. They're keeping up. They're spending $75 billion, but now, instead of the market responding when you say you're gonna spend $75 billion on AI,

They're not going, oh great, they're going, oh, do you need to spend that much? Maybe that's actually not a good use of funds. But with that said, let's move on to something that happened this week that I think impressed one of the most impressive people. I'm gonna throw this on, I'm gonna share my screen right now. We've got Dharmesh, the CTO and founder of, well, I don't know if he's the CTO anymore, of HubSpot, and he was talking about the OpenAI new deep research feature.

Brandon Corbin (14:41.503) Right.

Sean Hise (15:03.674) So I think we're gonna move on to that real quick. First of all, has anyone used the OpenAI's new deep research feature?

Jacob (15:08.558) It is not available to me as a poultry. I only pay 20 bucks a month. I gotta pay 200 bucks.

Brandon Corbin (15:13.846) P.ON

Sean Hise (15:15.194) Yeah, yeah, have to. yes, this is going back to Altman's really, really research based idea of how to charge for their pro account. So with open eyes and chat, GPT's pro account, pay $200 a month and you get access to more stuff. One of those things is their deep research. Which is the video.

Brandon Corbin (15:19.179) Ha ha!

Brandon Corbin (15:31.991) Unlimited Sora too, apparently. You get unlimited Sora, which might not be that bad. Like, he just did a tweet, sorry I decided to lie on us here, I do that, but on a, he just did a tweet where Sam was like, you know, originally we thought that the only, people would do the deep research or 200 bucks a month, but then we give you unlimited O3, we give you unlimited Sora, we give you this and this and that, it ultimately ends up turning out to be a fairly good deal.

And he's not necessarily wrong if if really the source stuff could start producing some like long format and like, you know, like what the Pica we didn't even talk about that, which we probably should is Pica.art's new one You know if they can do 1080p and like 15 30 second long videos that becomes really not a bad price point if I can just generate, know continually from you know off for 200 bucks a

Sean Hise (16:24.092) So basically this guy who's really well respected said that this feature is a game changer for him. Disclosure at the very end, he's an investor in OpenAI, but I thought that there was a lot of discussion that kicked off in the internet about these soft skill consulting type jobs and how they're irrelevant based off the fact that you can create truly in-depth research. My question is where does it go from

when it's, I'm just asking for answers about something to actual research. Like what defines that? How does it cross the line there? Is it going out and trying to collect new information? You know what I mean? What is actually happening there that qualifies this as research?

Jacob (17:08.366) Yeah, I know that Google's deep research goes out and it'll search like 50 or 100 actual web pages and collect all that information and then distill or give you whatever the question you're asking, try to answer it from that information plus the model, I think. I'm assuming OpenAI's model is working the same, but Brandon, you might have some other thoughts on that.

Brandon Corbin (17:30.048) No, I would assume the same thing, basically. Now, again, though, that the fact that, you know, flash can sit there and consume or Gemini flash can consume, you know, know, foul to 20,000 page PDFs. I mean, I would be more inclined to trust, you know, a deep research coming from Google than I would potentially open AI. Right. Like they're really going for being able to consume just a

crazy amount of information in one shot. And I think that that's, I think that that's the right approach, because that's where we ultimately need to get like, I can just build something, throw in a ginormous document, and then ask anything about it. That seems to be the most relevant. But yeah, I don't know, I don't do any of I haven't used either of them. I mean, I use a one and I use three for like doing some deep reasoning and like trying to do but I haven't done anything in terms of like that deep research.

Sean Hise (18:18.106) No, it says so.

Brandon Corbin (18:27.121) Apparently, OpenAI also offers one like if you're willing to wait 24 hours, you can do batch. Did you guys even know this you can do like batch processing and that then they'll just return it like 24 hours later when it's done and like super cheap. I don't know if the reasoning is available on that model too.

Jacob (18:43.362) You think they're just sending that to some someone overseas and they're like, Hey, well, what was that one? What was that one company that did that? I actually worked for it. like you'd ask it cha cha. It's like cha. Yeah. Yeah. Yeah. I mean, not, no, I didn't work at cha cha. I did like the, I was an answer or an answer of questions. So I would Google shit for people.

Brandon Corbin (18:46.613) Probably.

Brandon Corbin (18:56.508) yeah, Cha Cha. Right. You worked at Cha Cha?

Brandon Corbin (19:05.141) okay. All right. I got so I interviewed at Cha Cha and this was the first time that I met Doug Rao. If you guys don't know Doug at Profile Tracker, Doug and I go way back now and that was at when I interviewed there. I got into an argument with the I was coming on as a UX potential UX role and I got into an argument with the developers on how you should actually be building it. And the whole interview just went to a fucking argument. And then at the end

Doug was like, dude, what the fuck? And I'm like, sorry. That's when he realized I'm the most unhireable person in Indiana. He's just like, ugh, this is ridiculous, Woo, cha cha.

Sean Hise (19:41.37) You are, there's no doubt. would never, I don't hire this guy. So according to OpenAI, how it works is somewhat vague, but it says, deep research was trained using end-to-end reinforcement learning on hard browsing and reasoning tasks across a range of domains. Through that training, it learned to plan and execute a multi-step trajectory to find the data it needs.

Jacob (19:47.31) Ha ha.

Sean Hise (20:09.232) backtracking and reacting to real-time information where necessary. The model's also able to browse over user-uploaded files, plot and iterate graphs using the Python tool, embed both generated graphs and images from websites and its responses and site-specific sentences or patches from its sources. So it sounds like it's a mix between its browsing functionality, additional model training, and the Python library that they've been using that basically.

makes pretty graphs and turns them into PDFable content. And I think that is part of the chat GPT and all these interfaces is the ability not just to return chat based responses, but to actually take data, visualize it, and do certain things with it that's just beyond giving you some text.

Brandon Corbin (20:40.435) Right.

Jacob (20:57.568) Yeah, that reminds me the other day I asked a question and like it was some math question and it went and like actually did the Python inline and I forget what it was but I was pretty impressed. Like to your point Sean like these things are getting more powerful. You used to ask it like what's two plus two and it just have no fucking idea how to respond to that and now it's like well I know how to write I know how to write a

Brandon Corbin (21:21.013) Mm-hmm.

Jacob (21:23.426) function and, and, that's a simple example, but I noticed how to do math using Python now and it'll just show you the formula. And I'm like, cool. It was some, it was some like profit calculation function that I wanted to do over time. And I was like, damn, that was, that was pretty sick. But I mean, I'm impressed with the continued improvement of like inline tooling and, features outside of just chat.

Sean Hise (21:46.712) I want to share another tweet since we're doing I like this. This screen sharing thing as the host I get to do things and throw things. So I've got another tweet that I want you guys to react to. So this was one that I loved. So I've been posting every time anybody posts about opening I lately I've been shitting on them because of what Altman did and how much hole this guy has that and so this is this is from an engineer at X and it says and it's from February 2nd. So that's one.

Brandon Corbin (21:52.261) Yeah, did too, it's great.

Sean Hise (22:14.96) bajillion years ago, February 2nd, today's February 6th. So it got 1 million views. And it's an engineer at X, which is Twitter for those that actually I don't call it X. Please don't do that. My dad calls it that it kills me. But he says, don't he I think it's a she I don't know who it is. Do not invest in OpenAI. Do not sign any agreements with them. Do not do any business with them. I've been busy all weekend and just got plugged in.

Brandon Corbin (22:32.274) No!

Sean Hise (22:43.77) I've seen enough things across 4chan, my x-feed, and my own experiments that confirmed it's over. There is no moat.

Sean Hise (22:53.708) I agree. I agree. The model is not the moat.

Brandon Corbin (22:57.126) So, okay.

So the model is not the mode. However, as we've talked a thousand times over and over and over again, not raw dog in the model anymore with OpenAI. There's an entire application layer that sits on top and they want a trillion fucking dollars to basically make it so they could continue to take on more more and more. Like if you've actually tried to use DeepSeq at any level, you'll see that it's hard to keep that shit up and running where they're like, sorry, we're at capacity. sorry, we're

Sean Hise (23:13.274) No.

Brandon Corbin (23:32.814) Like it's hard to make these things kind of keep running and doing it so they need all this money to kind of do it But there's a whole application level layer that I think that that's where they're gonna the battle is really going to go and Grok like that person sitting there with Grok sitting there being like hey Write me some Nazi poems, you know and Grok will do it for you

Sean Hise (23:44.22) That's where the battles run. That person is with Rock. Couple things. Couple things on that. One thing is this person may or may not understand game theory, right? Which is the first mover advantage, right? There is a huge value in their ginormous user base. We're sitting here trying to figure out why people don't use Gemini and it's embedded in Google products. Another thing is I think there's a lot of press

and attention that gets put to the models. But that doesn't necessarily mean that's exclusively where the company's focus is. That's where the attention goes to is, the model, the accuracy of the model, the new model, the this. But I agree that there isn't a moat in the model because you're seeing people just steal these things and retrain them and just basically siphon all the work and just layer on top of it and release it as their own. But I thought that...

I think that it's a really important consideration. And I think that it just points to the fact that how much longer are we going to be focusing on the models themselves and really moving on to more applications of it. Which brings me to my real real something that that Brandon brought up that I thought really could create a moat, which is opening eyes filing for these trademarks. So I don't know that this was this week.

Brandon Corbin (25:11.815) Yeah.

Sean Hise (25:13.776) But they're going out and one thing is that they did and it was something that you kind of haven't heard much about. But Johnny Ive, so for those that don't know Johnny Ive, Johnny Ive was Steve Jobs designer for 20 years, Or I don't even know how long.

Brandon Corbin (25:32.153) Yeah, he's a sir. I mean, like the dude is literally like one of the best industrial designers that we have had in the last. I mean, yeah, in the last 30, 40 years.

Sean Hise (25:45.722) Yeah, and he started, so he quit Apple and he started an agency called LoveFrom. And it's a small, I mean, it's probably a small agency that's already doing $100 million a year in revenue, whatever. But they contracted with Opening Eye. And so one of the things that Opening Eye has in their pocket is this guy, right? And so what you're gonna see, I think, and potentially, and that's what Altman and they're trying to do, they're trying to get out ahead of all this stuff because they're asking for billions and billions and billions of dollars and they're losing billions of dollars.

But I think it's an interesting thing that we're that we are totally short that I'm totally like counting out is the fact that this guy's in their corner and they're going to start coming out with products. They're going to start coming out with more. So you're talking about brand power. If you're a brand person, right, you're sitting there tooting your own horn going look at these guys. They've got product that's not any better than than this this or that or something from a really another really powerful brand.

and they're bringing in even more brand power and even more product power. So now the pendulum swings back the other way. Is opening eye gonna dominate because of all this stuff, right? First mover advantage, product design, user base, you know? And so, you know, we sit there and crap on them because they went and asked, you know, the Saudis for 500 billion and they want to control these data center builds out, but build outs, but maybe there's other things going on that, you know, and now we start to shortchange Altman a little bit.

Brandon Corbin (26:49.871) Right.

Brandon Corbin (27:09.007) Right. So some of the key things that they call out during these for the the trademarks are headphones, goggles, glasses, laptops, mobile cases, smartphones, smart jewelry, VR headsets, right? So like they're just spreading across the entire spectrum of potential products where you would be interacting with AI. So if they can kind of

take that position now and be like, we're gonna be that next hardware play. Because again, we all have to admit that Apple hasn't created anything new since Steve Jobs died, right? It's all just been plays off of what they've currently done. Again, props to Cook for turning it into an $18 trillion company or whatever the hell it is right now, right? Like he made it mad, but the creativity is gone and Ives is gone.

And jobs is gone. now jobs is over here kind of working on like what's going to be the next human interaction of technology with AI. It could be that could be the remote. mean, again, what everybody likes to say that Apple just waits and then copies what other companies do. Right. What people don't really realize and think about is that Apple created the entire fucking iPhone market. They created the phone market.

They created all the design, they created the design patterns, they created the physical hardware, they created everything that everybody copied from. And that was all frigging eyes. The guy's a genius. And now he has the ability to kind of take it and to reimagine what this actually looks like in today's age with AI backing everything. I do not bet against Johnny. That's it.

Sean Hise (29:00.324) I don't either and I do think if there's any sign of. Of a product that could have the ability to compete with with an Android device or even an iPhone. I mean you're talking about. You know years and years and years of expert engineering that goes into the iPhone. You can't even can never discount that. However, if you move on from the user interface as a.

as the interface as the as the interface of choice right and you move to it a voice based interface right. You know you just look just the first place you need to look is opening eyes app on iPhone. I the I mean it blows Syria out of the water. And in a a world where you know they don't even have to service and one of the problems with tech companies is that they have to.

Brandon Corbin (29:43.343) All right. All right.

Sean Hise (29:55.96) service all the things that they've had. Apple can't come out with a screenless phone, right? Or an audio-only device, you know what I mean? It almost doesn't even incentivize them, right? Because what's gonna happen? Well, their services revenue could get cannibalized through in-app purchases. They're de-incentivized to fundamentally change how people use computing devices because they're pushing out billions and billions and billions of iPhones every year.

If opening eye is to have an advantage moving forward, I would put it more in the hardware side, the robotics, the interfacing, not necessarily the models, the applications, and maybe we're thinking about it wrong. We talk about, you know, the applications of the models being like, create agents and, you know, integrate software platforms together. Maybe they're more, maybe their competitive advantages may be creating a more general computing device that utilizes their.

Brandon Corbin (30:34.968) Right.

Brandon Corbin (30:54.41) Right. Are you?

Sean Hise (30:55.866) their large model.

Brandon Corbin (30:57.644) Are you guys using any of the the open AI tasks thing yet? Have you set any of those up?

Jacob (31:04.438) No I haven't.

Sean Hise (31:05.37) No, I set a birthday reminder for my mom after I deleted Facebook.

Brandon Corbin (31:08.622) So that's actually an app that I'm considering building, by the way, which will sideline that. So I set one up, which was just basically, I want you to find something that is a horrible design, a horrible user experience, bubble it up, and then every day send me an email that basically says, hey, this design experience sucks. We need to have a better experience for it. And I just set it up just kind of playing around to see what would happen. And it's actually awesome. Like every morning now, I get an email and it's like,

Hey, gas station pumps UI is awful. And here's how you can try to think through it, right? And so like every morning I wake up and I go through this design thinking process where I'm like, yeah, how would I solve that problem? How would I solve that problem?

Sean Hise (31:44.07) That's amazing.

Jacob (31:44.856) Dude.

Sean Hise (31:51.078) That's a great way to get the dopamine going for an ADHD more like you.

Brandon Corbin (31:54.072) Totally dude, it's amazing. It's so fucking awesome. I'm loving it. So that's that is a really cool feature I am like him that I didn't think I would

Jacob (32:03.97) Yeah, I guess I never thought I would be so angry at a gas station pump. I feel like, you know, like the Apple pay doesn't work half the time. Like sometimes the card reader doesn't work, have to go inside and I'm like, do I, do I pay you? Do I just steal the gas? What do I do here?

Brandon Corbin (32:20.878) Or doors so so I just sorry

Sean Hise (32:21.756) So you mentioned that it can email you. How does it email you?

Brandon Corbin (32:26.862) So every time it runs, I think that that's just how it does it. That's how it notifies you is that it ran, is that it just emails you, and then it's like, hey, here's what it is. And you click that link. Now what's interesting is it takes you back to that same chat that you had previously. So when you create it, you actually select the model is going to be 4.0 with tasks or something like that. And you select that model, and then you say, hey.

I want to have every morning an email that basically does or I want to have every morning a report that does this and then it's like, all right, I'm going to email you. And so it just basically says, Hey, this has been generated. It gives you a tight little title. You click that it takes you back to the chat. And so now my chat just is like keeps getting longer and longer and longer. But yeah, so I don't I don't know like long term like what happens after you know, three, four, five, 10 years of this, like is this chat can just be a blow up or whatever.

Jacob (33:20.44) Dude, I would love, so to your point earlier Sean about reminders, so I sent out a save the date for an engagement or for my wedding and basically the app I put it into to collect everybody's information is cool because it's like trying to get me to buy more product through them. So sending more cards and all that. So it'll be like, hey Brandon's birthday is coming up in two weeks, do want to send him a card? Right?

Brandon Corbin (33:20.641) But it is, it's a cool feature.

Brandon Corbin (33:42.401) Ha ha ha!

Jacob (33:48.78) Well, I already have everybody's birthday in Facebook, right? Or in whatever, my phone. what if I could, what if there was a device that I could be like, hey, go look at my Facebook scrape everybody's birthday that I have in my phone that I've actually contacted in the last two years. And then, and then, yeah, but I, so like, these are the things that are more useful to me. Like Facebook's great for, like it has all my repository of information.

but I don't want to go there every day. And also, am I going to be looking three weeks out of who's coming up?

Brandon Corbin (34:20.75) Dude, this is it. This is what you should have, like if you would have waited just like two weeks.

Sean Hise (34:23.136) Jacob, just said it and Apple just went and built it.

Jacob (34:27.982) no shit, this was on the fourth

Brandon Corbin (34:28.685) Dude, yes.

Sean Hise (34:29.924) Yes, dude. Yes, so two days ago, this actually popped their stock. I was watching it. So Apple invites came out.

Brandon Corbin (34:37.357) Which is wild, it's a fucking application dude. It's an app. They built an app. They built an app and it pumped their stock.

Jacob (34:45.41) Heh

Sean Hise (34:46.812) So I think it's actually going, I think it's actually, mean, this is one of the things that Apple's doing and I think that this is where Cook is smart, which is, you know what, we have this ecosystem and no one can keep up with us. Just keep ripping industries and throwing them into our ecosystem. Now is there a threat to that? Yes, if people stop using iPhones. But like, you look at what they're doing from a health perspective, right? I mean, if I'm Eventbrite or any sort of, you're done.

Brandon Corbin (35:16.841) yeah, you're shitting

Sean Hise (35:18.49) You're done. You're done because what Jacob just literally described a lot of the problems something like this would solve like your people, you know what I mean? Like think if you if you sent out like, you know what I mean? Like you like just think about all the things that this integrates into your life and they just boom, they just drop the app and they do it.

Jacob (35:34.872) You and you talk about moat dude, like I would never fucking leave Apple if I had all these like reminders and like and what eventually needs to happen is the context I need to be able to say hey, whatever, you know, you know your name I Might my anniversary is coming up in two weeks and last year I did this Sent me a reminder to make sure or whatever the reminder I know you can do some of those things now but like more context and conversational stuff like that where it's like hey you might think about a

booking a hotel as well or whatever,

Sean Hise (36:05.232) By the way, they snuck in something that I thought was really interesting, which is iCloud Plus. So iCloud Plus Premium, three features that they released is some, really, I don't know if you guys have been watching TV the last couple of days, but there's tons of Apple security email. They are doing a marketing campaign around privacy. That is happening. I mean, every commercial that they're dropping right now has to do with privacy. And there's a huge Safari push.

And I don't quite know why they're doing it, but it has something to do with, think, trying to get people to use iCloud Premium. So like, for example, you can, yeah, they just released all these new features, but they expanded their family sharing. You can do custom email domains now. You can review, if you're using HomeKit, you can review your footage from your house, like, in a totally encrypted end-to-end thing. But like, hide my email, you can generate random unique email addresses whenever you need to use them.

And then this private relay thing is what they've been talking about even on their commercials. So I don't know what this is all about, but it just seems like they're seeing the fact that their services business is growing so much and they're just leaning into it. And maybe it's a reflection of the fact that they realize that there's only so far they can go with the hardware iterations now, like from a growth perspective and they're trying to move the needle.

But I don't mind a company, I don't need my iPhone to be that much faster, let's be honest. I don't need the camera to be any better. Do I want the ecosystem to be healthier? Do I want Swift to be better from a developer experience? Do I the core Apple apps to be better? So it's kind of an interesting dynamic where Apple's at the peak of their technology prowess and their ability to build awesome products. And then you have kind of,

going back to the punchline is, is it gonna end up all for not because somebody like Open Your Eyes is gonna work with Johnny Ive and build a screenless, device-less, you know, I think you might be muted, Brandon.

Brandon Corbin (38:20.331) We go alright I'm back sorry. So here's one thing that's kind of interesting about the the and I've said this for a long time. The best user interface is no user interface right as is as something I've just said for a long friggin time that that if there needs to be a user interface if you can figure out a way to get rid of the user interface and still solidify and still deliver the you know the solution then that's awesome. I think though that when we come to this product.

Sean Hise (38:21.35) Sorry.

Brandon Corbin (38:50.516) this product standpoint, it's going to be a very interesting, we had the humane pen, right, which was trying to be like this kind of like no UI thing. We had the the rabbit R1. Now again, those were those would all have been the old flip phones of the yesteryear, where Apple was sitting around being like, these things are dog shit, we need to make a better phone. Right? That's what's happening there. So again, I can't I can't even pretend to think of like what this might be. But with Ives, you just you just hope then again, he did come up with a hockey puck.

The the mouse hockey puck if you guys have ever used that and that was a horrible product So he isn't God and he does have bad production that he does do however Yeah, yeah, yeah, yeah, yeah for sure

Sean Hise (39:29.744) Well, he's a he's a minimalist. He's a minimalist.

Jacob (39:33.024) Yeah. I've said it before, but I think I'll reiterate. I think that, wearables wearables are the next generation of like things like, cause we want data on ourselves. We want to be able to be like, so I have a whoop. I'm a huge, huge advocate of like, I can test how, if I drank yesterday, you know, how, shitty do I feel today? or whatever. No, but like, I think like capturing video capturing like your day.

and what's going on is a huge opportunity. And maybe Apple doesn't make every single one of those or open AI doesn't make every single one of those, but there's gotta be a common language. That's why Android and iOS is like really the bedrock. Like they're gonna own all that space and you're gonna interface with them, right? Like my whoops sends heartbeat data to my iPhone, right? So, and.

Brandon Corbin (40:16.199) You

Brandon Corbin (40:25.756) Right, right.

Sean Hise (40:26.928) Did you guys see the, speaking of third party peripherals for that, integrate with that? Because whoop has, do they integrate with the health kit? Is there some sort of actual Apple? Okay. So they allowed that product to integrate into what they do. Okay. Well, Zuckerberg, when he went to Rogan.

Jacob (40:35.533) Yeah.

Brandon Corbin (40:36.808) Yeah.

Jacob (40:41.112) Right.

Brandon Corbin (40:43.721) Well, Apple just opens that up. HealthKit, can just integrate HealthKit right into your Swift. Yeah.

Sean Hise (40:47.996) It's an API. Well, Zuckerberg was on Rogan complaining about Apple. I don't know if you saw the whole thing. But one of the things he talked about was how all they do is squeeze people and that they haven't innovated. And they kind of echoed what Brandon said, which I don't totally agree with, but I can understand where they're coming from. But it's kind of the pot calling the kettle black with meta, in my opinion.

Brandon Corbin (41:10.279) Yeah, what the hell is Zuckerberg built in the last five years? Ten years.

Sean Hise (41:13.932) Yeah, the metaverse that they're literal, they're CFO or whatever is like, we're gonna have to like shut this down at some point. But he was talking about AirPods. And he was he was just hitting on AirPods. He's Did you see this? He basically was like this AirPods are perfect examples, the AirPods you you put them in your ear and they connect, you know, it's integrated. It's it's it's it's unlike any Bluetooth experience that you that you'd ever have. And they purposefully they built their own protocol.

So there's this sub protocol that AirPods uses that they don't allow any other company to leverage. So there's a protocol for AirPods. if you, he, so he went to, to Apple and was like, hey, can we use this protocol for our devices? Because we'd like to have a better experience for iPhone. And they were like, no, no, you can't. Well, Zuckerberg spent a couple minutes talking about how dog shit the protocol was, how it wasn't secure.

Brandon Corbin (41:44.094) Mm-hmm.

Sean Hise (42:11.376) how it was like not even encrypted and like, was like, this is just a bad product and the only reason why it's any good or people use it is because Apple won't, it's exclusive and they're doing monopolistic tactics. And it's kind of interesting how big of a moat Apple has with that kind of stuff. you know.

I don't necessarily think that what he was talking about was true because you have things like health kit that are allowing people to build third party devices that integrate and all that. And maybe the AirPods was just like, you know, a side project. I don't know.

Brandon Corbin (42:45.897) Yeah, so Apple has innovated. There's no question Apple has innovated but Apple's innovation are all still built off of the backs of what was created 10 years ago. has, mean, Vision Pro, you can make an argument that that was like a truly kind of, truly revolutionary kind of, right? Like we knew that that was a direction that was going, but it wasn't a successful product. You know, like when, but when we, you know, if you think back to early 2000s, man, we had

We had the iPhone, we had the iPod, we had the iPad. had the hell, had the Mac Mini, the Bondi Blue Mac Mini, right? All the way from the beginning of when Jobs came back, and it was just transformational. And it was just year over year over year of just transformation, transformation, transformation. And now we're just to, what kind of software can we build on top of it? We're gonna build an invites app. That's the fucking innovation?

I mean, again, no problem. The invite app is cool. However, you have to be an iOS user to be able to create the invites. You can accept the invites and be participant if you're on an Android, but don't pretend for a second that you're creating a fucking invite if you're on an Android body. Only people who are having parties are on iPhones.

Sean Hise (44:05.612) They, people that, the Android people aren't real humans. They're actually, they're a lower class of human.

Brandon Corbin (44:12.69) Their birds aren't real. The Android people aren't real. Like they're all fucking NPCs.

Sean Hise (44:20.284) Speaking of the Big Cheese AI stock portfolio, we have an absolute rip roaring stock announcement today. Palantir hit what, $108 a share today?

Brandon Corbin (44:35.322) Who's invested in it? Not me. Sean is? Sean's the only one.

Sean Hise (44:37.589) I am. Hey, come on baby. I told you guys to buy it at 75.

Jacob (44:37.841) on.

Brandon Corbin (44:44.878) Did you tell us to buy it at 75?

Sean Hise (44:46.712) I did. Actually, I have just said that I did.

Brandon Corbin (44:48.658) Dude, I'll start listening to you this time.

Sean Hise (44:52.538) Yeah, you need to. That's 35%.

Jacob (44:53.506) Hey, I'm in on Cloudflare. They're up a little bit over the last couple months here. Yeah, they're up 56 % the last three months, but I am a huge user of their products.

Brandon Corbin (45:08.818) Hey, speaking of this, this will be a good conversation for AI, is that you and I talked about Cloudflare tunnels, and you're like, yeah, it's simple. You just do this, this, this, and this, right? And I'm like, all right, fucking Cloudflare tunnels. So I decided to have AI do my Cloudflare tunnel, and motherfucker, I went through two days of screaming at AI to get it, and eventually I was like, I'm done. I'm not gonna have AI do this because you're stupid as shit. You obviously can't figure this out.

and I was able to do it myself in about 15 minutes.

Sean Hise (45:39.654) Well, here's the problem that I think and I already know what your problem was. How long has Cloud for Tunnels been out?

Brandon Corbin (45:46.364) Doesn't matter! That's my point. That's my whole argument, right? Like everybody's like... Yeah.

Sean Hise (45:49.83) But I have the same problem right now, which is technology. Okay. Here's a, here's a conundrum. Okay. AI scrapes the world's internet. AI creates model. Model creates people interact with model. Model updates, new model comes out. Okay. Real time web scraping and all this research stuff, expensive. Right. And so like every, but what does AI also do? It accelerates development.

Brandon Corbin (46:12.668) Yeah.

Sean Hise (46:17.968) And what do you need when you come out with development products like Expo, Versel, CloudFlare tunnels? You need documentation. You also need people to adopt the usage of the product so that there can be Reddit posts and there can be things consumed on the internet that get sucked in by the models. And so there's this latency that's happening with newer technology that's coming out where I'm going and I'm asking questions about, I'm like, hey, build this. And the AI is like, I don't.

Brandon Corbin (46:26.76) the latest.

Sean Hise (46:46.608) know anything about what you're talking about. Because I'm not necessarily as fully up to date on this as I could. You know what I mean?

Jacob (46:51.244) No, it's like, it's worse. It's worse than that. It'll, it'll go build it, wrong, but it, but it doesn't even tell you it doesn't know shit. It's like, we go. Here's this thing that's definitely not going to work, but looks okay. pisses me off.

Sean Hise (46:57.116) it'll yeah

Brandon Corbin (46:57.35) Eww eww eww eww eww eww

Sean Hise (47:06.396) Yeah, and so I think that there's there's gotta be a and maybe maybe there's there's a way to do this, but like it would be great if cursor or other platforms that are where you're doing generative AI. You could say hey, I want you to use this model, but I need you to actually like fully suck in this entire documentation set as well. Maybe that's not even enough.

Brandon Corbin (47:20.218) Yeah, so...

Jacob (47:24.631) I was-

Brandon Corbin (47:24.644) Yes, it exists. It's your cursor rules file, right? So .cursor rules. Now you can actually search for .cursor rules files. There's a whole website that's out there that basically says, hey, I'm building a Swift application. Here's a cursor rules file for Swift. And it's like the latest version of Swift. Make sure you do this. I've actually been building them out as I've been working on these Swift applications. Like every time it...

Sean Hise (47:48.54) can you really dump the whole doc set? What if it's 100 pages of docs?

Brandon Corbin (47:52.455) Hell yeah, baby, use Flash! Gemini Flash 2.0 apparently is gonna be able to consume all of it. The documentation though, if you think about like the documentation between this old version and this new version, the migration documents, are usually not that thick, right? They're all either, I mean, you maybe, Right.

Sean Hise (48:10.726) For something like that makes sense, for, know, for, and maybe it makes sense for something like, you know, for seller expo. But yeah, I mean, there's definitely this, this issue where I think a lot of people think that AI is going to be great and it's going to write other code for them. But like they're also, you get to the point where you're like, where you're like us, which we use all the best and breed stuff. And those people are constantly coming out with new versions and the AI models just aren't ready and aren't as up to date on that.

Brandon Corbin (48:25.08) out of. Yeah.

Right.

Sean Hise (48:39.696) Think about that, that's gotta be the same in the medical field, right? It's gotta be the same.

Jacob (48:42.51) Look at this. Look at this shit. So, I'm so skeptical. Can you guys see my screen? I'm so skeptical of this shit, Brandon. Like, okay, so this might not be the best example. Blow it up, all right.

Sean Hise (48:47.568) Yeah, yeah.

Brandon Corbin (48:48.324) Mm-hmm. Mm-hmm.

Blow that up, would ya? Can you increase that for my shitty-ass vision? Perfect, thank you. Hey, you dickhead.

Sean Hise (48:57.279) my God, you can't see that, you old.

Jacob (48:59.82) Okay. So like, okay. So I agree with what Sean was saying. It'd be nice to, if I could embed, documentation and like, cause we use direct this and swift, they all have really good docs, right? What would be really nice is if you could just, and maybe you can do this and I'm just ignorant to that, but like just embed the docs and then you could make your rule sets or whatever. But like this kind of stuff I am so skeptical of, like

Brandon Corbin (49:14.158) Right. Right.

Jacob (49:26.732) Where does it say, you will always use the latest version of Swift UI and Swift.

Brandon Corbin (49:31.408) Yeah, yeah, yeah, yeah, yeah, so that's bullshit. That's that's stupid. There's like it has no fucking concept of what the latest version of Swift is. So yeah, yeah, that's stupid. Yeah.

Jacob (49:37.154) Yeah. Yeah. Well that, I see that shit on like LinkedIn all the time where it's like people are, people are like, yeah.

Brandon Corbin (49:42.758) 100%. Don't make things up.

Sean Hise (49:43.612) Does adding to a prompt, does adding a layer on top of a prompt, is that's not the same as training a model on that data, right?

Jacob (49:53.783) Right.

Brandon Corbin (49:55.522) No, no, yeah, it's not like you're trying to do a reinforced learning model based on this thing. So, but again, like the key, so for me, when I first started, it was with Svelte 5, right? All the AI models are trained on Svelte 4. Svelte 5 was a pretty dramatic shift. So, I pretty much just ended up going and taking the migration documents from 4 to 5, making that into, copy and paste that into the cursor rules, and that's fine. The other thing that you can do,

Sean Hise (49:59.526) No.

Brandon Corbin (50:23.394) So here's a very important thing, is as you're going through, you're gonna find you get these weird bugs in cursor. What I want you to do is when you copy and you paste the error log into cursor, I want you to say, I want you to actually think through why the fuck this happened, and if it's something that's worthy to remember, go update the cursor rules file, and it will do it.

Sean Hise (50:29.67) Mm-hmm.

Sean Hise (50:46.15) Will it do that?

Jacob (50:46.432) shit.

Brandon Corbin (50:47.672) yeah, yeah, yeah. So then it will go and be like, shit, yeah, this is actually something that I'm going to do over and over and over again. And like, there's, you know, different ways of doing I decided to try to build a game right with with felt and like all these vector ways that you like interact between 17 and 18 are totally different. And so then it kind of slowly started building. And my cursor file got pretty big of like, here's all the things that are relevant to this specific product, because like a cursor file or cursor rules file for a game would be different than an app.

Right? Like, because you're using like scene kit in here and you're using like, you know, the different kits. You actually, so you create the cursor rules file exactly like you would like a gitignore. So it's just a dot cursor rules and anything that's in that dot cursor rules will automatically get added to the context every time that you're chatting with it. So it just automatically injects straight into it.

Sean Hise (51:23.42) So where's the cursor rules? I'm sharing my screen on my current cursor problem.

Jacob (51:32.419) Hmm.

Brandon Corbin (51:45.314) And that's basically, hey, this is my application. Here's so what my cursor rules end up using look like. Here's my directory structure. Here's the things that I want. Here's like high level requirements. It should. And if not, then sometimes like if it's not behaving, sometimes I'll even pull that damn cursor rules file into my context. The other thing that I'll do a lot now too is I'll include markdown files for every feature and I'll like go through and I'll like write out like this feature is supposed to do this.

Sean Hise (51:53.168) Can it work?

Brandon Corbin (52:14.244) Cause that thing loses its fucking mind every once in a while and just like rewrites complete like functional requirements that we have spent two hours working on.

Sean Hise (52:22.556) By the way, I've been trying to get this, I'm building a native app without writing any code and I cannot get cursor to take photos in the right aspect ratio. So this is my code, right? And it's just React Native Project. But basically, for those who've never seen cursor, you ask it questions and then it'll give you answers and it'll actually go through your file and suggest exactly what it wants to change. And you can just accept it line by line or you can accept.

Brandon Corbin (52:28.59) Hell yeah!

Sean Hise (52:51.566) all of it or you can reject it. I know we've never demoed that, but.

Brandon Corbin (52:55.458) Are you guys using, are you guys doing, has anybody done windsurf, windsurfer?

Sean Hise (52:59.714) I have not. I have not.

Jacob (53:00.012) I saw someone recommend that the other day.

Brandon Corbin (53:02.028) I need, yeah, I'm gonna try to play around with it this weekend just to see, cause I hear a lot of people talking about it.

Sean Hise (53:10.34) But this app that I'm building is 100 % a situation where, and I think you have to recognize that as a developer, like Cursor's not gonna answer this problem. It doesn't have enough information. Like this, there's not enough training on it, not many people have asked that question. It's a very platform-specific plug-in question, and it just doesn't have the answer. And I think maybe that's where you go ask a couple different models, or just go to the docs and read it yourself.

Brandon Corbin (53:19.811) Mm-hmm.

Brandon Corbin (53:36.567) Start the ride at yourself, bitch!

Jacob (53:38.574) Did you guys get my did you guys see my meme that I sent you yesterday? Let me pull that shit up. I thought it was hilarious. also, I Just saw that Google flash 2.0 is is a benchmarking slightly better just slightly but slightly better than Claude saw in it 3.5 so

Brandon Corbin (53:42.947) Yeah, bring it, bring it, bring it.

Sean Hise (54:00.4) I'm gonna try, I'm gonna try, so I'm gonna take the next week. I'm gonna switch cursor to Gemini. Flash. I'm gonna try it.

Brandon Corbin (54:06.115) Yeah, yep, let's see how it works.

Jacob (54:08.066) Yeah, give it a try. I'm just, you're not gonna be able to see this. Let's see if it'll focus in, but basically, basically someone was like, and I've actually said something very similar, so I'm an idiot. Sometimes in the process of writing a good enough prompt for ChatGBT, I end up solving my own problem, even without even needing to submit it. the AI folks have now discovered thinking. Like shit.

Brandon Corbin (54:13.059) Ha

Brandon Corbin (54:28.567) Yeah.

Sean Hise (54:35.984) Yeah, I think it's important. think, okay, we all want to one shot apps. We all want to do this, we all want to do that. But at the end of the day, it's probably a better use of your time to have AI do as many ubiquitous tasks, project setup, initial code, getting your functions, getting all your stuff. And when you get super stuck on a very specific plugin configuration or something that, yay.

you know, unless you're Brandon and you just want to yell on LinkedIn and say, fuck a lot, you know, go and look it up yourself, you know, go. Because at the end of the day, we still need to continue to ask questions on the internet. I know your point has been made though, and I feel like we just need to move on. No, but I mean, I like, I think one shotting these apps is totally possible. But you, lot of those think situations you need to be.

Brandon Corbin (55:11.906) Hey, it's a point, it's a point, I'm making a point.

Sean Hise (55:28.636) doing something that is, you know, it's a widely used API. It's a widely used thing. The more niche it gets, or the more new it gets, the harder it's gonna have building stuff for you and doing all your work for you so that you can just do nothing.

Jacob (55:41.186) Yeah. And again, that's why these replets and platforms that have all these modules are going to do pretty well because it's very fixed and they can fine tune those models to work really well with their components. You still need to understand how to build them. You still need to understand like what you're doing, but it's better than like just random ass react code, right? Cause it's so open-ended. Like the way we build apps is just, it takes an immense amount of creativity and experience.

And I think we're moving towards a future where maybe there's a little less creativity. They removed some of those variables.

Brandon Corbin (56:18.434) All right, so I'm gonna drop a name here real quick before we wrap this thing up. And so this is Lord Aster West. And so this is a girl that I've now been interacting with online on LinkedIn quite a bit. She is the CEO of a company called RadHash. And I think I'm still really trying to wrap my head around what she's building here, but it's basically a framework for startups to basically be able to build all of these platforms using AI.

But what they've done is that they've actually like, from what I understand, I've been asking for like, and so this is my plea for, give me a frigging demo of this damn thing so I can actually talk about it intelligently. That it seems like what she's decided to do is that we're going to go train AI models on very specific applications or functionality or frameworks that exist. And then when you go and you want to build your tools, you're just working within those existing frameworks. So instead of just saying boiling the fucking ocean,

Sean Hise (57:14.244) I love that idea.

Brandon Corbin (57:17.355) Here's a handful of frameworks that you need. You need auth, you need database, you need this, you need that, you need that. I think that that's a really interesting model. She's actually going and hitting hard. She's an awesome one to follow, because you can tell she's hungry, she's out there, she's really loud, and she's trying to raise money for it. So she's a very fun, I think she is.

Sean Hise (57:35.164) Well, she's also right. She's also right. Because the least variance you can create in this ecosystem, the better. Because it's not going to hallucinate some random thing. It's not going to send you off on a wild goose chase.

Brandon Corbin (57:44.927) Yeah, exactly.

Jacob (57:51.19) I, yeah, I still think templates, that templates plus that Brandon, because think about this, like we w when we're building something, we make decisions all the time of like, okay, let's think of like, authorization and permissions and roles and all that at first. So we, you could build a very complicated, system for that, but your app might not probably doesn't need it right away. So what do you do? You reach for something that's a little simpler because it's easier to implement, right?

Brandon Corbin (57:57.601) Mm-hmm.

Jacob (58:20.59) And then down the road, if you want it to switch, it's a big pain in the ass and you're never going to do it. You're going to have to rebuild it or, burn it to the ground. Like with something like that, like to your point, Sean, like you could use a more sophisticated system and implement it in a simpler way. So, and then build on top of that later on. The reason you don't use it in the beginning is because there's too big of a barrier to entry to use that system. But if that's the one they chose or you know, you could, you could replace a.

Brandon Corbin (58:26.773) Riot. Riot.

Jacob (58:50.25) authorization with any anything right like and maybe there's not one one solution for all and maybe there's flavors of each but that to me is very attractive.

Brandon Corbin (58:57.093) Right. Yeah, yeah. So Lord Astor, give me a demo of the damn thing. Come on now.

Sean Hise (59:05.276) All right, Lord, after we're coming for you, maybe you can be a guest in our podcast and we'll do another remote. well, thanks for sticking with us. we appreciate your patience as we get through our mood.

Jacob (59:05.294) I'll see you then.

Brandon Corbin (59:08.993) That would be awesome.

Brandon Corbin (59:27.521) Whoo, awesome. All right, guys.

Jacob (59:28.462) See you.