This Week in Startups - AI Demos: ChatGPT-o4 Canvas, Notebook LM & Meta's Movie Gen | E2026

Episode Date: October 16, 2024

This Week in Startups is brought to you by… Lemon.io - Hire pre-vetted remote developers, get 15% off your first 4 weeks of developer time at https://Lemon.io/twist Fidelity Private Shares℠. Manage your cap table and data room, get faster, more accurate 409A valuation and fully automate your next financing round. Visit ⁠https://www.fidelityprivateshares.com⁠! Mention our podcast and receive 20% off your first-year paid subscription. CLA. Innovation takes balance. CLA's CPAs, consultants, and wealth advisors can help you get from startup to where you want to end up. Get started now at https://claconnect.com/tech * Sunny Madra joins Jason to discuss Tesla’s Robotaxis (2:17), demo ChatGPT-4o new Canvas interface (25:23), break down Meta’s Movie Gen model (41:11), Notebook LM (54:41), and much more! * Timestamps: (0:00) Sunny Madra and Jason kick off the show (2:17) Tesla's Robotaxi reveal and challenges with regulatory approval (10:00) Lemon.io - Get 15% off your first 4 weeks of developer time at https://Lemon.io/twist (11:22) Self-driving architectures and potential applications for Tesla's Cybervan (18:15) Update on Tesla's Optimus robot and the future of autonomous vehicles (22:04) Interest rates and startup funding challenges (22:13) Fidelity Private Shares℠ - Visit ⁠https://www.fidelityprivateshares.com⁠! Mention our podcast and receive 20% off your first-year paid subscription. (23:14) Traffic congestion, urban planning, and future players in autonomous vehicles (25:23) ChatGPT-4o with Canvas demo (31:28) CLA - Get started with CLA's CPAs, consultants, and wealth advisors now at https://www.claconnect.com/tech (41:11) Meta vision models and Groq demo (42:35) Apple AirPods updates and AR glasses privacy concerns (47:46) Meta's new vision models, open-source impact, and OpenAI's business model (54:41) Notebook LM's tools (1:00:13) Capturing family history with AI and Meta Movie Gen's potential * Subscribe to This Week in Startups on Apple: https://rb.gy/v19fcp * LINKS: Check out OpenAI’s Canvas: https://openai.com/index/introducing-canvas/ Check out Notebook LM: https://notebooklm.google.com/ Check out Meta Movie Gen: https://ai.meta.com/research/movie-gen/ Check out Meta Llama 3.2: https://ai.meta.com/blog/llama-3-2-connect-2024-vision-edge-mobile-devices/ * Follow Sunny: X: https://twitter.com/sundeep LinkedIn: https://www.linkedin.com/in/sundeepm * Follow Jason: X: https://twitter.com/Jason LinkedIn: https://www.linkedin.com/in/jasoncalacanis * Thank you to our partners: (10:00) Lemon.io - Get 15% off your first 4 weeks of developer time at https://Lemon.io/twist (22:04) Fidelity Private Shares℠ - Visit ⁠https://www.fidelityprivateshares.com⁠! Mention our podcast and receive 20% off your first-year paid subscription. (31:32) CLA - Get started with CLA's CPAs, consultants, and wealth advisors now at https://www.claconnect.com/tech * Subscribe to the TWiST500 newsletter: https://ticker.thisweekinstartups.com/ Check out the TWIST500: twist500.com * Great 2023 interviews: Steve Huffman, Brian Chesky, Aaron Levie, Sophia Amoruso, Reid Hoffman, Frank Slootman, Billy McFarland * Check out Jason’s suite of newsletters: https://substack.com/@calacanis * Follow TWiST: Twitter: https://twitter.com/TWiStartups YouTube: https://www.youtube.com/thisweekin Instagram: https://www.instagram.com/thisweekinstartups TikTok: https://www.tiktok.com/@thisweekinstartups Substack: https://twistartups.substack.com * Subscribe to the Founder University Podcast: https://www.founder.university/podcast

They've taken all the image editors and all that. Oh, that is not a, so that one is a good job. I mean, that actually, I look like a hitter. Like, hey, would you like a soy latte? I feel like I'm making you a crapped cocktail. This week in startups is brought to you by lemon.io. Hire pre-vetted remote developers. Get 15% off your first four weeks of developer time at lemon.i0 slash twist.

Fidelity private shares. manage your cap table and data room, get faster, more accurate 409A valuation and fully automate your next financing round. Visit FidelityPrivate shares.com. Mention our podcast and receive 20% off your first year paid subscription. And CLA, innovation takes balance. CLA's CPAs, consultants, and wealth advisors can help you get from startup to where you want to end up. Get started now at cLA connect.com slash tech.

All right, everybody, welcome back to this weekend. Startups. Our guy, Sindit Madra, is back. He's been busy. Grock is cooking with oil. How are you doing, my friend, Sunny? I am doing excellent. It's been a really, really busy.

And we said this. Remember, it's going to be a Q4 plus one. So we're now actually in Q4. And we are cooking. We are cooking. It is, we're 10, 11, 12 days in by 10 we publish this. And J-Cal, tell me about your September in Q4 and, like, what's going on? Because, you know, I am settled in here in Austin.

It's been absolutely fantastic. And we're doing our next class of Founder University, not our ninth cohort. You know, that's where we have people come for 12 weeks. It's virtual. 250 teams, two or three founders, each team, hopefully. a technical co-founder or two. And then we invest in the top 10% of those. We might put 25K in if they want like that first friends and family check or put 125k in, you know, if they want the standard like incubator deal in Silicon Valley. And then, you know, what will be different

this time is I'll meet them all for the week one in San Francisco with some of our partners. Then my plan is in week six, nine, and 12 perhaps. The top performing companies, the ones we invest in, have them come to Austin for the day or two, hang out with me, go to the salt lick, have a little brisket, some bison ribs, you know, the one that broke my tooth and a half. Oh, yeah, yeah. And maybe a little Terry Blacks, beef rib, which I took Chamatsu for the first time, and, you know, just jam out with them about their vision. And I am super excited just because so many of these companies are able to go so far with such a small amount of capital, especially in their teams of two or three because of AI specifically. And it's really wonderful to see how

fast people are working. And then I'm very excited about 01. I've been using chat GPT-O-1 version to the point at which it gives me like a warning about credits, you know, and says, hey, you've got to take a pause. That's how often I'm using it. So it's really amazing. I had just, while we start here, we're taping on a Friday. I'm not sure when this is going to come out, but we taped on Friday, October 11. Last night was the Cybercab big reveal. I wasn't able to make it. Sorry to my friends over at Tesla, but I was traveling this

week for Saxes LP day earlier in the week, so I couldn't do two trips in one week. What were your thoughts when you saw the cyber cab, the cyber bus, and optimist especially in relation to AI and where we're at? Yeah. Okay.

I think it's a great leadout. And I've been thinking about this all day and we talked a little bit about it in our group chat. Okay, let's go in that order. I think the cybercab is really a culmination of everything that, you know, Tesla has been doing, which is building cars, building great electric cars, you know, building on autonomous technologies with sort of all the in-camera learning, and, you know, their ability to basically, you know, put that all together in a form factor now without a, you know, steering wheel and or pedals.

A quick aside, I heard a number of people say, put a steering wheel and pedals on that and I'll buy it. I mean, I literally said that as well. And I think that this was in the Walter Isaacson book about Elon where Franz or whatever, you know, it was a big internal debate. And I think Elon went with the kind of burn the boats according to that approach of, hey, we just got to force our autonomy. We got to take the steering wheel out, et cetera. And so maybe they'll be what they were calling, I think, the model two. which is the two-door hatchback, I want to buy that two-door cyber cab so much.

I would buy two of them immediately because they just look so fun and cute and zippy to drive. Can you imagine a Tesla power train on something, you know, of that footprint? It would be unbelievable. It would be like that Nissan or is it Toyota, the GTR, that all the racers like to be over drifting and stuff? It would be like a mini Cooper, you know, you get that really small thing.

I would love to put fat tires on then and feel it. I love the design, too. I like the fact that this feels like kind of the love child of the cyber truck and the model wire model three. It's kind of got its own, like, thing going on there. So aesthetically, I love it. I love the vision of it, obviously, for self-driving. What did you think of the bus?

To me, that I think it's an RV. I looked at that thing. I don't know if you saw, do you see my tweet that I did about it? Yeah, yeah. So let's talk about the bus, because I thought the, Well, let's finish on the cap. Okay. So I think what they've been able to do is bring together all the pieces.

Now, the final thing, which, you know, I think there's the market reacted negatively to Tesla. But I think that's probably a buying opportunity because, look, the stock went down, whatever, five or ten percent. Yeah. Yeah. But look, think about what they put out there and what it's capable of. And I personally think given, you know, it's. there's a couple of things that Elon really touched on in the talk track was, look, we collect millions or I don't even know what the number is,

but let's just say a huge amount of data. And there was a video that he showed during that, where it showed all these, like, crazy scenarios of, like, people jumping in front of a car and, like, you know, a car crash happening right in front of you or someone spinning out or, you know, someone crossing the road really, really fast. And so all of that has been kind of given to them through their exact, car network, which then they can use to build models around.

And so I actually think, and look at how confident they were yesterday. They allowed people to just use it inside, obviously, a square, you know, no safety driver because there's no steering wheel. So. Yeah. Yeah. And they were letting anybody do it. And it wasn't just sort of.

On a controlled environment. So obviously, it works in a controlled environment. So on a grid system, it's ready to go now. But you have regulators. And regulators. You know, they like steering wheels. They like review. They like side mirrors.

So this is, I mean, really going to be challenging with regulators, I suspect or not. I don't know what you take on that. Yeah. So I think, like, you know, what we'll probably see is like the crawl walk run because they know they can do it there. So you can imagine like in a, and I think companies are going to buy them first. So think Disneyland, right? Think all, you know, all these kind of places that operate.

Think about like a retirement community in Florida, right? So think about all these. these like, you know, places. Like a campus. Yeah. Yeah. Like, exactly. Where there's golf carts or those little things.

Yeah. Exactly. Easy. And I think you sell in there first. And then as the technology evolves, because he did say end of 26 or 27, I think by that point, we get to kind of widely generally available release. And I think that's the underestimation because they had enough confidence to allow people to, you know, just use it openly.

obviously in a fixed area, but they weren't scared. They had no safety person in there. They had nothing on it, which is incredible. So that's my summary there. I think we get it in, you know, what you call, like, fixed, fixed cordoned-off areas. We get it there sooner than I think people believe. 2027 would be a pretty good estimate, I think. He said he could have those cars being manufactured.

It was interesting because he was like kind of, and I'll say 2026, but maybe 2027, he kind of, He kind of hedged a little bit there, which kind of, maybe that's why the stock went down a little bit because maybe people want everything yesterday. But the truth is, you got to get this right. And when you do, they have the ability to produce, I think they did 1.9 million cars last year. So he does have the ability to produce that volume. And so, listen, you know, he can put millions of these on the road, you know, in a short period of time. However, you know, just doing back of the envelope math, there are a billion rides a day in the U.S. A billion rides a day in the U.S. A billion rides. Of those rides, about 12 million of them are Lyft and Uber. So, but 3%, 2% of rides in the U.S. are done by ride sharing. It is a minuscule amount of the overall rides. And we're not even taking into account. When you do that math, of course, public transportation and other forms of transportation,

bicycles, et cetera. So there's big pie here. And of course, if this became cheap and ubiquitous, people might stop buying cars and use these. And people would induce more usage. So it's a bright future. I do believe there'll be three, four winners in this because it seems like everybody's solving it at about the same time. But there is a difference between Waymo and Cruz writing code top down versus building it off the data set.

All right, founders, are you tired of doing all your own software and development? Do you need help? But you can't afford all this time it takes to find great talent? Are you dreading the endless interviews and email chains just to find somebody great? It takes six months. It takes a year. Well, what you need is Lemon.io. Lemon.io has thousands. That's right, thousands of on-demand developers who can help you,

and they've done the work already to vet these developers, making sure that their results-oriented and that they're super experienced. Of course, they've got to have competitive rates, so they're going to take care of that as well. And great developers are so hard to find it integrate into your team, unless you're using lemon.com. because they handle all that for you. They only offer handpicked developers

with three years of experience at a minimum, and they have to be in the top 1% of applicants, right? Something goes wrong. Don't sweat it. Lemon.io will find you a replacement developer ASAP. So many of our launch founders have worked with Lemon.io, and they've had great experiences. So here's your call to action.

Go to lemon.io slash twist and find you're a perfect developer, or the perfect tech team in 48 hours or less. That's right. And Twist listeners get 15% off the first four weeks. Stop burning money, hire developers smarter and faster at lemon.io slash twist. So maybe explain to the audience in plain English as best you can the two different architectures of these systems. We know one has LIDAR, the other ones don't.

Or Tesla doesn't use LIDAR, the other ones do. So there's one issue here, but we beat in that. that dog to death a million times. Let's talk about the model and how they're doing it. And so, you know, it's actually kind of related to that, right? Because, you know, what Elon says as a human has two eyes and sort of a head that moves around. And that's all it has. And we're able to drive cars, you know, perfectly fun, right? That's what we're trying to mimic. And the reason, you know, others add sensors is all the, you know, these eyes, the eyes, the cameras can get covered. So you have an extra set of sensors that give you another view of the world and then use

that. So now you take that into the model, right, where if you're building a model that's using vision and using like a LIDAR and using, say, ultrasonic's, you have to write code that brings that all together. And so what you're doing there is basically sort of, you know, I would, I just call like sort of the traditional approach across all those things and you're double checking and you, you know, it's, it's valuable that you have all that, but it's much more difficult because you're, you're having to have the code and the drive and all those things make decisions, right? What Elon and team did, and this is a gross simplification is, they just took a model that would learn to drive by watching video.

All they fed it was hours or maybe hundreds of millions or billions of hours of video that they had from the cars. And so just figure out what humans do and follow what they do. See how they sit in lanes. And so instead of Cody and saying, you know what, what people tend to do is they don't drive right in the middle of the lane. They tend to move based on if the hug the left or if there's a car to the right of you, they go to the right or if there's a bike.

They even go out. Give it a little extra grace in that period. I'll give you an experiment you can do. If you make a right turn, but there's no one in like, say, the oncoming lane into the lane that you're turning into the road, it'll actually go wide and go out into the lane into the oncoming traffic. And that's because it's learned that from humans, right? Like, you don't need to hug the right side curb if no one's in the, it's, yeah, you're kind of scared

to hit your rims or whatever it is, right? And so what it's learned to do is like, just do that by watching humans and say, well, there's no one that lane, just stretch out into the left turn lane, the oncoming left turn lane. And, you know, you can't write that in code. That's the nuance, right? And that's what you have to do if you build kind of a more traditional system versus this one is just watching all the nuances of humans and then implementing them on how they take curves and how they move around people.

And, you know, now it has to follow a certain set of rules. Like, one of my biggest frustrations is it does this, like, full, full stop. And I'm always afraid of, like, someone from behind hitting the car. Yes. Yeah. This is, well, I mean, a Waymo issue as well. I saw somebody on Threads today. I went over to Threads to see what the reaction was there.

And that's, like, that's, like, really the people who hate Elon or, you know, they're the left. So they hate anything that's Republican or GOP. And, you know, one person was talking about Waymo, and they were like a San Francisco. Francisco bike person, and you know how those people are, like, they kick your car, literally. Like, I had one kick my car and almost got into Bistachuffs with the guy. Started yelling on my pregnant wife. That's all another story.

Putting all that aside, the really interesting thing he was saying was, I actually was, I used to screw around with the crews and the Waymo's, you know, I was like one of the guys who put cones on them kind of thing. And he's like, and now I walk with my dog and I love them because I know that they're not going to be looking down at their text or distracted and roll through a, a stop sign and hit me. There's no way they can hit me and they stop perfectly. So the other thing people are

complaining about and somebody on our group chat said this is it's kind of annoying that they're such square drivers and they kind of take their time too much and that's been my issue with using the Tesla full self driving in Austin. Yeah. And is traffic

circuit and it is very, it's very cautious. But you know you can put it on aggressive mode. I have it on aggressive mode and it's still a little scared of traffic circles, I think. So for the team over there, I think that's like, probably traffic circles are not that common. So maybe it needs more training data. And then how people behave in traffic circles is wild. That's one of the other things I've learned in Austin. Like what I thought was the proper way to use a traffic circle is not how people in Texas use them.

They use them like, I mean, they'll be going through them. They don't slow down. Yeah. They just, you know, yeah, it's a pretty interesting moment there. Let's talk about this. the bus. I thought the bus... Robo van. I'm calling it the cyber sled. I think that this could change everything. If you look at this thing, he did mention it could be used for transporting goods. Okay, yum, yum. You imagine this thing with an Amazon person in it? Imagine there's an Amazon rep inside of this thing, and it's driving to the next destination.

That rep is getting the box ready. It opens the doors, it jumps out, it tosses the box. Okay, that's going to save a lot of time. But what were your thoughts on this? It looks gorgeous, obviously. But I have my own thoughts, and you saw my tweet, so we can jump into that. Yeah, look, like, my personal thoughts, like, very similar to yours. Like, this could be, like, a great basis for, like, a whole new generation of RVs, long-distance commute, all that type of stuff, right? Let me get in.

Like, you could own one. It could have your whole office desk in there, big monitor, sit down, work, you know, Starlink on top. It's like a, you know, all things. Mobile office. I've been looking at what I've been. When I moved to Texas, I was like, you know, I'm going to get one of these mobile offices, like a sprinter van. And I started looking into them. They're not cheap.

They cost like 200 grand. I mean, because they dialed them in or 300 grand or whatever. So I was like, well, that's a little ridiculous. But anyway, you can have a desk and you can have like a full-on, you know, basically like it looks like a private jet inside. But I thought these would be incredible for a mobile home. Imagine autonomy, a giant battery. And then you have a hurricane occur in Florida. And they say, okay, send 10,000 of these to, you know, this stadium's parking line.

or to these Walmarts, people come and you could rent one. Or like you're at the World Series of Poker and you want to have your own trailer or you're at Burning Man or, I don't know, you have your in-laws coming in and you order one to your house for a week for $100 a night and you park into your driveway and now you have an extra bedroom. These things are so compelling to me because it's somewhere between a car and office and ADU. You think about the giant battery in there.

How long could that battery last? That could be a battery that lasts 30 days and has enough roof space to put solar on it. So congratulations to the team. They're optimists, looks like it's doing well as well. Well, last thing on optimist, just like the car, they had it out in the wild. They had it serving drinks, they had it

just walking around talking to people and reacting to people. So, you know, I think what I really liked about this was they made it more real, right? And it's kind of funny because the reaction wasn't that way, but by letting car and letting people drive around and letting the robot walk around amongst people. Those are big accomplishments.

And what we forget, Jason, is just two years ago, Elon did the, or you know, Tesla did like the announcement. Remember they had the guy, the suit, like, kind of acting as a robot? I don't know if you saw yesterday, they had that one in like a gazebo and they were dancing. And like you think about it, two years ago, it was a dude in a suit. And then yesterday, it's like, you know, robots actually dancing like in a gazebo. I think, you know, it's like bankruptcy. They asked somebody like, how did you go bankrupt?

And they said, slowly then all at once. And I think that's what's self-driving is. It's like, this thing is taking forever because of edge cases and regulation and safety and competition and, you know, what happens when there's an emergency vehicle? How is the public going to respond to it? But it's pretty clear that this is going to be the future. And then the buildout is going to take time. Because even if you built two million of these a year and they did 30 rides a day,

you know, two million of them doing 30 rides a day, 60 million rides, you know, you're still, it's going to be like 1% of the rides in the United States, I mean, let alone the world. And then you have BYD that copies everything Tesla does. You've got Cruz, Waymo. I mean, there's a lot of players in here. So everybody wants to ask me about like my thoughts on Uber. Well, Uber went up tremendously because I think people realize if every single little J-Cal yum yum. A little young, you know, listen, and I own a lot of different stocks. Like, it's not a big deal. like if Uber goes up and down, you know, build out of this will be a two-decade build-out. I think 10 years from now, 10 years from now, will be at 20% of rides being automated.

So 20, we're almost in 20-25. So 20-35. 20% of rides globally will be autonomous. 80% will still be done by drivers. Oh, dude, in 10 years, I'll take the over. I'll take the over, like way more than 20% in 10 years. Perfect. Okay, we got a $1,000 bet. We'll put that on the bets, this week and start off.com slash bets. There's an easy bet because here's the thing. I think what people don't realize is the number of rides out there is so giant and the number of rides are going to triple. What I learned from being an early Uber investor was that we were comparing it to cab rides,

but then it quickly exceeded the number of cab rides. So you said, well, what happened here? Well, people left their cars at home because they were having to drink or was more convenient, et cetera. And then they started using it for commuting where they might have previously taken a bus. and they said, well, I'll go point to point because I can afford it because it's $8. So a lot of great things going on here, but the number...

Javen's paradox. It's Jeevan's paradox. You make more of a... So, Jevan's... Yeah, Jevan was an economist that basically came up with this theory in the late 1800s, British economist, that said, if you make more of a resource available,

people end up consuming more of it, right? And so it was kind of in the steam engine era, right? Whereas like, oh, like steam engines are going to change everything, but they start in what's it going to do to coal? And then all of a sudden, they start consuming more coal. And so anytime we put something out there, actually a better example here is much more is that farming, right? So when we went to the industrialization of farming, what ended up happening is the consumption of food really increased, right? And like, you know, Americans or everyone's more obese because it's just more of it's available.

we eat more of it. So it's just something that we tend to do is cheaper. Yeah. All right, listen, interest rates are coming down, but it's still hard to raise a new round. Everybody knows that. The last thing you need is to get bogged down in the minutia while raising. So if you're a founder overwhelmed with your cap table management, due diligence, and frankly, managing your investors, you need to check out Fidelity private shares. Fidelity understands that dealing with equity can be incredibly time-consuming. That's why some founders joke that fundraising is a full-time job. I'm not working on my startup. I'm managing my investors. I'm doing fundraisers. And it shouldn't be, right? With Fidelity private shares, you're going to easily manage your cap table and your data room. You're going to get a more accurate 409A valuation. And you're going to get it quicker. You'll also be able to model the future of your company's equity, which you need to do as the founder. Because this dilution adds up. People start signing notes. They don't know what the impact of those notes will be. And you know, you want to model that out.

And you can fully automate your next funding round. So you stay on track. You seem professional. Even if you're scattered, you know, you want to come across as very professional and buttoned up. And that's what having a great partner like Fidelity will do. And you're going to get great support from Fidelity, a great startup community. They do a lot of cool events with founders and VCs. I'm going to be going to some of them.

So check out FidelityPrivate shares at FidelityPrivate shares.com. Mention our podcast. You're going to receive 20% off your first year paid subscription. when I lived in LA, they were constantly making the 405 one lane bigger. And like when it stopped, when they stopped one project and then they'd have two years, what happened was people from people would move further out from the city center because they're like, oh, I can live now in El Segunda and make it to Santa Monica or Century City for work. And then it fills up again.

So the traffic, no matter how many lanes you did, would instantly fill up with that. induce traffic. But congratulations to the team there, and I think it's awesome. And I think there are going to be, I think, you know, we talked about this. There's going to be three or four players who are all going to figure this out to different extent, I believe, in the same 36-month window, which means at the same time. Now, I know that's hard for people to understand that figuring out a technical problem in the same 36 window is the same time. Because there is a physics constraint here. You need batteries and you need resources. It would take literally to do the number of rides just in the United States every there. There's billion rides. You're talking about

50 million of these cars. Like, I don't know if the battery packs are available for 50 million cars. And if Tesla makes two million a year and they were to double it every year and they went to four, then they went to eight. I mean, I don't know what the physics constraint of the factories and the batteries are, but, you know, it's going to take a lot. And so that's why I think 20% being automated, you know, in 10 years as a realistic goal. Right now, we're below 1%. I think that means the pie is going to get so big. Yeah. And I think one, two, and three are pretty clear, right? Waymo, Tesla, and Tesla. And B.D and some of the Chinese companies, I do. Can we give, we can add in Jobi? Sure. I mean, and then going in the air is going to be part of the Jobi,

Archer, and the other one. Yeah, there's, it's pretty clear that we're going to really have a bright future. So congratulations to everybody. Let's talk a little bit about demos here. We've got to get back into it. We've got to get back into it. All right. Yeah, I'm going to go a little bit of different order because I'm generating something and it's taking longer. So my first demo, actually, Jaycah, I don't know if you've used this yet is, let me, let me get my chat GPT open here.

And I don't know if you've tried this. So let me get my share going here. So they have a new mode here, and it's called ChatGPT40 with Canvas. Have you seen this? I have not used it. I have seen. This came out this week. Yeah.

Rate me an itinerary for a one-day tech conference in Austin. I'm just going to do it just for you, J-Cal. And so what you're going to see here is, so now what it does is it on the left-hand side is your chat, GPT, and you can do this for writing code and a bunch of things, but I think this is much more applicable to folks, right? And you can basically take a section of it, highlight it, which I'm going to do here.

Let me just highlight this properly. How did the chat window move over to the left? Because that's what happens when you use this mode. And now, so for these two sections here, I can say, let's just say, make these more about intros and a special guest. And so now, see what it'll do is it highlights it right in place. Oh. Yeah.

And so if you're working on something, basically what it's doing is it will highlight, it'll let you work on the document while you're kind of editing it on the left-hand side, which is really cool. So weird that mine's not doing that. It's not putting me on the left-hand side for some reason. No, but are you doing 4-0 with Canvas? I am on 4-0 with Canvas, but it's not showing me the left-hand... I don't know why. Maybe I've got to log out and log back in, but

this is really cool because it's editing the same document. Yes. This is what we've been waiting for because it's constantly, yeah. So this is great. This is a 10 of 10, obviously. This is absolutely wonderful because we could put in today's doc or like a news program doc and make it shorter, make it longer, add something to this, and something to that. I did something to edit it, put it in a table, yeah. Yeah. And so, So, yeah, we're starting to see the, this is what we saw in Blade Runner when she is asking to analyze some video clips and he's zooming into the left and the right, and it's responding

in real time. That was a vision from Ridley Scott, you know, from the 80s. And, yeah, this is fantastic. Absolutely 10 of 10. Yeah. Just great. Yeah. Right.

And then this is daily use. This is use all the time. This is much better to iterate when it comes to, you know, you know, you know, you know, you're usability. And so you give it a, give it your grade, J-Cal. I'm giving it a 10. 10 of 10. Oh, 10. Yeah. Yeah. Okay. All right. Yeah. A-plus for me. What about for you? Yeah. Yeah. Yeah. I mean, I look, this was the one last thing that I always kind of struggle with with Chb-T is like, oh, you're always having to copy, paste or rewrite the whole thing. And now you can edit it in place. And so I want to make sure you get it enabled, J-Cal. Like, I feel for you right now. I don't know what. Well, you know what it might have been? It might have been that I didn't wait till it was done.

Like, when does it actually open up the left? It kind of starts, I mean, I can do another one. We can just do another one here. What were you trying to do? I'll do your one here. Well, I just said, you know, create an itinerary for going to Tokyo and Naseco to go skiing this winter. And it gave me like three days in Tokyo and four days in Naseco.

Yeah. But it didn't put it on the, oh, 4.0 with canvas. Nope. Try. Yeah, so see, it's kind of, it's interesting. So for this one. Yeah, so it's open the left-hand side. I wonder why it's not doing that. Hmm. Somebody from, uh, one of our friends over there explained to us why it's not opening the left-hand chat for every one. Yeah. So, yeah, because I, you just had that happen. Yeah, it just happened there. But it's now it's not doing it for me either. Weird. Weird. Very strange. Well, anyway, this is absolutely fantastic. And, uh, I think this is the game. changeer, yeah, that I've been waiting for. So congratulations to the team over there. And, you know, just using 0-1 all the time and seeing it show its work of what it's doing

has been extraordinary for me. I have been doing so many interesting. And like actually speaking of what we're talking about here, I was looking at, you know, I was having these conversations with people about, let me share my screen one moment. So I was having these conversations with people about car production. Because I was trying to figure out how fast could we actually make enough cars to do all the rides in the U.S. and who makes the most cars? And I got it.

Did a table with this. And, you know, then it was missing some companies. So I was like, hey, where's BID and all this? They make 4.5. And then I said, well, where's Tesla? And I said, well, Tesla's making, you know, put Tesla in there at 4.5. And I was like, wait a second. And that's not the number of cars they made.

They made 1.8. So is that right? and then I was asking it to, you know, correct itself and it did memory updated, et cetera. And it's really like coming together. And then I asked it how many cars are sold in the world per year. And it told me 70 million cars are sold per year. So then I was able to back into, well, in the United States, we need like 50, 60 million cars in order to have a fleet big enough to do all rides. Because I was like, what's all rides?

All rides is probably 50 to 75 million of these cars because you've got to realize some are going to be in a dense urban area. will be in the suburbs where, you know, you might do an hour-long ride, and then it takes 45 minutes to pick up the next person because it's across town or whatever, you know, it's or half an hour to pick something up. Long story short, it would take every car manufacturer in the world just to make enough cars to do the United States. And then on to the next country, onto the next country, you know, it's just not going to be easy, is the point. The world of tech startups is a whirlwind. Man, you got to navigate financial management, global expansion, and of course, strategic growth.

It can feel like plotting a course through uncharted waters, but fear not. CLA is here to cut through the complexity. From entity selection to mergers and acquisitions, they'll elevate your startup journey. Simplify your financial management, and they're going to be your navigator. CLA offers a comprehensive suite, including industry-focused wealth advisory, digital solutions, plus audit, tax consulting, and even outsourcing. What sets CLA apart? Their expansive footprint.

They have more than 8,500 dedicated professionals across nearly 130 U.S. locations, all driven by a global vision. Their promise is simple to know you and to help you. CLA is more than an accounting firm. They're your partner to success. So you need to have a trusted advisor by your side. One is going to help you navigate through all these important tax issues, get your accounting tight, and do everything in between. It's time to take action. Visit cLA connect.com slash tech.

And don't forget to mention that your boy. J-Cal sent you. That's CLAConnect.com slash tech. Start today. Jaycoe, we can do your demo. Try this. So try it again. Okay. And say use canvas, if it's not. I just was getting annoyed, yeah, and then see if you can do it. Okay, let's see. I'll pick. Or use Canvas mode. Use Canvas mode to. Well, pick the Canvas mode, type it, and then if it doesn't do it, when your next prompt, to say use Canvas mode. Use Canvas mode to explain. the total number of car rides in the United States, give us citations as well. And let's see if it does it. Oh, it did it. There you go. So, you know, and this is what I was trying to explain. Yeah, sorry. I've got to stop sharing. So, you know,

$344 billion rides annually. I mean, it's just nuts the number of rides that are occurring. And, you know, if you say 344 billion, it's 365 billion days a year, I think about a billion rides per day. And so, you know, now if I went to the side here, let's make a financial model how many automated cars would be needed to service all these rides. All these rides. Okay, let's see what it says. Let's see if it doesn't. So here we go. It's giving us a total number, the total number done.

Average rides per day. Assuming a car can put 30 rides a day, that's actually a pretty good estimate, by the way, because that's what an Uber driver can do is like 15 to 20, so two shifts, you could kind of get there. Days in service a year, this assumes they operate 365 a day. That's a big assumption. We're probably going to do that. Total car rides, per car is, you have to bring them in to change tires, to charge. And the charging time, as you know, is to fully charge. Well, maybe if you're doing a supercharge, it'll be less, you know, but it's still going to be sometime. So it's saying here, approximately 31 million.

And I say, let's say, assume cars are on the road 80% of the time and charging 20% and that they cost $40,000 to buy and, let's say, $10,000 a year to insure and clean. And then we can, like, let's see if they can do this. Now, this is something you would be talking over the shoulder to an analyst, right? If you were in a venture firm or whatever, or you were at Morgan Stanley, you're an analyst. And you start, now you're here. You've got the financial model says, cost per bar 40. annual insurance 10, total cars needed $40, $39 million, $1.5 trillion to purchase the cars.

$1.5 trillion is a lot of money, folks. Just to give you an idea, I think Apple may be sitting on $100 or $200 billion at any given point in time. And so the operating cost would be $400 billion a year. So, you know, it's not easy, right? and what would the revenue be if each ride was, I don't know, let's say, $20,15. Now we can start to see what the revenue would be here. $15 ride seems reasonable for a half-hour ride, and you can start to get to, let's see. revenue 5.6, $5,000.

Oh, my God. That's $15, right? So then you take $15, you take out the cost of buying all these cars, 1.5, the cost of maintaining them $400, and you can see there's a big business to be at here. Who knows what the margin would be, you know? But this is incredible. I mean, wow. What I love about this is, when we were talking about crypto,

we were trying to figure out the use case, right? You were in crypto only for five years, right? you have your, you have almost, you lost a half decade to crypto. No. I mean, you did okay. You made a bunch of money on your expenses. But my point is, as a technology, we put all this effort into it. And then you look at what is happening with AI.

Like, this is actual real value. If you're an analyst, I had to have a whole talk with my team the other day because they were giving me work before doing this. And I was doing this while I was asking the questions. And I'm like, your work sucks. chat GPDs is awesome why don't you use chat GPD and then come to me with your work

after you've polished what chat GPD has done and they're like are we allowed to do that? Is that fair? And I'm like, fair. What are we talking about here? Like let's kick everybody's ass into I think that's like going from typewriters to word processors. Like, oh, I can't use I'll check. Yeah. I got to I mean like literally people are taking white out.

People don't know what white. Look up what white out is. There was a huge business in what they called whiteout, where you would take a little bit of a nail polish and you would, you know, they misspelled. If you did a typo, you would put whiteout in and then blow a drive and then retype over it. Okay, we're not in whiteout mode anymore. This is a 10 of 10.

I could sit here with you all day and come up with the examples. And now imagine this thing could put images in there, you know? Like, and then I could say here, put this in a simple table. You got to share it. Got to share it. Oh, sorry. You're doing demos. I know.

Sorry, I was trying to be gracious here with you. Okay, put this into a table. Yeah, that's nice. Make an infographic. Do it different. Do it slightly different. Okay. What I'm seeing is you can highlight like a little paragraph on the right hand side.

So just highlight, say, see the paragraph that says that, no, no, on the right hand side. On the right hand side. Okay. Yeah, highlight that paragraph is just total number of rights. Just highlight that paragraph. right up a bit. Oh, okay. Yeah, highlight that.

Now, wait, wait, wait. Now, see, use as a shit, GPT, you can say, make this more spicy or expand on this. So just, you know, say something. Yeah. Make this more concise and, yeah, make this more concise. Let's see what it does. Yeah. And see what it'll do is it'll rewrite that.

Oh, wow. Boom. This is the superpower, J-Cal. Now you don't have to do it on the left-hand side. You can just go to each bit and you can say, now you can say, oh, you can say, oh, you can say, oh, put an image of that, so you highlight it again and think, can you turn this into a graphic, right?

Of the same paragraph. Say, I want to see this as a graphic or something, right? Yeah. Pretty amazing. I mean, this is what we've been looking for. Well done to the team over there, A plus, A plus. I've been waiting for this too because it's so obvious that this is not a difficult thing to do.

And this is, you know, I gotta give the team over at Open Eye credit, and I gotta give the people at Gemini a bit of a warning here at Google. This entire workspace is dedicated to the UX of ChatGPT. You can't do this with Gemini because you shoehorned it into Google. Google should buy chat.com, the domain name from whoever. You've been saying this. You've been saying that.

And make a dedicated new service. You can leave Gemini inside of search, but I need a domain name I go to that's different than Google that does this. And I can then, it could be power by Google. You know, it's from alphabet, chat.com. and then have a great app that complements it on iOS. Can work on the iOS app first. We'll start releasing stuff to iOS first.

You get those top users. Everything is hidden inside of an app. You ask somebody, how do you get to Gemini? They're like, I don't know. Somewhere in my Google interface, I got a hunted pack for it. 100% of the user interface must be dedicated to the AI. But wow, we're quick with all. Give us another demo here.

This is a great one. Okay. Thank you. Yeah. Awesome. And you're going to, it's getting daily use. You go, the weekly, active, daily active, remain solid for Open AI.

Okay. Three or 400 million people a month is what I heard. Yeah, some 250 or 300, some really massive number. And a big portion of those paying 20 bucks a month, which is also incredible. Okay, we're going to just give a quick shout out to Meta and a couple of demos real quick. And we're going to start with one of their models,

which we have running on Grox, so I'll just show it there. And so what meta released a few weeks ago, two weeks ago now, was what they call vision models. And so I'm going to give a quick demo here. Let me get my screen share going properly. And I just did it before, so I'll reset it here. So let's just clear this out.

So basically, I'm going to use the 11 billion. So it's Lama 3.2. It's called vision. And so what you can do is, you know, you can give it like a picture of anything, and you can give it a task, and I'll just kind of turn this down a little bit. So I just said, so for the folks listening, I gave it a picture of a plate of food and said, tell me everything that's on this plate, estimate the portion size and calories. And you do that. And basically, what you'll see here is, and you see how quick that is now?

So this is what's starting to happen in this universe, right? Which is we're getting the same models. But instead of, you know, this used to take like 30 seconds, 45 seconds, and that's happened in 300 milliseconds. Right. And so now you have to think about this combined with glasses and other things, Jake, how this is the era we're going into where you can be looking at information and the processing of it's even faster than our own brain. There's a really interesting thing.

It's really good that you brought up interface on the new Apple AirPods. And I'm not an AirPods guy. I like the pixel buds. But in this new version that I just got the fours, with noise cancelation. Same here. Not the anterior ones, but the fours are the noise cancel.

I got the same ones. Which I prefer. Yeah. There is a very interesting feature. When Siri is going to read you something, it says you can shake your head up and down to say yes or left and right. I saw that in the settings. Okay.

Yes. And so it's really compelling because it will say, Sonny just texted you a long text, read the text. And I go. and when I just nod, it reads it to me. Yeah. Yeah. Or you say no.

Or it's like, hey, because I use the walking thing, I like to track my walks as exercise. And it says like, hey, would you like to end your outdoor walk? And I just go, nope. Or I go, yes, and it ends it. So this is becoming really compelling. People are going to start doing shaking of their head or they'll have something on their wrist to kind of do small gestures to interact with. the glasses that we saw from, you know, meta, and you put this all together, we're going to be

constantly interacting with data and information if we want to. And it's going to ruin social settings and everything, but it's going to make people incredible. Like, the idea that in a poker game, you could be wearing glasses. Well, they're going to have to ban them. You're going to have to ban them. I think we're going to have to ban those glasses in restaurants, obviously in gyms. And I think, you know, I haven't been to a public gym in a long time. I go to my YMCA here in, in Austin. I like the YMCA. It's great.

Well, it's good fun for families and kids and stuff. But, you know, like they ban, I think you can't have your phones out in the locker room, I think. But, I mean, these glasses, that's a whole other level of, you know, invasion of privacy. And then also distraction. So do you want those glasses? Do you want the AR glasses? Yeah. Well, so, look, I think the era of us, you know, getting superpowers, because, you know,

other than getting like the brain implant and everything can't be done over audio. So what I kind of think about is, what if you're wearing those glasses at work? And then it's just analyzing everything you're looking at. So instead of you having to kind of keep telling your, you know, employees to keep using chat GPT, it's just automatically like looking at whatever you're doing and do it. giving you guidance and advice on it. Like, hey, I can help you make this better. Or what do you think about this?

So I think, you know, it has huge implications just in that setting. And you know, like, there's a crazy stat someone told me this week. You know what percentage of people like in America wear glasses? Yeah, it's like 60 apparently. So, yeah. And so it's, yeah. And so, and then, you know, naturally everyone gets old and you just wear glasses too, right? So it kind of adds up.

I mean, I wear readers because when I'm doing the show, I want to be able to crisply read it and if I don't, I have to strain. And so these 1.5s or 2.0s are... I'm the same. Yeah, I'm the same. One point. What did you think of the hack?

I had the two Harvard kids who built the facial recognition that would then go look up in public databases information feed it to them. And then they were like kind of tricking people into thinking they knew them. You saw the demo. I had them on this weekend startups

this past week with Alex. What are your thoughts on that? Look, I think like when people do that kind of stuff, it's like art of the possible, right? It's not like a practical thing because, you know, at the end of the day, the people that make those platforms won't approve apps like that. And you can do all these bad things like you can sit in front of a store

and capture everyone's Wi-Fi. Like you can do all these like nefarious things all the time, right? I think really what, look, you know who I want to have that? Is like the security guys at like a baseball or football game. Totally. Yeah, no, I know. If we started having, if we had terrorist attacks, God forbid, at a stadium, which has happened in the past in other countries, you know, people would be like, you know what, we need this facial recognition. James Dolan has famously installed a lot of facial recognition at Madison's work garden and little controversies there on the margin. But even, you know, I have a big security system at my ranch and it knows faces. And I put in everybody's face who works at the house, like we have a plumber or an electrician. When they come to the house, I put their face, I put their name.

Hey, that's Carlos. Hey, this is Jane. Whatever. And then if somebody comes on the property who we don't know, it tells me. And I get a little alert, hey, there's an unrecognized face or a license plate on the property. So it knows certain, you know, trash pickups license plate and others. And that's all coming to corporate software. and then it's going to make its way down to consumer.

Yeah. All right. So meta's new vision models, what's your grid? I mean, yeah, I got to give it a B plus. I mean, it's really solid. And it's going to get better and better. I mean, this, what do you give it? I mean, look, I think the fact that it's open source for me is A plus, right?

You are now able to do in open source what you can do with the best models that are there. And I think that's better for society. So I give them an A plus for putting vision models out there. And look, this. This, think about what it took to make a calorie counting app. J-Cal, probably someone came through your incubator at some point. We did. We did.

We did it in 300 milliseconds without. We actually had one years ago, like a decade ago that pitched us that was using people in Manila. Yeah. You know, like mechanical turking it. You would send the image. They would look it up online and they would make their best estimate and give it to you. And they would do it for, you know, two cents or five cents through mechanical Turk. And people were paying 50 bucks a month for the service.

So, you know, they have. Even if they abused it and used it 50 times a day, it still wouldn't match their subscription price. Absolutely fantastic. And let's pause for a second here. What does this mean, you know, that the language models are catching up to what Open AI is doing, the leader in the space, very quickly. And there doesn't seem to be that much distance between the models, I would say under a year, six months to 12 months behind. Yeah, six months. But the interface seems to be where Open AI is making really great advances.

So it's really fascinating, and it's a good point. You've highlighted it, right? I think what they've realized, and I think their revenue was like leaked, and it showed actually more of their revenues coming from chat GPT than their API. Basically, what I think is exactly what you said. ChatGPT is, you know, an open eye are becoming like a consumer-facing application, whereas, you know, meta, and we're going to do a couple more meta-demos in a second, is focusing on the fundamental technology and integrating it into their interface,

which are consumer-facing applications. So what you're seeing, and for them putting it out there in the open to make it, the community to make it better is important, but their core business is not to be selling, you know, tokens as a service to people or selling intelligence, right? And so that's the interesting intersection that I see, you know, or maybe crossroads is a better word, across what Chad GPT is doing now. And, you know, if you take a step back for all the, you know, let's just call it chaos for lack of better word that's happened at OpenEI. If you think about everyone that's left for a second, JCal, right? The people that have left have been pretty much like were focused, like not on, like, you know, they were there as like kind of core research types, right?

better for the world types, for real Open AI. And as Open AI commercializes, it's looking like a different business. And so... It's looking like an app business. It's looking like an office suite or something. It's starting to feel like productivity software. It feels like a Salesforce. Yeah, if you're an AI researcher, maybe you're kind of like,

oh, well, I signed up to work at Open AI and build intelligence for the world. And now these guys are like, we're just going to build like, you know, commercial product, which is totally fine. Yeah, and with all this money being poured into it, it feels like this is getting commoditide faster than anything I've ever seen. That's why them having the users is important, right?

Because I think, you know, having, you know, 200 million paying users, yeah, you got to give me credit for that. I guess so, yeah, that is hard to do to get people to take their wallets out. So a couple of million paying users is pretty impressive. But the other point of it is, like, I think, you know, using Siri, I've been using the beta, and I've talked about it on the show and all in a bunch. I feel like they're going to intercept the first 50% of queries are going to be done by Siri, Alexa, Microsoft co-pilot, Google,

etc., people who already have scale on their devices. But you're going to expect that for free versus what we were just doing. You're going to willing to pay for that because you're like, oh, this is my workflow versus I'm just, you know, if I'm doing Q&A, I just want it for free. At 20 bucks or 30, I think we have the enterprise version. So maybe we're paying 30 bucks a person and maybe half my people are using it. And I think I'm like if you're not using it at my companies, I don't, you're not going to be on my companies for long.

It's going to become obvious that you're falling behind. For $360 a year, if your average employee is, 50, 75, 100, pick your business, 150, depending on where you're located in the world. 1% or 0.01% or 0.1% of your employee salaries, nothing to pay. It's kind of grossly undervalued. If you told me this product was $500 a month, I would pay for five seats. I wouldn't pay for everybody. I would make sure they're using it, but I would have five people on it. I mean, things like the Wall Street Journal cost $400 a year or something. I don't know what it costs, something crazy.

So this is for all that stuff. That's something can add up. Oh, yeah. This is providing serious, serious value. And I guess Canvas and 01 are not available for free. So yeah, I do think there's a case to be made. But if you're a Microsoft or Google, I'm making all this free. I'm taking everything chat GPT does.

It's the teams. You're going to do the teams. And you're going to do the work. Yeah, I'm going to bundle it and make it for free. I'm like literally going to make it super for free. and just, you know, take the oxygen away from them, right? So that people who... But right now, this, you know,

there's no domain name for me to send my team to for Google to go use Gemini like this, right? And the multiplayer mode is the thing that I want. I want to use Canvas with you or with another person. So when we're building the docket, we're working on it together. That's got to be close. They can't be... Multiplayer mode.

Yeah, yeah, it's going to be super close. But you can already share it with each other. It just... But then it gets forked, right? When you share a discussion, it gets forked. And did you notice when you create them now, it asks you when you share it? It asks you if you want to publish it.

This is going to be crazy for SEA. So when you share a link and you copy, there's a little chat, but there's a little button. It doesn't work with Canvas, but with the regular one, yeah, it does let you share it publicly. So that means the Google Index is going to be filled with conversations. I mean, it's wild. What about the biggest hack, right, to get more traffic?

in is basically just open your index of questions up and then basically everyone's coming to you. It's so smart. It's all the work that went into Cora is now these these guys scrape the Cora and they're going to republish Cora a thousand times. I mean, if I'm Cora, I'm suing

the hell out of them. If they, I mean, I don't know what they're paying for. Maybe they have a deal. Maybe they have a deal. I think this just guts everybody who's got a business. Yeah. Okay, let's do another time. Okay, we're going to keep rocking. We've got a few more here. Okay, this actually finished. So I'm going to go back to, and I don't know if you guys did this

already. So if we did, we'll go quickly here. Yeah. So I had this like MIT management fall 2024 review. It's just like somebody who sent it to me on WhatsApp. And so basically I didn't want to read it. So I just dropped it in notebook, LM, and it turned it into a 13-minute podcast. And so basically, so like I'm going to just kick it off. We'll listen to the first. Hey, everyone. Welcome back. So it seems like you're all about diving deep into how organization you're tackling change right now. Those MIT Sloan management review articles you got stacked up.

NPR voice. Yeah, we see you. Don't worry. This isn't about drowning you in information. We're here to pull out those golden nuggets, the insights that will really get you thinking. And maybe spark some ideas for and here's my female co-host because we're woke and this is NPR coming to me.

Parallels there are. All right. Okay. So, Jacob, I actually like, I love it. I love it. I think it's right. that mode yet? Have you tried it? I haven't used it yet, but I think that it's fantastic that one of the modes of consumption will be make this into a conversation that I've listened to. Yeah. Fantastic. Like, I want to learn physics, but I don't want to learn it. Like reading the textbook, or I don't want to do Q&A with it. I want to listen to a podcast about it. And so,

Jake, the hack that people are doing now, and I'm going to tell you this, is they take their diary, right? And, you know, some people diary at different levels. And then, or, their feelings about a certain situation. So let's say you and I are having a conflict, right? And I go and I write myself like a three-page letter of like, these are all the things that like I love about Jake out of this. Yeah, making fun of me. Yeah, exactly.

Took my money and poker. Yeah, yeah. I write it all down. And then they put it into Notebook L.M. And then they have two people talking about. Oh, my God. And they listen to it. And then you get like a third-party view of your situation.

It's really incredible. I mean, think about this, like, for a coach kind of situation or a therapy kind of situation or a mediation situation. It's kind of in that zone. And I had a lot of people pitch me on coach, not coaches, but like therapists, like virtual therapists. And we had some spicy ones that were on earlier, like a year or two ago. So there is something here of being heard is a human, very cathartic for humans.

When you feel heard. And that's why I think, you know, if the AI can. can pull this off in an authentic way, and it makes you feel heard, that could lower people's anxiety where they're like, hey, I'm having feelings of jealousy. I'm having feelings of anger towards my boss. He's a jerk. His expectations are high. He's giving me a hard time about AirPods and all this stuff and how to do like for a production or whatever. Yeah, my boss is in full of things and stuff or whatever. Like, it's like, oh, yeah, you know, but your boss just wants you to get the best out of me. You know, the company just wants to produce great content.

There's a reason why, you know, yeah, okay, now I can sleep at night. You know, like, it's kind of cool, actually. I have no judgments. Like, I think it's kind of cool that people can find. So I want you to try it this week, Jake. I want you to diary one day. Okay. And I'll put it in there.

And then we'll do it next one. We'll do it, actually. Yeah. Absolutely. All my feelings. Yeah, from one day. Do it after you record the all-in pod. Absolutely.

You're doing great. Oh, I give that one a B-plus. I think it's like a cool feature. I don't think it's gimmick. You're going to try it next week. You're going to try it next week. I will try it. But I'm giving it a B plus because I don't think it's a gimmick.

And I think there's, if they did podcasting, which feels like a gimmick, they could also do a therapist. They could do a coach. They could do a best friend. They could do your mom or dad. And that's where I think this could get really interesting is if you took whoever the great advisor or friend you had in your life was and then you capture them, like, I mean, that could be very special for people. You know, if your grandma always gave you the best wisdom and advice, and then you said, hey, here's everything grandma, like, and grandma was one of us with a big online footprint, right?

Or you just had AI talk to grandma before you pass it for an hour a day and just asked her questions, and then you could say, hey, I got grandma forever. Yeah. Like, I'm actually getting a little sad about it, thinking about my dad, my dad's sick. And so I've been thinking about it and, like, him being gone and whatever, yeah, how long he's got left. Yeah. How long he's got left. And I'm actually, I'm actually getting a little bit of. I'm He's 81, you know, it's a cancer survivor. Like, I'm like, ah, my mom and dad, I want to capture it. So I was talking to my little brother, like, hey, maybe we should do like an interview series

with my dad, just record it, so we have it for our kids. Just record, just do it. No, honestly, just record it. And then, you know, not, the arrest, turning it into AI and all is super easy, but just record it. Just go and do a couple hours and you'd be surprised. Yeah. And do it as in like, I would say, do the history of the family

from his perspective and then do, you know, like, top, top big moments of his life. I need the prompts. I need like to get the proms. the prompts. So I think that's something where I... I asked CHTPT for the prompts, dude. I know. I was just thinking, ChatChapT-C-C-T-canvas would do it for me and get all the prompts together. And then I could actually have chat ChpT do it with him.

That would be very interesting too. Yeah, I think... This would be a great business. I don't know why this doesn't exist. I think some folks will do it. But like... Yep. But what I know, what I think you could do is just give the prompts to him and then have, you know, some combination, you know, of Josh or someone. Just go sit with him and record him.

Like, he can just read out the questions, so he doesn't have to come up with him. And then he answers them and then, you know, I'll help you do it, Jacob, if you want to do it. I think it's a great idea. Last two, last two. Okay. So, meta, meta has done two things recently, which are as follows. The first one is they haven't released this for use yet, but they've created a movie gen model. And, you know, just to kind of show you like what it's capable of, it's, you know, pretty else.

Yeah. And look, like, you know, we've seen this stuff before and this is tied into some of our bets, but like I think this is just getting better and better and better, right? The fidelity is getting there where if you weren't paying attention, you would think it's real. Yeah. And obviously a koala surfing, you wouldn't think is real, but the stuff that's not koala, you would think is real. Yeah, like, that's pretty good. And, you know, we're just kind of looking at some interesting videos here.

And then you can edit them with text, right? You can have this video and you can say, like, hey, add pom-bombs to this or replace some of the dinosaur. And then, yeah, this produced personalized videos. Like, a picture of yourself saying, hey, make me, you know, scientists do it. with a beaker or whatever it is. Soundtracks, sound effects, pretty incredible where this is going. So it really shows the power of them and having lots of data like they have. And they haven't released this to everyone, but it's very, very exciting.

So maybe we don't grade this one yet because it's not released, but it kind of shows the power of what's potentially coming our way. And that's Meta's video long version. Yeah, it's called Meta Movie Gen. Meta Movie Gen, yes. Amazing. I think Zuckerberg's, you know, really doing something special here with this open source stuff. I think we're going to look back on this and be like, hmm, this is, he made a strategic,

he made a very strategic decision that I think is going to pay off massively for him. I completely agree with you. Okay, last one here is. And that one's a B plus two. I mean, I, okay. But the processing speed of these is the other one I'm wondering about. It's like, remember they were all limited to five seconds? Have they gotten past that now?

No, they haven't released them yet. I think that's what they're all kind of working on. And so the last thing here is like meta-AI, I don't know if you use it, J-Cal, you should. And it's kind of like their chat GPT. Now, it's designed a little bit more for fun than it is for, I think, kind of like the workflow stuff that we were talking about.

But where it's super powerful, and I'm just going to do this, I got a picture of you. And what you can do is you can give it a picture. And this is like a more modern one. And then you can say, edit this image, and you can say, make me bald. Right. And what you'll see is it's like so powerful that like, you know,

all these image things that were coming out the last couple of years, they've combined all that functionality and basically just, you know, this is like J-Cal's Mark and Mark Andreessen. I don't actually, I kind of look good bald. I mean, it's a look. I look like I'm. Yeah, yeah. Who is that guy who played the crazy guy in the M.

Shaman, M. Night Shyamalan series. There was like glass and then there was the other one. And then there was just guy who plays like a guy with multiple personality disorder. I think was James McAvoy playing him. But I look like Professor X. I like it. I like my Professor X. Yeah. Yeah. So, yeah. So anyways, like, look, that's what you can do here. And you can say, you know, as you can say, give me a beard. All right. Oh, God. Here we go. Yeah. I like that. I can never grow a good beard. Yeah, for some reason it's not. Let me go back and do it. Let me, it might have to just re-upload or restart again here. Yeah.

Yeah. That's hilarious. Yeah. And so, yeah, so you can say, but it's, you know, it's like, yeah. It's going to be a lot of fun. I mean, and then I think they're going to have to start watermarking these, correct? Because, like, what is reality? Like, we don't meet people watermark face tune-up or those tune-up things that I know.

No, we don't. We don't. Yeah. Yeah. So, look, I think, you know, and you can do all kinds of fun stuff with this. We're just playing around here. I give this a B. Solid B.

Yeah, but, you know, they've taken all the image editors and all that. Oh, that is not a, so that one is a good job. I mean, that actually, I look like a headler. Like, hey, would you like a soy latte? I feel like I'm making you a crab cocktail. If you put suspenders on me. Craft cocktail, crack cocktail. Yeah, I'm going to make you a smoky, old-fashioned with a lump of

sugar in it and whatever. Yeah. Yeah. I'm going to make you an overpriced cocktail. Let's go in Austin. There's a bunch of those places in Austin. Yeah. There are. They're big into cocktail culture here. And that mustache and beard and that mustache. Yeah. That's a waxed mustache. I look like a supervillian. Exactly. Exactly. Yeah. All right. All right. Well, this has been an amazing episode. Welcome back. Let's do it every two weeks. You got to store all these up and we're going to get dialed in here. And he is Sundeepe, X.com slash Sundeepe. I am X.com slash Jason.

And let's give a little promo here for Grock. You got tons of developers. Anything coming up on the calendar? 500,000 developers. You know, we have the fastest automatic speech recognition. So look, Jacob, one of the things that we can do, if you haven't done it, is we can take almost all of, we could take all of this weekend startups. 2,000-plus episodes.

Yeah, and we could, and we do it at 250x real time. So basically, you know, whatever time frame that is, so let's say that's, you know, like how 2,000 hours divided by 250, and that much time we can turn, we can get transcript forever. Love it, 10 hours. Yeah, we can do it in 11 hours.

Yeah. Unbelievable. All right, everybody, we'll see you. And where can people find out more about GROC? Just go to GROQ.com. GROQ.com. All right, and we'll see you on next time on this week in startups.