This Week in Startups - AI Demos: Gemini Gmail, Search GPT, Groq with Voice & Playground | E2001

Episode Date: September 4, 2024

Timestamps: (0:00) Sunny joins Jason to jump into AI demos. (1:21) Analysis of Q4's impact on startups and Nvidia's market performance (5:09) Chip design advancements and AI prevalence in startup technology (8:12) Experiences with ChatGPT-4 and Google's Gemini in Gmail (10:14) Squarespace - Use offer code TWIST to save 10% off your first purchase of a website or domain at http://www.Squarespace.com/TWIST (11:36) Discussion on generative AI: filtered vs. unfiltered content (19:50) Vanta. TWiST listeners can get $1,000 off for a limited time at ⁠https://www.vanta.com/twist (20:03) Importance of SOC 2 compliance and AI's role in LinkedIn (22:42) AI integration in Gmail and throwing AI at business problems (25:28) AI personalization tools and grading AI email integration (29:01) Micro1. Visit ⁠https://www.micro1.ai/twist⁠ to open a talent search and get a 2 week free trial per hire. (32:22) Deep dive into Search GPT functionalities and Google's AI search evolution (39:08) Practical AI search tests and Yelp's data integration (43:17) Ethical considerations in personalized AI technology (45:26) AI-powered product recommendations and monetization of AI models (51:44) Legal challenges in AI and generation of web apps (56:15) Exploring practical AI applications and Claude AI's app development (1:00:14) The era of building your own apps and fast image generation tools (1:02:12) Comparison of Playground and Canva, impact on graphic design (1:04:28) Groq's itinerary creation and the future of user interfaces (1:07:14) Closing remarks and acknowledgments Subscribe to This Week in Startups on Apple: https://rb.gy/v19fcp * LINKS: Check out Gemini Gmail Chat: https://workspaceupdates.googleblog.com/2024/06/gemini-in-side-panel-of-gmail.html Check out Search GPT: https://chatgpt.com/search Check out Groq with Voice: https://groq.com Check out Claude: https://claude.ai/ Checkout Playground: https://playground.com/design * 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: (10:14) Squarespace - Use offer code TWIST to save 10% off your first purchase of a website or domain at http://www.Squarespace.com/TWIST (19:50) Vanta. TWiST listeners can get $1,000 off for a limited time at ⁠https://www.vanta.com/twist (29:01) Micro1. Visit ⁠https://www.micro1.ai/twist⁠ to open a talent search and get a 2 week free trial per hire. * 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

Chat GPT has this new function called search GPT. And so you can say, tell me about the... Oh, are you in search GPT? Yeah, yeah, yeah, yeah, yeah. I didn't let me in. It's only the waitlist is over. So this is perplexity from them, right? Yeah.

And then if Open AI puts a revenue model on this, like links out in an auction, you would have stolen somebody's content and then slapped a link on it. So let that sink it. This isn't monetized yet outside of the fact that you're paying 20 bucks a month to remove all the ads and have a cleaner experience. But in a way, they are monetized. it with that subscription. So when this search engine comes out, I think you're going to see 10 times as many lawsuits against Open AI. So there's a reason why they didn't open this to everybody.

This is Lawsuit Central. This week in startups is brought to you by Squarespace. Turn your idea into a new website. Go to Squarespace.com slash twist for a free trial. When you're ready to launch, use offer code twist to save 10% off your first purchase of a website or domain. Vanta. Compliance and security shouldn't be a deal breaker for startups to win new business. Vanta makes it easy for companies to get a stock to report fast. Twist listeners can get $1,000 off for limited time at vanta.com slash twist. And Micro 1. Micro 1 is an AI recruitment engine to hire world class engineers fast. Visit micro 1.a.ai slash twist to open a talent search and get a two-week free trial per hire. Welcome back to this week in startups. My guy, Sandeep Madre is here.

Of course, from GROC, you can go check out GROC at console.g.com if you are a developer. It's the fastest and cheapest inference in the world. And Sunny is an expert on AI. So we like to do a little catch-up here every week or two to stay on top of everything that's occurring. And we talked last week about open a new tab automatically open up chat GPD4. Man, it's really good. Chetachyp4 is so good. Yeah. I can't believe how good the results are. Well, we used

Anthropic to create our own plugin to do that so we don't have to. Yes. Yeah. So it's, we're kind of getting it all around these days. Yeah. Yeah, no, Anthropics pretty great too. How are you doing? How's, uh, how's work on? Everything's excellent, you know, really busy, um, you know, just heads down. This is, we said the second half is going to be crazy and it's already started that way. And, you know, Q4 starts in October, but really It starts in September. I always tell my teams like, look, Q4 is our longest quarter because it's four months. I mean, it's not technically, but you get what we're saying.

Right. People come back from their August break and they're on fire and they want to get everything done in September, October, November before the Christmas rush. And we've been in business. The holiday rush is here before you know it. Kids are back in school. So everybody tries to get as much work done during those hundred days from Labor Day. Budgets, too. And budgets get set.

for the next year. So it's just a very critical. And so what's happening with the pace of new product, new demos? Are people working through the summer? Do you get the sense that people have a sense of urgency or do you think people are taking breaks? No. So I'm going to quantify it in the following way. So first of all, you know, there's that meme, which is like the entire like fangs being held

up by, by Nvidia. Yeah. Yeah. So like, Nvidia came through, right? They came through and they proved with Or, you know, huge expectations. They still even beat those out. And on top of that, they're like, look, our stock is so expensive. We're going to buy back $50 billion of it as well. And we're not going to go deep on the financial side. But like, that's what we needed to get this, you know,

keep the momentum going, right? Because that could have been, you know, had it been a miss or something like that, it could have been a little bit, you know, slowed things down. But that just showed the demand is still growing, which, you know, I think those of us on the ground know that's happening, but a macro thing can always be a challenge. The stock is, yeah, it's still up massively year-to-date, 150%. In the last year, it's up 142% over the last five years. It's up 2,800%. But, you know, it's kind of feels like their success has now been baked into the stock.

It's all priced in worth $3 trillion-ish. And they did beat the revenue number a bit. They still have a huge margin. They haven't added a ton of headcount. But there have been some delays in certain products. Maybe you could inform the audience about the product lineup at NVIDIA and what you're seeing, you know, get deployed and some of those issues. Yeah, I mean, look, they have a new chip called Blackwell.

Here's what I'll say it. And, you know, we're all in the same ecosystem, but they're moving really fast. You know, chip design used to be like a three to five year thing where it'd be like that's a cycle it was on. You know, think back through, you know, your days when you were, you know, deploying Intel CPUs, right? Every three to five years is when you get a new one. Yeah, a new Pentium chip. Yeah, exactly. And Intel also had this like TikTok model, right, which is like they would create a new chip and then that was a tick and the talk was like a,

you know, like the iPhone S kind of thing. They'd make like a, and so these guys are coming up with brand new architectures and moving really fast. And so I think they really, you know, made folks feel comfortable that they have chips in customer hands. They've made some changes for their process that will help them yield more chips. And so we're very excited by that. But close out on my thought, I think starting with that, I'm going to kind of break it down into the following categories. And I'll throw one to you, but I'll start. I think this, there was a for the, there was a summer Y Combinator bat. Yes. And man, like I, you know, I have it in my like Twitter and, or X. And I kind of feel like every company in a Y Combs is a company. And there's

There are some amazing stuff. Like I look at it. I go, holy cow, like, that's really cool. So I'm excited to see what's there. I'll throw it back to you, like, in terms of, you know, the companies you have going on. Like, how many of them are kind of AI native, right? And so half are AI native or more. And the other half are using AI to accelerate whatever mission they're on or using it to operate to operate their businesses.

And so, yeah, it's very real. Now, one of our startups that went through Founding University or Accelerator also went to Icombinator, taxed GPT and they're doing fantastic. And so, you know, unlike crypto, which I know you were a fan of, sometimes technology is, you know, super compelling, but it doesn't get to a use case. And I think AR, VR, VR, and crypto would be the leading culprits there of a nanotechnology, perhaps quantum computing would fall into this where you see the promise. There's some great foundational innovation that's occurred. But it doesn't actually change. a human being's life. And with AI, it does. So very simple. I'm, you know, I'm at the grocery

store last night and it's called Central Market. It's like, uh, H-EB is like hipster brand. It's quite nice. And I'm the person who's bagging my groceries. Ask me how my day was. And in Texas, everybody talks. It's a very charming and personable place. It's a pretty, yes, ma'am, yes, sir. I mean, I've never been called sir or said ma'am so many times in my life. Um, but she was, a student at business school and she was bagging groceries. Give her a nice tip, obviously. And I was on this conversation with her and I was talking to her about chat GPT4 because she was talking about business school. And I said, you know, do you know what you're going to do in business? I said, whatever you do, just get on chat GPT4 or every day. And she said, yeah, you know, people have

been talking about that a little bit. And I was like, how does the school not have their students on it? And I'm like, and I told her, I said, I'm going to give you one piece of advice. Just use it all day and long. She's like, okay, I'm going to do that. I was like, pay the 20 bucks a month to get it on your phone. Yeah. And I was like, wow. You gave her a $20 tip so she'd get the first month free. But basically what I did.

Yeah, you kind of nailed it as a first month free on Jacob. So the point is, like, I still think people are not using this, which is bizarre to me. People who are using it are bionic. And my lord, it has changed how quickly you can research knowledge and then deploy it inside your company or for your life. I started doing something where I was like. Research manipulate the knowledge as well. like, you know, sort of, and then, yeah, or not, it's the wrong word, but I guess, like, you know, work on it, right?

So I started a new thread with my chat chip E.T4O where I said, you are my economics tutor. I want you to teach me and quiz me every day about and give me reinforcement of what a good job I'm doing on, you know, different theories. And I want you to use really illustrative examples. And so I was just trying to fill in my knowledge because I didn't take economics, macro, micro, anything really in college. And it's been amazing. I've done like maybe three or four sashes where it gives me some information about price elasticity or, you know, just buzzwords. And then it gives me a quiz. Is this price elastic or not? And I'm like, this is so compelling. Can I give you another hack? Sure, please. And man, I would do it real time in a demo, but it's probably a little bit too quick. Go to Harvard or MIT, depending on the type of course you're taking. Find the course. It's open source. They've open source all of them. Put it into your choice. of Chetachin, and then walk through that with it. Very interesting you say that because the way this started was, you know, a couple of years ago when All In was starting, I felt like, wow, I'm just a little deficient on this macroeconomic stuff. So I did find MIT's open courseware. Yes. And I took the course on YouTube because all the notes are out there. And so that's literally what I did. But you're saying, take this page.

If you look here, principles of macroeconomics, and this is the coursework there. And it has. as the topics, the video, and these are the lectures I've been taking on YouTube. Actually, this is a real one. So this is the same course taught by somebody else. I guess they updated it, and they have transcripts. So you're right. This is what I'm going to do.

I'm going to take these transcripts, which I assume. Topics and reading, right? Go right there. And then, yeah. Like, yeah, I mean, boom, topics and reading even better. So it's a brave new world. And you can learn anything and you can learn fast. and you get your own personalized tutor.

I don't care if you're building enterprise SaaS, a verticalized AI startup, or the next great consumer app. No matter what you are cooking up, you have a brilliant website. It's the first thing that people see or read from your startup, and it has to be beautiful. You've got to make a great first impression. People do judge a book by its cover. It's that simple. That's why we have that expression.

But since you're building a company, you're strapped for time. You know that. And wouldn't it be nice to have some world-class help? How about a beautiful website built to your specifications quickly and easily? It's 2024. If you don't already know, HTML, CSS, and JavaScript, that ship has sailed. Get on Squarespace and check out their Blueprint AI product. You answer five questions, and Squarespace's new AI tool is going to build you a gorgeous website.

It's so simple, it's so slick, it's just brilliant, it's the future. Even better, you don't need to worry about what's under the hood. Analytics, SEO, payments, memberships, e-commerce. You can do it all. It's all included, too. So here's your call to action. Check out Squarespace.com slash twist for a free trial. And when you're ready to launch,

go to squarespace.com slash twist to get 10% off your first website or domain purchase that's Squarespace.com slash twist. What have you seen recently? I've been using GROC a bit, the other GROC, not your GROC, but Elon's GROC, to create funny images. And I noticed they don't have a bit of a filter there

so you can do some wacky stuff and some friends who are doing some spiky. stuff, not that spicy, but there's a big debate going on right now, I think, between unfiltered, generative AI and filtered. You know, what that's actually showing, and this will be a challenge is, you know, these laws that are being proposed, right, and particularly the one in California that is trying to regulate this. And the challenge that's emerging is, and, you know, we're going to see some demos of this

today is that it's so easy to do stuff now that, like, before you could, you know, you could take software and do good things and bad things with it, but the buried entry to do it was quite hard. And now it's just minutes on, you know, in a prompt. And, you know, the combination of those images plus software you could create plus plus plus, you know, it's really, you know, I still think that, you know, the industry like we talked about needs to self-regulate. You know, you had a great suggestion, like tag those images or put a watermark on them and say, this is like it's, it's, it's made by, you know, this image generator and it's fake. That would be a way to, you know, do it, right, is to just simply put a watermark on these if you wanted to. But, you know,

it's kind of like Photoshop. I think everybody should assume every image and video you see is fake until you've confirmed it's real. Same with audio. So how do you confirm that? Multiple credible news sources, and this is where news can play a role, have confirmed that it's real. So when you see something on, you know, social media, you should just assume that's fake until a proper news source has vetted it. Yeah, so here's what I think is going to happen. And it's, it's going to happen quickly because we're probably already at a ratio of like, for every real picture that's taken, like a hundred fake ones are generated because, you know, it's just easier to do. I think what we're going to see within the next, you know, and this will be longer, because it's not a technology thing.

It's like an industry thing. Photos that are taken by a device will be watermarked and that watermark will carry all the way through to the point of publishing. And you'll be able, because I think it'll be more about saying this is an actual real image. So, you know, like certificate of authenticity on like a bag or a watch or something. Sure. It'll be like that thing. And I think that's where we're going to end up. Interesting. So instead, we'll assume anything you see that's not labeled is fake. Yes. Then if it's real, that's when it gets some sort of certificate, something on the bottom, a QR code, could be blockchain. Who knows? That says, hey, this is real. Yeah, and it basically go and it'll show like the device it came from and all those kind of things and, you know, where it was taken. And that's the only way we're really going to be able to, you know, because the generation is just too easy right now.

Well, I mean, the conspiracy theories now are colliding with fake images. And so it's kind of, I kind of feel like we're in a post-proof, post-reality era where just you see it with your own eyes on a computer or phone, assume it's not real. If you see it in the real world, maybe. It is real. Yeah. All right. So let's see some demos here. We've got some demos back that. I'm excited. There's some really cool ones.

Okay. So this one, it's actually in your email. It's enabled in your Gmail if you're paying for Gemini. And so what they've done is at Google. And so I've just pulled up, you know, Ben Thompson's strategy repost from,

you know, a couple weeks ago. And so what you can do is you say, ask Gemini. And basically you get this little thing inside your Gmail. Yeah. And basically here you can go as like, you know,

can you summarize, this email in, you know, for bullet points. Perfect. And because, you know, these are quite long and, you know, you should read them. But I just using it as example here. He's prolific. Yeah. And here you go, right?

Amazing. Well done. And then you could do follow up questions with it. And so, and you can apply this everywhere. So this is not just to, you can help this help you draft responses. you can have this help you look through your emails as well. And basically you can say, oh,

can you help me find an email with Jason and me and about poker? But it was from February. And you don't have to understand all those crazy parameters anymore, like the search parameters. So this is, honestly, since I discovered this, I've been using it like multiple times a day.

Superhuman released an AI search product. That's very similar, a little bit better on the, margins. And it automatically summarizes your email as it comes in, which is kind of great. There's like a one-liner at the top of every superhuman email now, which is really nice. And then it will summarize the entire email thread. So you can do like a thread where you had 20 people and say, like, catch me up. And I think that's kind of really interesting because if you think about busy executives, you know, some people are rank and file. They get an email. They have to go through

the thread, they got to figure it out, you know, they don't have an assistant, they don't have a chief of staff, they don't have a researcher or whatever working for them. Other people are just like, I don't have time to read this. And you just give me a bullet point summary of it. And some human word to do that. So now everybody becomes bionic. Everybody has two or three researchers working for them. So I think this static team size theme I've been talking about with Alex and you on this podcast about static team size. Fancy way of saying your team size stays the same, but your companies growing. This is the weird phenomenon. I think companies will get smaller and more efficient. It's going to be pretty wild because you'll have somebody in the organization who adopts the technology

gets three times as much done and then somebody else doesn't adopt it and they get fired. The position gets, you know, cut, whatever it is. And boom, then that person's doing three times the amount of work. It really is going to be dramatic. If you're using these tools to get through your inbox, box, whether using superhuman or g-mails, you're going to be, you know, somewhere between two and 20 times faster. That's a game-changing right there. It really is. And, you know, I am, I am just blown away, like, just by, you know, how much you can do, how effective it is. And, you know, it's, like, like I said, as soon as I got into it, it's like, and its recommendations are powerful. Where does it go next, Sunny? Where does it go next? Because right,

now, this is exactly what I would expect. Nothing more, nothing less. I expect it to summarize. I expect it to, you know, help me find stuff. What's next? Where does it go? It watches how you deal with them, and then it just becomes you. Okay.

So let's say an email comes in. It's from somebody pitching me, I guess, for the podcast. It knows. It knows that you usually take that email and forward to, you know, deals at, you know, launch.co. Plus you, and you usually add three lines. Like, hey, you know, A, B, and C. It just starts doing, it kicks that part off for you.

Got it. So I already have an Athena-A-Assistant. Go to Athenawound.com to get your own and get a month free on your boy, J-Cal, or something like that. It's a pretty good deal. And what Athena does is I have my assistant and in my professional inbox,

she pre-sorts it. So she is constantly, Nikki is, labeling things, and then she haven't had her start forwarding it, but it's all labeled. So it's all in, you know, nice and clean. So when I get in there, and I haven't given her the instructions of like, if it's a startup pitch, give it to somebody else, but I could, and I could have a route

it, but this is going to be very interesting to see, you know, you're outsourced Athena assistant AI working together to do this, right? And I think that triage is really interesting. Obviously, some of these are giving you example responses or proposed responses. I find they're terrible. I was looking at LinkedIn's the other day, and they were kind of like one-dimensional. They didn't have my voice yet.

So I do think you're right. Understanding the person's voice is going to be the next big piece. All right, founders, do you want to sell to bigger customers? Of course you do. You want to get that ACV trending up, up, up, and to the right, average customer value. And you want to lower your turn, of course. And if you want to get to major buyers, you're going to need to clear those compliance checks, that means you have to have your SOC2 sorted out.

What's SOC2? It's just a standard and it ensures that companies keep customer data safe. If you aren't SOC2 compliant, you're not going to get the big deals. But Vanta makes it so easy to get and renew your SOC2 compliance. On average, Vanta customers are compliant in just two to four weeks. And without Vanta, it's going to take months. Vanta can also automate compliance for GDPR, HIPAA, and all these other acrony, that are boring and tedious and it's hard work. And that's why you get a partner like Vanta.

So it's easy, piecey, lemon squeezy. You want to get the big customers. You got to have these acronyms all dialed in, and they're going to save you hundreds of hours of work, and they're going to save you like 85% on your compliance cost. It's that simple. And Vanta is such a big supporter of this program that they're going to give you $1,000 off at vanta.com slash twist.

That's vanta.com slash twist for $1,000 off your sock, too. when you go on LinkedIn, here I am, this person, he, him, Eric is a startup coach, right?

Yeah. And here, there's a bunch of little AI wizards. I like that. We used a little star. What are the latest developments at Denver Union Station?

It knows he's at Denver Union Station. What does that have to do with anything? But I'm a premium user and it says, here it is. It knew that this person was taking a photo at Denver Union Station. Yeah. That's bizarre. He actually had a,

He added in his text. Oh, it does have it there. Excited to be working out of Denver. Okay, fine. And now, if I wanted to comment on this, what's very interesting is typically I have three choices here. It gives you, yeah, like a little same thing as well. I don't have that this time, huh? Yeah.

I know it depends on the message. I've seen that. Oh, it depends on the message, yeah. So here's an Airbnb one. And maybe if I go to this one, it will show me like my choice. but it's, oh, look at this. Another little one. Start a conversation easily by mentioning mission connections, introduce myself.

So now it's drafting a message. Hey, Jack. Whoa. I hope your week is going well. I came across your profile and noticed that we both connected with Mark Cantor. I am impressed by the work you did at, oh my God. I mean, that's wild. Wild times, folks.

It is literally starting me on third base. I just have to hit the send key. So, yeah, my in-mail is going to have even more people asking me to test their SaaS product or invest in their companies. I don't mind either. I'll be honest. I like getting a little in-mail. It is super useful. And then, you know, with that, like the Gmail demo, you can also be like, hey, can you find me that receipt from this?

Basically, your email, which a bunch of us use as a data source, it allows you to interact with it as a data source as well. Pretty interesting, yeah. What's going to happen, J-KL? What's going to happen? I mean, it's pretty clear to me that companies will stay the same size and then we'll get more done. And adding people, throwing humans at problems has always been rife with problems. It's been inefficient, but it does generally work. So, hey, sales is growing at 20%. Let's add more salespeople and hopefully you have sales go up to 30%, right? And we, you can, you keep throwing salespeople at the problem. Hey, our product is not keeping up with our, you know, competitor. Okay, let's throw some devs at it. You know, it's always like throw, oh, customer support times you're raising,

we'll throw some customer support people. Now, you know, I think managers are saying, wait a second, throwing people doesn't do it. We need to just throw AI at it. So the whole concept of throw people at a problem is going to be throw AI at a problem. Bingo. Yeah, we'll keep coming back to this theme. I think we're going to really change the workforce and find people that just use these tools natively.

And, you know, I keep going back to a time of my career in the late 90s, right, which feels like so long ago now, where, you know, and you'll remember this too, there was a huge part of the workforce that didn't use search. You remember that? Yes. Absolutely. There was. Yeah.

Yeah. God bless her souls. But a lot of those people are probably dead now, right? Because, you know, this was like 30 years ago. But, you know, I've said it many times on the show, paradigms don't die. People do. and I think there are people who are in that. And, you know, now you're going to have a verticalized AI that really understands LinkedIn,

that really understands Gmail, that really understands HubSpot, that understands Lead IQ or Salesforce. And, you know, these are the verticalized apps that are going to need a lot of fine-tuning. So when we saw that LinkedIn, you know, writing the custom message, you know, it may not want me to be so effusive, you know, or I may not want to be that effusive in the email, that I'm impressed by whatever. And so I just talk to it and I say, you know what, make the email shorter and more to the point.

Yeah. You know, I want to email all these CEOs and invite them to this conference, you know, and they should come to the conference because they're going to meet other great CEOs. The end. Don't sugar cut it.

And, you know, that I think understanding your style is going to be the interesting thing. And I notice chat GPT4O keeps adding to its memory of me. You know when it does that little thing where it says adding to memory? Have you seen this little? Oh yeah, yeah, yeah. It's doing it all the time now. When it realizes that you use the name or like a, that you have a dog or a cat or whatever it happens to be. It says adding to memory. And this is, I think, that's great. That's personalization. Right. So personalization is going to change this. Right now, as powerful as this is. Without personalization, it kind of hits a cap. I don't want to click on way to go. You know, I don't say way to go. You know, like that's just not in my lexicon. So, So, you know, if chat GPT or LinkedIn or superhuman, you know, picks that, not for me, but LFG

explanation point is. So if it knows, I like to say, let's F and go. Yeah. That's LFG explanation point with the rocket ship, which is my response to people, you know, it's going to learn that over time. So pretty exciting stuff. Okay. Good segue as usual, JCal.

Oh, actually, we didn't grade last time. Let's grade. Let's grade the Gmail integration. Yeah, it's a B. Solid B. Really? It doesn't do, it's not blowing me away. It's not making me go, oh my God, I got to tell somebody about it.

It would be, you know, just like when you do a net promoter score, it's like a seven or eight. It hasn't blown me away to the level that I would tell other people on my team about it because it's what I expect. Now, I'm going to give the LinkedIn score an A for specifically the introduction. Okay. I'm not giving an A to the responses it does, the canon responses, but I am. I'm giving an A to that introduction. Introduction thing because that was impressive. That is what, you know, somebody might take a LinkedIn course on how to use LinkedIn,

and that would be the template in the course, right, is, hey, show some interest, say something about the previous job, let them know, you know, have a closing sentence with a request in it. Like, that feels like it's training you on how to do it well. Yeah. Okay. So I give that an A. I give the other one a day. What do you give this to those?

I think, like, yeah, I'm with you. Like, I want to see a little bit more person. personalization. You know, like, look, Google Workspace is what we use. You know, I use Gmail personally. So like for me, it's like, I'd be like, hey, you have to add this now. So it's like to your $20, you got to add this $20. But so it's in the recommend category for me. But more personalization. And then sort of it to just, you know, take the task on. And it can do. It can write the intros and all that kind of stuff for you as well. Like, you know, I can have it write an intro to you and it'll do a pretty good job. So that sounds like you're going B plus or

something. Where are you at? I'm B plus with that. And I'm lined up with you. I thought that intro thing was pretty cool. That was surprising. It's the first time I saw it too. Yeah. And that's, I think, you know, if you look at crypto, I didn't, I never had a moment like

that in crypto. Yeah. I never had a moment where I opened up LinkedIn and it was like, hey, would you like to put this on the blockchain, your connection to this person on a blockchain? It's immutable. And then we're going to create an NFT of your relationship. You know, like, and then we're going to make a Dow out of your, you know, Google group. And it was, none of that's happened.

for me. There's no blockchain email. There's no NFT email. There's no Dow of my closest connections. This is what I really think is so important for technologists and startups to understand is, you know, think about the average person. Obviously, you could think about the most advanced people, the vanguard. But, you know, if something is too hard and there's friction, man, your first job as a founder is to get rid of that friction, right? When I see these products, these are zero friction products. Just click. Yeah. Like my total commitment is clicked. Yeah. That's pretty impressive. I want to wave about a new company I've recently invested in. It's one of my favorites. I'm so excited about being on this journey with the team.

You know, I love all the companies, all my kids the same, but this one is really special. So I want to tell you about micro one. They're going to help you scale your product with a team of developers in days, not weeks. When you're handling the search process, when you're trying to get great developers, the vetting and the onboarding and the paperwork, that's going to take weeks. for each one. It's painful. It's one of the hardest parts about being an entrepreneur. Well, imagine if you just tell AI what you need on an engineering basis and then AI delivers with the team at Micro 1 exactly what you need, you're going to be in great shape. And Micro 1 built this incredible AI engine. This is why I invested. They interview 20,000 engineers every month. They pick the top 1%. And then

they help onboard them into your company. It's simple and it's effective. They do all the legal. They do the compliance. They do the search. You don't have to do a mountain of resumes. You don't have to worry about somebody flaking out. You don't have to budget a huge recruiting bill. Micro One's got you covered. And you'll find your next engineer in 48 hours or less. That's the promise Micro One has for you. So here's the call to action, micro1.aI slash twist. M-I-C-R-O-1.a-I-slash twist to instantly ramp up your product team. And if you make a hire before October 1st, they're going to give you two weeks of free development per hire.

I just did this test here and I hadn't seen this one before because I've only been using it with my own email. So maybe it's going to, it might go just a little bit down, but this is a good feedback for the teams. You can see right here I said, can you help me write an email to Jason Calcutt as inviting him to Vegas to play poker? Be very convincing. But it says, sorry, I can't generate emails to celebrities. Well, I mean, but it's an email. Fair enough. But I'm writing an email to you. And it should know that you and I email all the time. We're, on an email thread. That is a fail.

And I'm not a celebrity, so. Yeah, I don't know. I don't know. I don't know. So I think, I mean, it's gotten a little weird. I will tell you, I was like, you know, when it gets weird is when, you know, it's not that somebody recognizes you. It's when like three people in a row recognize you.

So I literally was at Seoul House in Austin. I was just hanging out with a friend of my Will Barnes who's a VC. And I get up from the table to leave. And I had no, you know, you notice people kind of like clocking you. I noticed that because I'm like, from Brooklyn. I just always keep an eye out. Yeah, yeah. So I get up and these two guys beeline it for me. They say hi. The person at the next table gets up and says hi.

And then as I'm walking out, the third person gets up and I said, Matt, you know, like, because they had to ask for selfies. So I was like, sure, I'll take a selfie. She goes, no, no, I don't want to take a selfie. I want to know who you are. And I was like, oh, okay. It's like, listen, it's just a stupid podcast or whatever. And I, and I, and I, she's a most podcast. I tell her the name all. And she goes, oh, my boyfriend, listen. to that. Yeah. And I was like, okay, three for three. There you go. That's what you know it's getting a little bit weird. Take it all in, Jason. Take it all in. All right. I'm in. I'm in,

you know, I'm going to take it down to a B minus because I needed it. We're, we are organizing a poker game in Vegas and I need to read you a convincing email. And so I mean, you convinced me already. You said Vegas and poker and you. Yeah. Three of my favorite things. Yeah. We'll do it. We'll do it. Awesome. I can't wait. All right. All right. So let's keep going here. So Jason, this is one where I think you're really going to like it. And I don't know if you've started using this yet or you've been enabled. And if you haven't, it's really, it's really useful. So same thing here.

Chat GPT has this new function called Search GPT. And so you can say, tell me about the... Oh, are you in search GPT? Yeah, yeah, yeah, yeah, yeah. I didn't let me in. It's totally the waitlist is over. So this is perplexity from them, right? Yeah, in Austin during September.

Okay, here we go. Tell me about upcoming concerts. That's a pretty good query. Yeah. Upcoming concerts. Wow. And so, you know, basically it's breaking it down, right? You've got incubus coming.

Oh, you got Green Day. You can go and do that. Okay. No links, though. Yeah. So, you know, you've kind of... Yeah, but it's kind of giving you some... And so, like, you know, you could do the J-Cal thing as it can do you put these results in a table, okay?

And then do... Yes. Yeah, okay. I mean, this is impressive. If it's correct, I'm assuming it is. I don't know where it's getting it from, but it does look like I add citations and links. Yeah, great.

Let's see what it does. Oh, yeah. Add citations and links to purchase tickets. And, well, this is where if you click on one of those, it says purchase tickets, I don't know what link it's going to give you. Yeah. So this is where, you know, I'll do incubus. Okay, Incubis. There you go.

Okay. Seekis. Okay, perfect. Yeah, perfect. Yeah, amazing. And Green Day, let's see. Song, oh, song kick. Andrea, that's...

Except all the cookies. Okay, let's see. Yeah. That's just in Austin. Okay. Yeah. So, I mean, it knows, I mean, what it's doing, to be clear, is it's searching the web. It's got its own crawl.

We know it's crawling, right? That's how it built the language model. So it's got a crawler. The opening eye crawler has some number of these pages already indexed, correct? It's in its index. It's not pulling them live. I think they're doing a new crawl now for this, right? For this, okay.

And they may have partnerships. They may have built their own. I mean, they've got all kinds of resources, right? And, you know, we know there's different ways to do this because we've seen other companies to it. Wait, it says set as search engine. What is that? Oh, I think this, you know, it lets you set.

I mean, I, yeah. If you click that, it will set it up as your default search engine. That's crazy. Yeah, yeah. Oh, okay, well. Yeah. So here we go. It does that.

Now, explain to the audience what you think is happening here behind the scenes. You have a new crawl and then it's taking those web pages and then saying what to the inference engine. Yeah. So great question. So what most of these things are doing is like following a form of rag. And in this particular case, what happens is the user asks something. The LLM figures out what the user is asking. And then the LLM has access to what we call tools, right?

And those tools can be anything, right? And this is like, this is a really big, you know, sort of shift in paradigm and how people are building software. One of those tools would be a web crawler. Now, you can use for the web crawler, there's existing services like, you know, SERP API and all these other folks that exist. You know, Brave has one, too. Or you could build your own, right? And so, my guess what's happening is when you put the query in, it realizes that it needs to basically use a tool. In this case, the tool could be very specific. It could be very generic in that I got to go crawl the web, or it could be a very specific tool that's built looking for concerts. So if I was building it, I'd build lots of different tools so that I could make them better

versus one generic one, which may not be good for across all these things. And so. And RAG is retrieval, which is like getting that data together. And then the second step is augmented generation based upon what you retreat. So there's two steps going on when you do rank. Correct? Exactly. That's why it's kind of written. It's like the generation was sort of was part of the first part.

Like I asked it what to do. Then it does a retrieval to augment that generation. So you're not just getting results only from its training data. It knows how to go and use a tool. That tool will then go and do a search that will send results back to the LLM. And the LLM will reason over those results and create a much cleaner result here. like you see on perplexity or you see right here with chat. In this case, it got tickets, it got concerts.

It knows concerts and tickets somehow are related to Seekek and Songkick and other places. Boom. And then it retrieves that information. These guys could have done deals. We don't know, right? They're out there. A lot of folks are doing deals as well right now where they've just done deals with these folks too. Well, you know, this is, I think, the innovator's dilemma that we were talking about for Google,

which is, you know, Google's results. don't feel like generative AI. They feel like blue links or some documents reported back in tables, etc. Depending on what you're doing, flight shopping, tickets for stuff, et cetera. But this is a new feeling, right? This is a different feeling. And do you perceive that users want this type of experience where it's almost like they had an assistant do the searches and put it into a Google doc?

Doesn't that feel like what's happening here is that there's a research assistant between the search engine and the human? It's a really, really insightful question. Here's what's happening in terms of how we're interacting with the internet. Before, you know, really modern search engines, we had Yahoo, which was just an organized set of links. Correct, right? Exactly.

In a taxonomy. And you kind of had to go through and find stuff. Then you end up with search and different companies try different algorithms. Google comes up with a really good one called PageRank, which basically, you know, a quick summary is it looks. at the links between pages to determine the ones that may be the most relevant. Authority, yeah. Exactly, because they're referenced across other pages. Now the internet has content that's not always referenced, right?

Because, you know, page rank is not as, you know, as relevant because there's well-known sources and things like that. And what's really happened is the way we interact is still kind of in that progression where we were in terms of where, obviously no one goes to a directory anymore, but people still use search and they hunt and peck through a bunch of. listings. And what I think people want now is like, you know, how many times you just find is like, give me the just damn result of what I'm trying to look for. Yes. Solve my problem. It's solved my problem, right? It's interactive. And so, you know, even what we were just talking,

I had it add like a rating for each of these, right? I don't know what its basis was, but it added, you know, 4.5 out of 5, 4.3 and, you know, whatever it happens to be, right? Ridiculous, yeah. Yeah. Well, actually, here, do this. Say, get a column with a link to their YouTube page. and their Twitter hand. Let's just see if it can actually do that research task, right? Let's see if they can do it. You know, this is something Gemini did well for me, but, you know, it's trying to link to a YouTube channel. We broke it.

So here you go. Once again, J-Gal broke it in the first instance. So Sam Altman still work to be done here. But, you know, I do have great success with this with Chat. Cheap D4O, not the search engine. I do have success with this with Jeannie. Yeah, really good. Okay.

Do another one here. Give me the top 10 barbecue joints in Austin. All right. I'm using joints to see if it can navigate. It should be able to figure that out, yeah. I think so. Here we go, Franklin, La Barbecue. Those are definitely up there, Terry Black, number three.

Yeah, these are the ones I don't know, but that's okay. And yeah, I don't see the Salt Lake. There it is. There's the Salt Lake. Yep. Now put, include the Yelp rating. Now this is taking IP from Yelp. This is a pretty contentious issue.

Let's see what happens here. Include the Yelp rating for each. And it could get the Yelp rating from somebody who wrote a blog post about the Yelp ratings. But here we go. The address and the Yelp ratings coming in. So, yeah. And if you said sort by Yelp rating,

that's pretty crazy if it does it. Yep. That's, I mean, right about now, think Jeremy Stoppelman is writing a letter a cease interstance to. No, maybe they've done a deal. We don't know. It's possible.

Yeah, I mean, but this is what I want from Yelp. I mean, I know I'm like the, you know, I know everybody's like, Yelp, what are you talking about? But I'm old too. I love Yelp. I trust the data on there.

And I do all kinds of searches like this. But this is a, this experience would make me not go to Yelp. Yeah. So, you know, we're kind of, and I remember we did this and you're putting, can you add the distance from the proper reach? Yeah.

So that's where you like to stay, I guess. And there we go. Yep, there it is. 2.5 miles. Yeah. 1.8 miles. Yep. That is accurate and crazy.

So here we go, folks. I give this A minus. Okay. Well, we broke it on the YouTube thing. That's why I'm kind of giving it an A minus. Okay. Yeah. It feels, I mean, it's a B plus A minus.

I mean, the way I'm looking at this is we did two searches. You did one for concerts. I did one for barbecue. Both of them gave great results. So I'm just basing it on, did I get a great result or not. Kind of did. If you were to ask it, what's another thing you might ask a search engine? How much is a business class ticket from San Francisco to Dubai in November?

Let's see. Over Thanksgiving, there we go. Perfect. Weekend. Love it. Spelling mistake as well. I mean, this is, it's okay. Business class to Dubai, 7,500 is about right.

And it's getting this from other airlines. Yeah, well, it's getting it from a blog post. And then I guess it might have, if you click on, oh, yeah, look at that. Remember, they had that concept of. apps and we've reviewed it early on. Very early on. Very early on. It looks like instead of adding the apps, they're just built in.

Scroll down, there's three logos at the bottom, sources. And the sources have the Favicon and, oh, wow, you click sources and it gives you the actual pages it pulled it from. Yeah, it's just pretty great. It's Google's in trouble. If, you know, this gets much better. Yeah. What does this need to do in order to get people to switch from Google?

Well, I think kind of, you know, we've already touched on one part of it, right? It should be personalized. So like it should already know that I have travel coming up to Dubai and I have to go there based on my calendar and all those kind of things. And like the agentic side of it. And then really I just wanted to book, right? There's so much pain that I like, you know, we've been booking travel for so long. But like that last bit, we don't really think about it. But it ends up taking like 30 minutes, Jason, right? Like, you know, from the minute you decide on the flight to the time you get booked. all the way through. So imagine you could just say, book me that tick. That's what you mean by agentic, meaning an agent does it. So agenic is evocative of an agent. Is that a new word in the lexicon or is that a word that existed before AI agents? I think like people have been using it a long time, but it's definitely been attached to the AI world in terms of like how we expect these, you know, to work for us. I was fascinated when I learn a new word because agentic is not something that I've heard anybody ever say, and I guess there is a psychological term, person's ability to control their destiny and actions and a mindset that allows someone

to follow orders from an authority figure, agentic person, a person who is self-organized, proactive, self-reflective, and self-regulated. Interesting. Or it's been co-opted down by the AI industry. Uh, agentic. I like it. Um, yes, I'm going to give this, I'm going to go back. I'm going to give it a B plus. I like what you're saying, which is it has to do a little bit more. Now, when everybody freaks out, oh, my God, it's the end of Google. What I want to remind people is habit. And Google can improve and is improving, and they're investing a lot in this. So even if chat GPT is six months or 12 months or 18 months ahead of Google,

just like when I met this young woman bagging groceries while going to school, just bagging groceries at night, going to school during the day, lots of respect there. You know, and she's not even using chat GPT-4-0. They're not teaching her in school and not forcing her to use it. It's going to take a little while, right? Yep. In order to switch from one service to another, something has to be, in my mind, not 20 or 30% better. It has to be two or three times as better, which is 200 or 300% better.

So to dislodge Airbnb, to dislodge DoorDash, to dislodge Google search, to dislodge, you know, eBay. It takes a lot. It takes a lot to get people out of a habit. That's why Craigslist still exists. eBay still exists and prives. PayPal still exists and primes. And Amazon with the worst interface, the clugeist, most disgusting interface in the history of interfaces is still magical. And people are still loyal to it. So this has to get a lot better. Let's do one more. Let's do your headphone one. That's always a good one. I'm looking for the best headphones for listening to high fidelity music. Money is no object. Give me the top. five in a table with the price and common reviewers notes. Yeah, common pros and cons. Perfect.

I think it's good. And rating. Perfect. So this is like a multi-step. And oh my lord, this is pretty amazing. The top five high fidelity phone for all of all files, focal utopias, focal bathies. I have the utopias right here. And I have the battees.

The battee, the focal. Utopia is your 5,000. The bath is there are 700. You can go to Headphones.com. They're not a sponsor, but they, uh, they're friends in mine. These are $5,000? Holy cow. Yeah, I mean, they have them right here.

And when you listen to music on these, your brain melts. And then these are the $700 versions. They're both focal is, um, this, uh, well, it's a great story. The, the guy who started headphones.com had a different domain name, but he saw me interview Alex fromcom.com. And I talked about the value of a domain name and had to negotiate. domain names, he negotiated it, and then he, um, has a gift to me years later, sent me these headphones. Wow. And, uh, Focles a fan and they sent it as well. I usually don't accept

free stuff, but I, but the guy was like, listen, you, well, the guy said, listen, I, if I had, he said he had no sales, because nobody trusted his website when it was like headphones, best headphones.org. Yeah. When he finally was able to negotiate headphones.com, I think he started making tens of millions of dollars. So he's like, trust me, I can afford. it. Can you add sources? Perfect. Yeah. In this case, it didn't give sources, but yeah, headphones.com comes up. Wow. There you go. So when it's giving all these great, these are all the sources it used. Wow. So Jason, you look at, that's why I pulled this one up, just as one last one. So I'm going to say B plus, you know, and you know what I think is a big part

of this is returning stuff without ads, croft pop-ups, cookies, a big part of what you're seeing OpenAI and Chat GPD do here is remove and present information in a prettier way. Google results are organized, but sometimes cluttered. And then when you click on certain websites, they're pretty gross, right? Yeah, yeah. You go to a content website that's not a subscription service, and it's like, oh my God, how many ads, how many pop-ups, audio, all the stuff. So you can make a great business by just removing all that. Look at this.

This exact same thing in Google. Yes. Yes. And it dumped us into their search, which, you know, it's, it gives you on the left here. It's giving us the shopping experience by default. And, you know, this is not terrible, but to my point, this looks unorganized. This is maybe more powerful for doing searches, you know, if you. But according to you, Jason, the results that we had in the, yeah, this is not the right results. I think these are, because it's sponsored the first one, and, you know, Chatsy, if he doesn't do sponsored, that's. Go back to the Google one for a second. I think what you're seeing in the Google one is the fact that they're making money. So these, you know, focal is not trying to sell them. So they're not in this list. But if you scroll down and, you know, we...

Well, I got to go read the wire cutter or I got to go through this Reddit threat. You got to go to Reddit thread or to PC magazine, all these other locations. No way. Yeah. So, you know, this is... If you summarize all those other pages and those people get no track... Now we're back to the IP issue, which is I didn't need to click on any of those sources when it represented this information.

And so there needs to be some compensation to piece, and it didn't give us the sources here. So this is highly unfair what they're doing in my mind. But it said this is it. You had to ask for the sources. You had to ask for the. Yes. Yeah. So it didn't put them here by default.

So I just want to give a note, the open AI team. I know this is beta. but this is profoundly unfair and you're taking all the value away from the people who did these reviews so you got to compensate them somehow to ingest them which is what I think they're doing right

they've done a bunch of deals now I don't know if they did deals with headphones or PC magazine or Tom's hardware but they're gonna need to have done that so I'm gonna get this a B plus you can ask this guy I can ask them no I know they haven't

there's no way they've gotten down to you know the long tail or the fat tail, the medium tail, they've done, like, deals with news sources. But this is unfair when you think about fair use. Wouldn't you say? You know, like, here's what I don't really fully know. I think if they're not paying the sources, then, yeah, there's a problem because the people that are incented and, you know, it's crazy because I don't know if you saw this today,

but did you ever use a site on UnTech? You must have, right? Yes, it's shut down today. Yeah. A non-tech shut down, 27 years. It's because you can't. Well, because journalists cost, you know, 50 to 150K. Let's say the average journalist makes 75K.

Yeah. If they make 75K and they, you know, there's 50 weeks in the year, you've got to make triple their salary, double their salary to have a business. Let's say double their salary is 150. You got to make $3,000 a week off of their content. You know, $3,000 a week at a $50 CPM, you know, it's a lot of views. And so these websites are upside down now. They cannot afford to pay the journalists to do the work. And this is going to break it even further when it just gives you, here's the top five, go buy them.

And then they're going to, you know what's going to happen. Obviously, chat GPT is going to charge people to be the link. So independent of where it got the information, they could just say, anytime focal utopia comes up in the price of it, we'll link to you. Or they could just say, would you like clicks that are related to headphones.com? Anything that's headphones related, we'll just send the traffic to you for a dollar or click. For $10, it would be an auction. And they just are embedding clicks here. So this could be PC magazine wirecutters and wire cutter and OpenAI are suing.

This could be based on their content, and it probably is. And they'll be able to prove this pretty easily, I think. And then if Open AI puts a revenue model on this, like links out in an auction, you would have stolen somebody's content and then slapped a link on it. Yeah. So let that sink in. This isn't monetized yet outside of the fact that you're paying $20 a month to remove all the ads and have a cleaner experience.

But in a way, they are monetizing it with that subscription. So this is going to be, when this search engine comes out, I think you're going to see 10 times as many lawsuits against Open AI. So there's a reason why they didn't open this to everybody. This is Lawsuit Central. Okay. There you go, folks. I give it a B plus.

It's compelling. I'm not stopping Google for it, but it might take a couple of my searches away from Google. Yeah. I personally think perplexity is still better. And perplexity is upset. That founder is upset about chat GPT, rug pulling him. I noticed. Well, I don't, yeah, but look, that's just the era we're in right now.

But it will maybe come down to user interface or something like that as well, right? Because maybe you're right, it'll come down to who can cut the best deal with all the data sources, right? I mean, that, that, when you frame it that way, it's easy to, you know, I'm more on the technology side. You kind of lose touch. You're like, yeah, just go crawl the stuff and put it in there. But like, you know, when you see an untech shut down, you're like, man, that, that's real people.

That content is gone now. Well, there's not going to be anything added to it. So who does the next review of, like, yeah, the historical stuff will be there. But when the next X CPU or whatever comes out, we're not going to get there anymore. Nobody to review it. It's over. Yeah. And then they're like synthetic data.

I was like, what am I going to do with synthetic data? Like, synthetic data doesn't help me. You're just going to give me a bunch of made-up BS. That doesn't help anybody. All right, let's do another review. Okay. This is great. We got through the chat, ChbT search.

We got through LinkedIn, Google. This stuff is really starting to hit the mainstream. It is everywhere right now, right? And so, okay. All right, this is one I actually have on my phone. You know, my first company was in apps. And so this one hits close to home for me. And so this is with Claude, and it's kind of building on something we showed before,

but they continually do really awesome stuff. And so I'm going to basically say, you know, can you make me a bill splitting app and use from my phone? And so, you know, this is a very simplistic example. And what you're going to see now is using that same artifact feature that we talked about before. Yeah.

I now, it made me a real-time app. So let's just say it's $100. Give a weird number, like $126.72. And number of people, let's say it was five and split bill. And there we go. Let's zoom out here. So you're making a, can you make me a bill splitting web app? Yep.

Yep. And it says each person should pay, you know, so there is five of us and each person should pay 23, you know, thing. Now what's awesome is you can say, actually, can you, add a tip section, right? So you're literally making a web app that somebody would have written over a weekend 20 years ago. And look, and now anytime you have a problem,

Jason, so what I'm going to challenge you here is, and so now it's the same thing. So say, you know, our bill was 1,2765. Now put a 15% tip in. We'll give you a dialogue to pick from the most common ones. So where is it getting this knowledge and information from? Is there an open source page? Did somebody make a tutorial?

How do you think it had this information in its model? Okay. So I think there's a few things going on. I think all of these folks are crawling open source code repositories. And then two, there are companies like Sourcegraph who actually make code available as well, right, to these folks. And so I think they're just kind of building up their strength. There's so much open source code out there at this point, right? That, you know, they're putting together all the bits and pieces of making these, you know, simple applications and functions and features.

And so I'm... So if it hadn't, you know, indexed and, you know, ingested into the model, all that open source data, it found on somebody's GitHub that they had built this before and it essentially is spitting it back out to us. I'm not convinced. I think the models are so good now. It didn't have to have seen a web tip calculator. Oh. So it knew what a calculator was maybe? Exactly. It knows what a concept of a bill splitting app, Bill splitting is. It knows what adding tip is. It knows I do the math. And so I think these are getting so good now that it wouldn't have to have seen this before. Got it. So, because, you know, if you're writing code for something, you're like, how do I make a, um, you know, an image gallery where people can upload five images and then put it in a carousel. It's like, there's a code base and open source repository for uploading multiple images from your camera. There's a repository and open source library, whatever, an SDK to do image correction or

to present the gallery. So, you know, that's always been a hack for developers just to find those things and adopt them. But here you're saying it actually understands what this is. A calculator is this and this calculator is supposed to do why. And it just does it. It's pretty impressive. when you think about it.

And so what I suggest for you, Jason, just a little task for you and maybe other team members, like, anytime you're doing something now, you're like, oh, I wish had an app to just do that. Just try it. So you should definitely sign up for Claude and be in their page. I am paying for Claude AI, but they have an, is their app good? Their app is excellent because, like I said, you could do these things. And so next time you have kind of like a weird, make an app, right? I love it.

Great idea. I mean, I got to give this, what? What do you give it? What's your ranking here? This is A for me, you know? Yeah, I was about to say B plus or A minus. Yeah. Like, you know, I'm going to give you an example because I know you're doing this out in the, in the new life that you're having had.

But like, you could make a custom app if you're barbecuing something. Or if you're like doing a, you know, let's just say, you know, you want to task manage yourself and you're going to cut the lawn and, you know, feed the animals and, you know, whatever. The ranch work. But yeah. Make me a ranch management app. Yeah. And just for that moment, just for today. Yeah, just for now.

And say, like, you know, each one put a counter and say, I'm going to cut the lawn for 30 minutes. I'm going to feed the animals for 25 minutes. And just like, so, right? Like, I mean, I'm really blown away by this functionality. It's pretty crazy. Does squad have the best app development AI, you think? Well, what I think they've really brought together incredibly well is a really powerful model. So you got to give them kudos for that.

But they've brought that like, like, you know, it executes it for you. And so not only is it like right, because, you know, all of these have been writing code and we've done plenty of examples like that. I think we did a calculator one quite early on. But like for it to just run it on your phone as a little web app, you know, I think that that's well done, really well done. I mean, and they also doing charting really well. I noticed that they added charts to a lot of these language models now.

They have, yeah. They've done that really, really well. I'm going to give it a B plus out of the game. Oh, my God, Jason. I mean, it might be higher, but I haven't seen it do like a more complicated app. So I'm leaving room for improvement. I said B plus A minus. Maybe I'll give it an A minus.

I'll give it an A minus. I'll give it A plus, Renee. I'm at A plus. Like, I think this is really cool. I think, you know, I started tweeting a bunch of these examples out. I called this the B-Y-O era. Like, bring your own. You know, build your own.

So I build your own. And instead of hunting and pecky-o, through, like, I hate getting those really, like a tip calculators in that because they always end up having ads and all that kind of stuff. And now you could do it. And you could, you could even give it like a weird thing, like, oh, you know, split the bill, but make it like two-thirds Jason and one-third me because, you know, that's what we decided on. Yeah, incredible. Awesome. All right, let's do one more. You got one more interesting one for the fans? Let's do one more for the fans. Yeah, I do. I do. I actually got a really short grok one

that I want to show you as well. Oh, great. Awesome. I think you get a kick out of. Yeah. Okay, let's do Playground. This one, this one's really,

really cool. Man, and all this inference is getting really fast, huh? All this inference is getting much faster, yes. Yeah. So this idea of like waiting and waiting. I mean, for images,

you're still wait and wait, but for text, it's pretty quick now. Yeah. All right. So basically, so He was a previous multiple successful founder. You'd built like one of the Google Analytics equivalent,

mix panel, right is with the company he built before this and so he's been working on this it's really cool let me pull it up here and basically it's image generation but it's tied to um you know he's kind of like formatted it in a way because like a lot of these places is just too loose and so it's like oh are you trying to do a mock up a t-shirt what are you trying to do here right and so you can go in here and basically what's awesome is there's a bunch of pre-built stuff which is good to start with, right? But you can basically take something and say, I can take this one right here. And I can say, you know what changes. I say, you know, make this in a, you know, hue of blue, right? And basically,

you know, it'll go off and it'll do it pretty quickly. And then it kind of takes you through the workflow to like create the thing that you want, which is, I think is really, really awesome. Yeah. So this is for making graphics. Graphics, but, you know, I like, I like the way that Dave, And so you can say you can apply this style. Oh, wow. Yeah, exactly. And then, you know, you're just off to the races, right? And you say, I have this.

I export it. And, you know, you're good to go. And so they've done a really good job. I'm really, really, you know, kind of. Playground.com. Yeah. Yeah. Playground.com.

Great domain. And then. Yeah. Yeah. And this is to create logos, images. Yes. Just design. Yeah.

And you can start from a design. And, you know, they've just. It is. But with generative AI at the core and, you know, and a bunch of like things to kind of inspire you to get started, right? Which I think is really cool. You can make T-shirts, anything. Yeah. Pretty great. Yeah. Yeah. And it's got a bunch of templates and you can just start with one of their templates and then work from there. Wow. Yeah. Pretty impressive, I have to say. And you can do memes, all that kind of good stuff. Wow. Yeah. Amazing. I love the memes. The memes have a great potential. What's interesting about the memes is they're taking famous memes. Yes, and then allowing you to, yeah. Mess with them, yeah. Yeah. So they have that famous meme of like the superhero with the two buttons,

but it doesn't look like the original. It's pretty clear that this one has been set up through AI. And so, yeah, wow, pretty impressive stuff. I'm going to give that a B plus as well. Yeah, I'm kind of the same place as you. I think it's early and, you know, just the way. workflows are there. But good start and I would use it and, you know, really have been. I mean, sorry to graphic designers, but, you know, I think when you started outsourcing graphic design around the world and it just

plummeted in price, you know, it was a bummer for designers, but people started to design more things. There were more logos designed in the world. There were more T-shirts designed, you know, in this long-tail of stuff. So you had this sort of abundance. I think what's going to happen here is like all of this like commodified design is going to make elite design even more value. So if people start making a ton of schlocky AI design or, you know, randomness, then somebody who truly has an aesthetic, I think, is going to stand out. But it's going to be hard to stand out. So, like, what is elite design is going to become an even smaller percentage of design? Because it's going to be such a sea of garbage. It's kind of like on Spotify, like there's so much bad music that

people are going back to old music looking for musicians who actually have talent. So you can here's something that doesn't sound like it's all digitally produced from samples. Yeah. Amazing. Crazy world. Okay. You got one last one you wanted to do, you said? Yeah, one last one.

Yeah, let's do it. Okay. So, and this is just kind of like proof of concept of where, you know, I think this is going to go really, really quickly and we'll have fun with this one. I think I'm going to Austin this weekend, can you come up with a quick itinerary that touches on some good food places and some key sites I should visit? And so you see, it kind of does that. You say.

Amazing. Yeah. Actually, can you put that in a table? My God, look how fast this is. This is using Grock, obviously, yeah.

Can you add a, yeah, can you add a column for duration? Amazing. You know what? I changed my mind. Can you change this to Houston? And so,

you know, why I wanted to kind of show this, Jason, is like, when we, when we're thinking and look, like you're doing this in,

you know, kind of launch, right? and the accelerator. This is probably where we're heading, right? Where the speed of iteration and the ability to change it and interact with it is this is even faster than search. Yeah, and this is like Minority Report when he was, or in Blade Runner, where they're like,

zoom in, zoom out, do this, do that. And like, this is 750 tokens a second. Yeah, yeah. Speed. And that's not even our fastest. We can go faster than that. By the way, this is such a game changer because it changes the whole. nature of inquiring about information when it moves this quickly.

Yeah. And, you know, if you think about this with the glasses that are coming, you know, I know if you have the Ray bands from Meta, I really recommend you get them. They're super useful. They kind of have that bone induction audio and they can look at whatever you're looking at, so you can ask questions. I really think the user interfaces are going to really go more towards voice and vision in these type of ways.

And we'll sort of look back and we'll be like those old folks that weren't using surge. We were in some console typing things out, but imagine being able to iterate like this. So that was one of the things that the team has worked on. We've added voice and a bunch of other things. So that was pretty exciting. That's an A plus for me, man. Seeing that level of speed and stuff. Yeah, no, you can give it a grade. I'm giving it an A plus because this moves so quickly, I could see this, you know, as it applies to making images, video, you start thinking about all the demos we saw today, how much more powerful they would be. Yep.

And if I, you know, my wife and I have been using the chat GPT4 interface. We drove to Houston to go do a couple tasks. And while we were doing it, we were asking questions about different health things. And it was reading it back to us. But it still was a little CB radio, right? It wasn't snappy enough. Kind of like GPS used to be or photography used to be. When it wasn't snappy, you just didn't use it.

You only used it if you had to. But when GPS became crisp and fast and, like, Like, it was accurate. Like, you can't not use it. You have to use it all the time. So I think this is like, of all the demos, that last one is the one that's got me thinking, okay, this is going to get a little bit crazy.

All right, this has been another amazing episode of this week in startups. You can follow Sunny. He is at X.com slash Sunddeep. I am X.com slash Jason. We're on LinkedIn. You can send your AI to introduce us. And everybody go to console.rock.com. and if you're a developer, go start playing in the playground.

We'll see you all next time. Bye-bye.