What AI Actually Does for Me: Codex, Claude, Blender & Everyday Tasks

So Lilly has this thing called Wild Rebellion that she's been doing. And it is, you know, it's pretty exciting to me to see her. She's been doing these drawings for me. And I was blown away. So I tried this six months ago with Opus 4.6, I want to say. And it was not capable of doing it at all. And now with ChatGPT Astra, right? So you can see ChatGPT Astra was helping me set up this, giving me the thumbnail and stuff to set up this live stream.

So with ChatGPT Astra, we now have the capability to do this using Blender. So I was really excited to see this thing actually come to life with her characters. It just failed all over the place with Opus 4.6. It had a lot of problems with the camera. So the way Blender works, I, by the way, it's been almost 20 years. I can't believe this tool has existed all this time. And it's still a really good tool for what it does, right? 3D animation and really archaic game design. Like it's pretty powerful to do something like this.

And I'm just blown away this tool still exists and is still in use. And, you know, and it's pretty cool to see something whenever I was in high school that we were using to try to build this kind of stuff. But basically you have a camera and it's just like shooting a video where you're building these 3D models and then the camera has to move around. While Opus 4.6 could not move the camera in the right position to get, like, it just kept getting in its way and it couldn't create the scenes that I wanted and then tie it all together. It was just failing over and over again. So now, I mean, you can see with this, it is so much more capable, right?

Like, and not only that, episode two, which will be coming out, has even more animation, has even more, has even more, you know, has even more interaction. It's pretty cool to see the voice stuff that it can do, like where it actually, you can see the lips move. It's pretty amazing what Astra can do. And episode two, we'll have even more in there. And it's all based upon Lilly's drawings. It's all based upon what she wants to build. We're also working on a few other things that I'm pretty excited about. So she plays a lot of Roblox and I wanted to build a version of this inside Roblox for

and it looks like I actually switched tools for that. So inside that, inside, this is Claude code from the command line. So I actually built that inside Claude code from the command line and it, and that's using Opus, I think five, five is the latest, right? That's the new Opus that just came out. Opus, I'm pretty sure. Five, five. Five. Yeah. So Opus five, five. So Claude Opus five, five. This thing is unbelievable. Like I could not believe Claude Opus five, five's capabilities.

I would say it is not only on par with Astra, but it is so fast in comparison to Astra. I cannot believe the speed. I've never seen a Claude model with this kind of speed. So I was really, really impressed with what Opus five, five could do in terms of building the wild rebellion Roblox version. So now it's building a Roblox version of the game, like an actual game. And then it's also, and then we're also building, you know, based upon her drawings, episodes

of the show. And I'm just really, really happy with what AI tools can do today. It's all like Lilly's imagination can come to life. And that was just not possible six months ago. For people who say AI is, you know, not progressing. Like I hear a lot of different things where like, you know, I tend to listen to a lot of AI news and there was a diary of the CEO had on a whole bunch of AI naysayers on there. And they were saying that it's never going to be successful. It can't be successful. And I'm like, you guys are insane. Like, I don't argue with, you know, are these companies going to turn a profit?

That kind of thing. That's not, that's not where, well, first of all, NVIDIA was a small holding of mine and now is, you know, my largest holding because, because I made a bet years ago that Intel, in terms of, you know, this is not, it's not stock advice or, or, you know, like investing advice in any way. I'm not a licensed, whatever that, I'll put the disclaimer below. I'm not a licensed, you know, whatever, you know, in terms of the stock market.

But first of all, I think it's something like 90% of the S&P 500 gains over the last 10 or 20 years have been tech companies. So betting in tech related fields is, is statistically the best thing. And then you look for, and then in my opinion, I just look for, uh, what company is failing miserably. And that company was named Intel. Intel was failing miserably. So I just put large bets on AMD and NVIDIA. And those large bets became, uh, you know, way more in my portfolio than I expected.

So, uh, but I have a mostly tech focused, uh, portfolio in terms of what I do even in my retirement. And that has been very successful for me. And it's just statistically, you know, tech tends up making, but do I have any AI specific bets right now? No, because I don't know about profitability. I can tell you, uh, the race to agentic workflows and the race to, uh, become the best model. These guys are neck to neck. Uh, what's funny is by the way, in terms of my day to day usage, uh, for anything that is

super important to me. So like the wild rebellion episodes themselves, those are still, uh, being done in Astra. So those are, um, this, you know, this app, uh, being done in Astra. So being done using Astra, usually high, right? By the way, that's very important. I didn't realize that switching this, switching from high to medium to low effort, realize you no longer get cached reads if you're inside of an individual thread. And so that actually can cost more tokens than just leaving it at high. So I think that's a really important thing for people to learn, uh, in terms of these

things, switching, this is not a free action, uh, and may not actually result in you using less of your weekly usage. Uh, I didn't realize that. Like if you're going to switch this for sure, create a new thread and feed it in the context that you want in terms of the information you want to be able to get started and not do as many tool calls to find what, what it needs to find about what you were trying to build. But otherwise, so we have this in Astra. This is like the most important wild rebellion, uh, this air live stream, right? Anything like that, that I deem important that gets done inside of this usage.

And this is a hundred dollar a month plan. I thought about bumping it down to the 20 because I don't always fully use everything. Um, but the problem with it is the second I do that, I'm going to use, I'm going to, I'm going to, I think you lose your banked, uh, resets. So I have a, uh, like three resets and I think you lose those. And every now and then I go over my usage for the week and I have those resets. So this is at, at that, this is at the hundred dollar a month plan. Then I have Claude Code, right? And this is $20 build yearly. And this, this is my opinion, the best bang for the buck to just be able to play around

with things. So I didn't think this Roblox game was going to even be buildable. I had zero confidence in it. I figured we'd just try it out and see, see what it looks like. And I have been blown away with what it was able to build so far. So we're going to do some testing rounds tonight on this. And this was all built with Opus five, five and with the $20 plan, which I think went from the least usage and kind of not, not that valuable to with Opus five, five and with them giving us a reset, uh, everyone has a reset so that they can, they can get even more usage.

This became a super, super valuable plan. It definitely varies. Uh, sometimes it's amazing. Sometimes it's not amazing, just depending on where they are in model life cycle. When fable was in it, it was amazing, but I burnt through the usage immediately. Right. When, uh, it kind of became stale and it was when, when, when it was just Opus five, but with five, five, it seems like it's the perfect balance of, of performance and, and, uh, usage. Then I have Antigravity and believe it or not, I still use Antigravity. I use it for questions that like, this was one of my questions.

I use it for any kind of question that I don't want to spend, uh, resources on. By the way, I burnt through, uh, and I finally got it back. I burnt through my Gemini usage for the week. So I'd actually burnt through all of my Gemini usage. And, uh, so now I was stuck using Claude Sonnet 4.6 and I probably should have switched it to Claude Opus 4.6, but for this question, it didn't really matter. And so these are just generic questions. And if this didn't come with, with Google's pro plan, so this is free to me, uh, with that

pro AI plan that gives you more storage on your email and gives you a whole bunch of other things. Uh, one of them is no ads inside YouTube, which I talked about in one of my previous things. And honestly, I think I'm going to keep this no matter whether, uh, this basically only is used for kind of that situation where I just want to ask generic questions. Uh, so that that's in my lineup. And this is basically any kind of questions, file system related stuff, because search is still absolute garbage in Mac OS. Like it's just terrible. Like spotlight search is garbage.

It doesn't work half the time. It's so much easier to go to this file system, hit add. And then you can see this was find my MiniMax API key. I'm going to click on that. Cause it did find it. Obviously it found all the different locations where it was, uh, on my system. And by the way, then I deleted the key, uh, because it's actually not necessary for what it's doing. It's using OAuth for all of my, so I'm using MiniMax. That's the other piece. MiniMax code is actually being used and my subscription is being used. Uh, not, I actually don't use this MiniMax code app very much, but underneath the hood,

it is being used a lot because all of the audio tracks, all of these songs that were built for wild rebellion are actually using MiniMax. And I have, I have, uh, the starter plan, which doesn't even exist anymore, I guess. Um, that is, that is powering all of that. So the, the starter plan was an amazing deal where it was a hundred dollars for a year. And, uh, you know, I think a little less than a hundred dollars actually with the year break. And that was an amazing deal that no longer exists. And I don't know if I would pay one of their current tiers to keep this not with, if I had

to pick one of the two, I would actually keep, even though MiniMax does more, um, I could probably use Google's models to do the same thing with the audio. I'm not sure, but I could probably switch over to Google to do a lot of the stuff that I'm using MiniMax for, so that will probably not get renewed. Unless they offer me, unless I can stay on the starter plan and stay at that. If I can stay at that starter plan with a hundred dollars, even though it's discontinued, right. And not bump up to a new plan, then I'll just keep it, uh, because that was a great deal. Uh, but so MiniMax is still in the loop and it's being used to generate all of the audio

for these different things. Uh, I don't actually use the MiniMax code app very much, but I do install the latest. Uh, and underneath the hood MiniMax is actually being called, uh, by, by, uh, and I think people don't realize by the way, the ChatGPT can call other models like these, these, these, these, these interfaces can call their models and do a bunch of work that is not related to, you know, that isn't just routing through your ChatGPT membership. They can do that. You just have to ask them to, uh, and especially for songs and stuff like that.

It's just much better to get that from MiniMax. Cause I have been for a long, long time. So that's, that's in the loop. And then actually I just, what's funny is I, I, I have the meta AI spark thing, right. And the worst part is I can't figure out, uh, like this, this is a use case because a buddy of mine is actually finally repairing the, uh, finally repairing the, and it did find it for me. It found the Reddit teardown post, but, uh, my buddy, you know, finally is repairing those. So I'll have my tonal back soon with, with, uh, fixed cables, but, uh, you know, I had,

that was the only real use, you know, and what's funny is the meta AI Mac compatibility check this. So meta released this thing, the Mac version of their app. And what was even funnier is it was not compatible with, uh, with, uh, 27. So it was not compatible with 27. And I just thought that was hilarious that they released it not compatible with 27. And then they did push an update a couple of days later and now it's working. That was one of the questions to Antigravity like, Hey, can you fix this thing? No, they had to actually update their, they had to update their app to be compatible.

And now they have a five one version, I believe is the version number of this. It's compatible. I haven't been using it very much. So in my day to day usage, I use Grok to ask a lot of generic questions on my phone, um, with their free tier. And then this is also a free tier thing. So I'm playing around with free tiers. And, uh, this seems like it would be very capable, but I don't have a good use case for it. And I see online other people with use cases for it. But for anything like that, I'm going to, it's either going to be important enough for me that I use GPT itself, or it's going to be not important and I'll use Antigravity.

So I haven't found a use case for, or I'll be using Grok and I'll be talking to it because I'm out and I have a simple question. So I really don't have a good use case for it, but it just wild that, uh, Muse spark is so powerful. Oh, and that's the other thing. I still have my Hermes agent and that is now being powered by Muse spark, um, or the Muse, the Muse free tier contributor tier, uh, which is now free that I would strongly recommend if you want to set up Hermes and you want to do anything like that, that is a great starting point. So the first thing that I would like to do is plug in the contributor tier of Muse into

Hermes and you can have a completely free setup that works really, really well, uh, and gives you computer use. Cause the problem with this Muse app is it has no computer use capability. So for someone like me, it's not really the best, you know, the best solution. Uh, it really, you know, I mean for, but yeah, you can use, so Muse contributor. And I believe that's using open code. So open code.

So I believe it's open code that, that ends up doing that. Uh, yeah, open code. So, but, and that can be done all for free, which is pretty cool. So that is an, a truly free tier way of getting into this agentic world. And I strongly recommend people go that route. So that is a really, you know, that is a pretty, pretty cool solution. Um, so that's kind of my stance on all of this. There's, there's so much going on in the AI space, uh, grok, by the way, I would love to

test their stuff, but I'm not going to pay for any of those tiers because all of it's kind of paid and I do use their free tier, but I just don't think compared to what you can do with Astra, it's worth doing. And anything that's really important, like sending emails or, uh, believe it or not returns. So I never returned anything in the past, right? So if a device, so my wife's Amazfit, uh, BIP and Amazfit Helio both broke. And it's been about a year, which is pretty unfortunate that those devices broke.

And, uh, you know, I wasn't going to post any kind of reviews or anything about it. Just because I, I, you know, it's not fair to the company if you're not going to ask for, you know, ask for them to be returned and replaced to do anything like that. Uh, but by the way, you know, I, and I don't have time to do it. So I just ask ChatGPT to do it for me, right. To go through, uh, find my Amazon information, find all of the, the payment information and, you know, send over all the information to, to them. I had to take a couple of screenshots of the devices, uh, for it to get all the information

that it needed because one was dead. So it couldn't get the serial number unless you take the picture of the device, took the picture of the device and sent them all of the information they needed, went back and forth to, to get everything, you know, to get all, all the information. Cause it was a couple of back and forth and boom, I have new Amazfit Helios and a new, uh, a new BIP-6 for my wife and they just shipped them. Uh, so they're, they're a pretty incredible company, you know, because one of them, the warranty had definitely expired and they still honored the, they, they honored it even like slightly. It was like a couple months past the one year window and they shipped her new devices.

So, uh, we'll be testing those out to see how long they last, right? Cause the original devices lasted between six and nine months of her actually using them. The ones passed a year because it was refurbished, right? So I believe that would go from the original refurbished date. Um, or, or maybe it was just past a year in terms of either way they replaced it. So they replaced it and I didn't think they should have to. Uh, but I wanted to see because, you know, I wasn't going to say anything negative if they said, no, we won't, won't replace this. But that's an amazingly positive experience that they were just willing to replace both

of those devices. So I was really impressed with Amazfit. So what I'm talking about here is the Amazfit Helio, which is this device, right? Which looks a lot like a whoop, uh, works pretty much as well as a whoop. My wife, my wife used this for quite a while. Uh, and then the Bip 6 because she wanted a watch. So she wanted, and I guess the Bip Max is there, but here's the Bip 6. So she wanted a watch, right? So, and you can see these things are under a hundred bucks. So pretty, pretty amazing devices for very, very cheap prices.

Now they did break pretty quickly, but, uh, you know, they didn't have the longevity of my Garmin that's sitting on my wrist right now. Now this Garmin's been on my wrist for years. Uh, and I, you know, I wear it every day. They did not have that kind of, uh, longevity, but they replaced them. So that was incredible. But I wanted to give you guys an update. I wanted to hop on here. I haven't had a lot of time lately. I've been really, really busy with, uh, with the stuff for Lilly, with, you know, with jujitsu and with, with everything that I've been doing. I still have one more jujitsu video to shoot the brown and black belt.

Like I got halfway through brown belt and in that video, like, and I didn't finish it. And then I have a black belt video to shoot. So, uh, I just got my black belt recently. Um, and so I wanted to, you know, finish that series out, but that'll be coming soon. And I wanted to get on here and give an AI update and where I'm at with all of these tools, what I think of these tools. I think they're, you know, amazing. Uh, Astra is an absolutely amazing model. I've never seen anything like it. And then when Opus 5.5 came out, it's equally amazing.

So I cannot believe we have two amazing frontier models that are, uh, the, the pricing on them is incredible. I mean, to be able to do, uh, all of the stuff that I can do with these things, to be able to power blender and render things and, uh, build, you know, Roblox worlds based upon my, my daughter's vision is just absolutely incredible. It's never been a better time, uh, to be excited about technology. And we have so much doom and gloom. We have a lot of talk about P death and all of this kind of stuff.

And I think it's looking at, it's looking at such a negative light of what our future could be like. And I understand it. Like I'm not, uh, you know, especially whenever all of these companies and all of these, uh, you know, CEOs, like I got, I got to listen to every CEO, uh, talk about it. I kind of went to, to what Dario, what Sam and what Elon all said. And I actually love what Elon said, to be honest with you. Uh, I know there's a lot of people that don't like Elon Musk out there, but I think sometimes

he gives us the most real, you know, uh, opinion of like his real opinion. And, and it's, and I think it's the most accurate one in terms of, you know, someone was asking him, what's, what's our chance of death or, you know, what's, what's, you know, our chance of death. Uh, you know, and, and they forgot to include from AI and he said, well, we have a hundred percent chance of death. You know, we, we're all going to die. I don't know if you know that or not. Uh, you know, I'm pretty sure the track record of the death rate of individuals in the world

is, is a hundred percent, you know? And it was just, it just like, we can't predict, you know, this P doom, like everyone's throwing out 10, 20, 50. We can't predict that, but we can predict that we're going to die based upon the past. You know, it's the same way I do with anything. Like what's the percent chance of this being successful? I do it with everything in my life, right? The percent chance of success, um, with things with my daughter, it's more around trying to, to bring her imagination to life because we lose our imaginations as we get older. You know, I don't have the imagination I had when I was her age.

I was catching Pokemon in the backyard and I was doing all of this stuff, but I couldn't bring any of the ideas to life. And now we have the capability to do that for the next generation. And, uh, I think, you know, in a consumerism world, building something and building something that you're excited about, it means more than money. So that, that's my personal opinion on it, that, you know, this stuff matters, but it matters because you give it value. And the value is, you know, spending time together, uh, teaching, you know, like teaching my daughter about public speaking and, uh, analysis of what you're building and having structure and

having a plan. I think all of that stuff matters. And so it goes beyond, you know, what, what we're doing here, what AI can do. You know, I don't think, I don't think you should ask, you know, like, are you making money with AI? It's what can AI do for you? How can it help your everyday life? And I think that's really, really important. And I want to leave you with that. Uh, I don't, you know, I have my own estimations on, you know, what could or could not happen with AI in the future and whether we have a, you know, how much of a risk we have with these,

with these tools. I've been blown away with what they're capable of. And I can tell you, they're getting more and more capable. Uh, and, and some of the, the stuff that we had, like, uh, like the recent, the recent hacks that we've seen are pretty amazing that these tools can escape their containers and coordinate together, uh, when given an objective with, when given a task, uh, and their propensity to try to cover their tracks is scary and is something that unfortunately needs to be fixed,

but it has to be fixed at the frontier labs. It's not something you or I can fix, right? Uh, uh, individually and not using the tools will not stop them from continuing to progress. Uh, and I think when government gets involved in this stuff, it gets even worse. So, you know, especially in a world where we're competing with Chinese models that are getting very good. Uh, and so I really think alignment and that kind of stuff is important. It's just not, it's not something that we can directly impact.

And I do think Elon is right that, you know, that these, the fear mongering is, I, I don't know how helpful it is, but I, I do, you know, I do think that what we can do today with these things is so amazing that it's worth, uh, at least having the models we have today. Right. I mean, with Astra, with Opus five, five, these things are absolutely incredible. And, uh, I think they have all of the capabilities I could ever ask for.

Uh, and I can't imagine, you know, I can't imagine what the future is going to look like because these things are so phenomenal today and I'm actually excited for the future and what I can continue to do and what I can continue to build, uh, especially taking my daughter's imagination and bringing the things that she wants to life. Anyways, I wanted to leave you with that, but thank you so much for your time and I really appreciate it. Uh, send me any comments on anything you'd like to see me talk about in the future.