1 00:00:00,020 --> 00:00:04,320 Josh: Just in the last week, Google unveiled some pretty unbelievable new AI software. 2 00:00:04,640 --> 00:00:07,500 Josh: One of my favorites being this new thing called Lyria 3. 3 00:00:07,780 --> 00:00:14,560 Josh: So on screen here, I have a music generator that I actually am going to prompt to make a song for us. 4 00:00:14,680 --> 00:00:18,880 Josh: So what I've loaded up here is make me a 2010s Kanye style stadium anthem about 5 00:00:18,880 --> 00:00:20,860 Josh: the Limitless podcast hosted by us. 6 00:00:21,020 --> 00:00:24,920 Josh: And what I kind of want to do is, is have it do this in real time while we discuss 7 00:00:24,920 --> 00:00:27,240 Josh: the product while we discuss the software. 8 00:00:27,440 --> 00:00:30,660 Josh: So Lyrae 3, what you noticed on screen there, which was super cool, 9 00:00:30,780 --> 00:00:34,180 Josh: is you choose the genre you want, you feed a prompt to the model, 10 00:00:34,280 --> 00:00:38,880 Josh: and after you fed the prompt to the model, it will go and generate you lyrics, the full production, 11 00:00:39,120 --> 00:00:42,760 Josh: and a full 30-second song about whatever it is you want. 12 00:00:42,860 --> 00:00:48,100 Ejaaz: Yeah, I saw a hilarious example of a lady asking her husband to do the dishes, 13 00:00:48,380 --> 00:00:53,420 Ejaaz: but instead she wrote it as a message and created a song and sent it to Dan, 14 00:00:53,420 --> 00:00:57,760 Ejaaz: her husband and it played just like hilarious music all jokes aside i think 15 00:00:57,760 --> 00:01:01,200 Ejaaz: this is more than just a gimmick or a novel you're about to see the quality 16 00:01:01,200 --> 00:01:04,960 Ejaaz: of this thing through the jingle it's just generated but this could replace 17 00:01:04,960 --> 00:01:11,020 Ejaaz: a music series entirely at least for for 30 second excerpts for tiktok or whatever that might be are 18 00:01:11,020 --> 00:01:12,900 Josh: You ready to listen to it we have our output ready. 19 00:01:12,900 --> 00:01:13,720 Ejaaz: Let's go 20 00:01:44,250 --> 00:01:47,450 Josh: Whoa, even an outro too. Wow, okay, that's pretty good. 21 00:01:47,750 --> 00:01:51,830 Ejaaz: Just to be clear, that was generated in about under two minutes. 22 00:01:52,350 --> 00:01:56,730 Josh: I'd say less than one, fast. And the production is good. Like the lyrics sound good. 23 00:01:57,090 --> 00:02:00,910 Josh: It's like a proper rap song. And I think that's one of the novel breakthroughs 24 00:02:00,910 --> 00:02:04,850 Josh: with this technology is that a lot of these tools that Google in particular 25 00:02:04,850 --> 00:02:09,630 Josh: has been releasing recently are really enabling creators to have this powerhouse 26 00:02:09,630 --> 00:02:12,170 Josh: suite of tools that let you do things like create music. 27 00:02:12,550 --> 00:02:16,410 Josh: If we want to have an intro song, we can actually use that or you can kind of 28 00:02:16,410 --> 00:02:18,590 Josh: workshop it and change a few of the lyrics and change the style. 29 00:02:18,750 --> 00:02:22,870 Josh: But I think what's really fun is that you can actually choose the types of songs 30 00:02:22,870 --> 00:02:25,570 Josh: that you want to create and you can embed them into whatever content you're making. 31 00:02:25,710 --> 00:02:30,210 Josh: So there's a second feature to Lyria 3, which is around generating photos or 32 00:02:30,210 --> 00:02:32,090 Josh: generating songs through photos. 33 00:02:32,270 --> 00:02:36,070 Josh: So what I have here is a photo of cold and snowy New York. We just got a blizzard 34 00:02:36,070 --> 00:02:42,330 Josh: last night, and I'm going to say, generate me a song that reflects the mood 35 00:02:42,330 --> 00:02:44,950 Josh: in this very snowy picture. 36 00:02:45,490 --> 00:02:49,310 Josh: And yeah, I mean, we were under a blizzard warning lockdown last night. 37 00:02:49,490 --> 00:02:51,610 Josh: Things are really bad here in New York, and maybe it will. 38 00:02:52,350 --> 00:02:55,690 Josh: Yeah, well, that's right. EJ's went to Florida, but maybe what we can do here 39 00:02:55,690 --> 00:02:59,210 Josh: is kind of synthesize that feeling through music. So we'll see what it does. 40 00:02:59,430 --> 00:03:04,950 Ejaaz: This is something that Google is really well known for, which is multimodality when it comes to AI. 41 00:03:05,150 --> 00:03:07,850 Ejaaz: It's not just a great LLM or chatbot. 42 00:03:08,150 --> 00:03:13,490 Ejaaz: It understands visuals such as videos and images. It understands sounds. 43 00:03:13,930 --> 00:03:19,170 Ejaaz: Remember, this isn't the only, I guess, sound model that they've created before. 44 00:03:19,310 --> 00:03:22,670 Ejaaz: They've had live translation with their AI that translates any kind of, 45 00:03:22,750 --> 00:03:25,510 Ejaaz: I think, 40 plus languages into whatever language that you want. 46 00:03:25,870 --> 00:03:29,730 Ejaaz: They've had experience in this. And the fact that their models now have this 47 00:03:29,730 --> 00:03:34,650 Ejaaz: intuition and this judgment, this capability to be able to see a snowy day and 48 00:03:34,650 --> 00:03:38,670 Ejaaz: say, hmm, maybe I'll generate a vibey song for that is pretty impressive. 49 00:03:38,910 --> 00:03:41,230 Ejaaz: It looks like we have it already. Wow, that was under 30 seconds. 50 00:03:41,850 --> 00:03:42,810 Josh: Yeah, you ready to give it a listen? 51 00:03:42,970 --> 00:03:43,590 Ejaaz: Yeah, play it. 52 00:04:12,720 --> 00:04:13,500 Ejaaz: Are your spirits lifted? 53 00:04:13,500 --> 00:04:17,300 Josh: Sonically, it sounds nice. I feel like I'm listening to a like kind of knockoff 54 00:04:17,300 --> 00:04:21,900 Josh: Demi Lovato, like main character energy type thing. And I like it. 55 00:04:22,040 --> 00:04:24,720 Josh: And to your point, I think this is something unique to Google in the sense that 56 00:04:24,720 --> 00:04:28,520 Josh: because they have these world models, they have that understanding of what they're 57 00:04:28,520 --> 00:04:33,040 Josh: seeing in a way that I think is more intuitive than a standard language, 58 00:04:33,240 --> 00:04:34,480 Josh: text-based language model would do. 59 00:04:34,620 --> 00:04:37,660 Josh: And another interesting thing that I learned while kind of 60 00:04:37,660 --> 00:04:41,440 Josh: playing around with this is they have this tool called synth id and 61 00:04:41,440 --> 00:04:44,380 Josh: what it is is it's a digital watermarking service that's baked 62 00:04:44,380 --> 00:04:47,940 Josh: into these files that we're listening to so that 63 00:04:47,940 --> 00:04:51,800 Josh: if you have an algorithm that can detect it it can actually know with high levels 64 00:04:51,800 --> 00:04:55,080 Josh: of certainty that ai was used to generate the music which is good for copyright 65 00:04:55,080 --> 00:05:01,360 Josh: and for kind of trademark issues and the way it works is it injects this unhearable 66 00:05:01,360 --> 00:05:05,660 Josh: waveform within the music that we just heard that we can't detect as human beings 67 00:05:05,660 --> 00:05:07,280 Josh: but should you feed it to an AI, 68 00:05:07,480 --> 00:05:11,040 Josh: it'll be very easy for it to decrypt that and let you know that this was AI generated. 69 00:05:11,220 --> 00:05:14,820 Josh: So pretty cool stuff. I really enjoyed Lyria 3. And I think a lot of people 70 00:05:14,820 --> 00:05:18,180 Josh: would have fun playing with this. It's free to use, available in Gemini. 71 00:05:18,360 --> 00:05:20,480 Josh: And yeah, super fun for people who, 72 00:05:21,140 --> 00:05:24,780 Josh: wanted to be producers or just want to send their friends a funny birthday song. 73 00:05:24,780 --> 00:05:27,920 Ejaaz: Well listen my take on this is that it's actually 74 00:05:27,920 --> 00:05:31,040 Ejaaz: i'm going to go back on what i said earlier it's probably not going to take over music labels 75 00:05:31,040 --> 00:05:35,380 Ejaaz: anytime soon but for like v1 or v3 you can imagine what this probably looks 76 00:05:35,380 --> 00:05:39,460 Ejaaz: like in about six months time and if it's taken if it improves at any of the 77 00:05:39,460 --> 00:05:42,500 Ejaaz: rate that any of the llms have like this is going to be a pretty insane thing 78 00:05:42,500 --> 00:05:47,420 Ejaaz: to do um on the synth id thing josh the reason why i was wondering why when 79 00:05:47,420 --> 00:05:49,360 Ejaaz: you were explaining it why it sounds familiar to me 80 00:05:50,180 --> 00:05:55,580 Ejaaz: they do this with AI images as well where they kind of like mix up a few certain 81 00:05:55,580 --> 00:06:00,100 Ejaaz: pixels in an image that is generated to tell you that this is AI generated. 82 00:06:00,360 --> 00:06:04,320 Ejaaz: They do this with text as well where they subtly kind of like change certain 83 00:06:04,320 --> 00:06:05,960 Ejaaz: word choices to do the same thing. 84 00:06:06,280 --> 00:06:12,420 Ejaaz: My question to you is or rather to both of us is can you create models now that just don't do that? 85 00:06:12,620 --> 00:06:15,720 Ejaaz: Like this is a baked in watermark that Google kind of put up, 86 00:06:15,780 --> 00:06:17,720 Ejaaz: right? But presumably you could create a model that just 87 00:06:18,170 --> 00:06:20,430 Ejaaz: I guess, illegally copyrights a bunch of this stuff? 88 00:06:20,730 --> 00:06:24,850 Josh: Well, my assumption is that if you can detect it, then you can reverse what 89 00:06:24,850 --> 00:06:27,130 Josh: it's detecting by using the same tool. 90 00:06:27,290 --> 00:06:31,250 Josh: So like if an AI can be aware of the trademark, it can probably reverse engineer 91 00:06:31,250 --> 00:06:35,170 Josh: those few pixels or that waveform that exists and remove it entirely. 92 00:06:35,330 --> 00:06:41,310 Josh: So I suspect it probably goes both ways, but it is nice to know and nice to have. 93 00:06:41,470 --> 00:06:44,610 Josh: And I think it's funny because we are following this trend of, 94 00:06:44,970 --> 00:06:48,730 Josh: again, reverse CAPTCHAs, where we're building things for AIs to recognize and 95 00:06:48,730 --> 00:06:51,770 Josh: not humans to recognize and this is another step in that direction. 96 00:06:51,770 --> 00:06:54,830 Ejaaz: Well why i like it as well is um if 97 00:06:54,830 --> 00:06:57,670 Ejaaz: you are a music creator that's listening to this or even like an artist 98 00:06:57,670 --> 00:07:00,690 Ejaaz: that creates images you might be thinking well they might be stealing my work 99 00:07:00,690 --> 00:07:05,150 Ejaaz: and i get nothing for it with watermarks like this it'll be recognizable to 100 00:07:05,150 --> 00:07:08,610 Ejaaz: your attributes or brands and maybe you end up getting paid out for this something 101 00:07:08,610 --> 00:07:11,890 Ejaaz: like in some kind of future system that doesn't really exist right now and it 102 00:07:11,890 --> 00:07:15,910 Ejaaz: definitely beats any kind of archaic royalty system that existed before so So I don't know, 103 00:07:15,970 --> 00:07:18,890 Ejaaz: I just like the technology. I think it's more than just a watermark. It's pretty cool. 104 00:07:19,390 --> 00:07:23,910 Josh: Yes, and that is not the only Google Labs cool new thing because there was another 105 00:07:23,910 --> 00:07:27,630 Josh: really fun creation that they had that they published to the Google Labs team 106 00:07:27,630 --> 00:07:30,370 Josh: called Pomelli. Do you want to walk us through what this post says? 107 00:07:30,650 --> 00:07:33,810 Ejaaz: Yeah, so I mean, I hate to say it, but it's true. 108 00:07:34,250 --> 00:07:39,390 Ejaaz: From a single image of your product, you can now generate several thousand dollar 109 00:07:39,390 --> 00:07:44,110 Ejaaz: worth photo shoot, which includes people that feature your product, 110 00:07:44,110 --> 00:07:46,510 Ejaaz: different types of lightings and textures, 111 00:07:46,770 --> 00:07:49,350 Ejaaz: backgrounds, all in a matter of seconds. 112 00:07:49,490 --> 00:07:51,970 Ejaaz: And the reason why that's a crazy thing to say is 113 00:07:52,700 --> 00:07:56,260 Ejaaz: there's a lot of photographers that spend a lot of their time and expertise 114 00:07:56,260 --> 00:07:59,620 Ejaaz: creating these photo shoots that now are pretty much out of a job. 115 00:07:59,740 --> 00:08:03,260 Ejaaz: And I'd also take that a further step and say that there are a lot of fashion 116 00:08:03,260 --> 00:08:05,860 Ejaaz: models that also may not be able to benefit from this as well. 117 00:08:06,200 --> 00:08:11,500 Ejaaz: But I have to say the feature is so cool. The fact that you can go from a single 118 00:08:11,500 --> 00:08:16,300 Ejaaz: person sitting at home that's ordered or generated their own product to a full-on 119 00:08:16,300 --> 00:08:21,760 Ejaaz: advertised kind of website that features these slick different images. It's just so cool to me. 120 00:08:22,200 --> 00:08:25,300 Josh: Yeah. And we have some examples of this. I mean, one of them I used for the 121 00:08:25,300 --> 00:08:27,800 Josh: bankless website that we use and posted about it. 122 00:08:27,860 --> 00:08:30,800 Josh: And it got like a million views because it's really awesome. 123 00:08:30,880 --> 00:08:34,560 Josh: What it does is, is you feed it a URL. So if you have a personal brand, 124 00:08:34,660 --> 00:08:37,100 Josh: if you have a website, if you have any sort of content on you, 125 00:08:37,160 --> 00:08:41,140 Josh: you feed it the URL and then it populates this business DNA sheet where it shows 126 00:08:41,140 --> 00:08:45,620 Josh: you your logo, the fonts, the colors, your tagline, all the values. 127 00:08:45,620 --> 00:08:49,680 Josh: And it It extracts this kind of identity that you can then use to create this 128 00:08:49,680 --> 00:08:50,640 Josh: marketing material with. 129 00:08:50,800 --> 00:08:54,640 Josh: And another fun example that we generated with the Bankless stuff was a hat. 130 00:08:54,840 --> 00:08:58,360 Josh: We had a hat in the merch store and it's just this very basic bland hat. 131 00:08:58,620 --> 00:09:02,360 Josh: But what I did is I fed it through Pomelli and I asked it to do a product shoot. 132 00:09:02,680 --> 00:09:05,260 Josh: And what it generated on the other side was really impressive. 133 00:09:05,460 --> 00:09:09,500 Josh: It shows, yeah, we could see here it has a model wearing the hat that looks 134 00:09:09,500 --> 00:09:11,720 Josh: like it was shot professionally. The lighting is very cool. 135 00:09:12,200 --> 00:09:16,980 Josh: The everything looks real. I can't really like if if you were to show me this 136 00:09:16,980 --> 00:09:20,360 Josh: in a magazine or just scrolling on x I would think that's zooming. 137 00:09:20,360 --> 00:09:21,540 Ejaaz: In here josh this 138 00:09:21,540 --> 00:09:25,040 Josh: Like there's really incredibly real indications and 139 00:09:25,040 --> 00:09:27,980 Josh: this took 30 seconds to generate all by feeding him 140 00:09:27,980 --> 00:09:30,760 Josh: the hat and then it created this really epic product shot that was 141 00:09:30,760 --> 00:09:33,480 Josh: just this fun kind of hero image showcasing the hat 142 00:09:33,480 --> 00:09:36,440 Josh: showcasing the logo in the middle of it and I think to 143 00:09:36,440 --> 00:09:40,900 Josh: your point I mean this is something that a lot of creators would see as a huge 144 00:09:40,900 --> 00:09:44,240 Josh: barrier where it it costs a lot of money to get these professional looking shots 145 00:09:44,240 --> 00:09:48,900 Josh: to hire models to have the lighting and the photo shoots and the reality is 146 00:09:48,900 --> 00:09:53,680 Josh: is that it's really not that difficult to do if you use this tool so what i figured we could do, 147 00:09:54,300 --> 00:09:58,520 Josh: is we can try to use Pomelli here live and do a demo for Limitless. 148 00:09:58,560 --> 00:09:59,860 Josh: So I actually have it loaded up right here. 149 00:10:00,140 --> 00:10:03,980 Josh: We can run through exactly how it works and create our own business DNA. 150 00:10:04,180 --> 00:10:07,860 Josh: So when you click let's go, it asks you to use the website. We kind of have 151 00:10:07,860 --> 00:10:11,100 Josh: a website, limitless.bankless.com if anyone wants to go check it out. 152 00:10:11,260 --> 00:10:14,020 Josh: And I'll feed it in there. And what you'll see is it takes a few minutes. 153 00:10:14,140 --> 00:10:17,520 Josh: It'll analyze everything. It clicks through and it starts to understand what 154 00:10:17,520 --> 00:10:20,620 Josh: colors you use, what images you use, what type of copy you use. 155 00:10:20,840 --> 00:10:23,880 Josh: And while that's thinking, maybe we could kind of describe a little bit more 156 00:10:23,880 --> 00:10:26,100 Josh: about how it works and what it's used for. 157 00:10:26,720 --> 00:10:32,540 Ejaaz: Yeah. So the rough kind of steps would be you can enter your website URL and 158 00:10:32,540 --> 00:10:36,120 Ejaaz: Pomelli basically just scans your entire business and extracts, 159 00:10:36,180 --> 00:10:40,280 Ejaaz: as you mentioned earlier, something that would be referred to as, I guess, a business DNA. 160 00:10:40,480 --> 00:10:44,220 Ejaaz: So it kind of judges your vibe. It gets your vibe. And it's a theme of all these 161 00:10:44,220 --> 00:10:47,460 Ejaaz: different products. Google's models are very good at kind of gauging sentiment. 162 00:10:47,920 --> 00:10:51,240 Ejaaz: So it'll understand kind of like the tone of your voice, the color palette, 163 00:10:51,500 --> 00:10:53,800 Ejaaz: the fonts, the brand identity, stuff like that. 164 00:10:53,900 --> 00:10:57,880 Ejaaz: And it generates a completely kind of novel marketing campaign that hasn't existed 165 00:10:57,880 --> 00:11:02,380 Ejaaz: for any other product before or is based on the type of vibe that you generally like. 166 00:11:02,880 --> 00:11:05,760 Ejaaz: And with all these things, I think that, well, 167 00:11:06,270 --> 00:11:09,650 Ejaaz: For example, like in previous jobs that I've worked, Josh, I've worked with 168 00:11:09,650 --> 00:11:14,030 Ejaaz: massive marketing departments and a large chunk of their time and money is spent 169 00:11:14,030 --> 00:11:16,450 Ejaaz: on creating these types of photo shoots, gauging the sentiment. 170 00:11:16,630 --> 00:11:19,930 Ejaaz: Now, what I will say, OK, I have to argue the other side here, right? 171 00:11:20,370 --> 00:11:24,350 Ejaaz: I've seen a bunch of demos of these product shoots. And after a while, 172 00:11:24,510 --> 00:11:28,270 Ejaaz: you kind of see a kind of similar sense and style. 173 00:11:28,730 --> 00:11:32,650 Ejaaz: Some of the product shoots and angles are kind of repeated. So if you want to 174 00:11:32,650 --> 00:11:37,210 Ejaaz: get something unique, you do need to be very descriptive and kind of know and see what you want. 175 00:11:37,450 --> 00:11:40,870 Ejaaz: And that argues in favor of the product shoot photographers, 176 00:11:40,870 --> 00:11:44,590 Ejaaz: which already have the ideas and experience, kind of like a movie director. 177 00:11:44,730 --> 00:11:46,930 Ejaaz: They're not going out of job because of video models. 178 00:11:47,110 --> 00:11:50,950 Ejaaz: They kind of know the artistic shots and they just need to kind of upgrade their toolkit per se. 179 00:11:51,130 --> 00:11:53,830 Ejaaz: So I think we're going to see something similar like that rolling out. 180 00:11:53,910 --> 00:11:56,130 Ejaaz: I don't think it's at its final form just yet. 181 00:11:56,290 --> 00:11:59,410 Josh: I think you're right in the sense that like it can get you 80% of the way there, 182 00:11:59,550 --> 00:12:02,190 Josh: but it's not going to give you that professional look so for the 183 00:12:02,190 --> 00:12:04,970 Josh: people who don't have these budgets this is awesome for the people who 184 00:12:04,970 --> 00:12:07,830 Josh: do i mean there's clearly a a gap still that 185 00:12:07,830 --> 00:12:10,830 Josh: exists but you have to assume that gap is going to be filled fairly quickly 186 00:12:10,830 --> 00:12:14,550 Josh: so it generated our dna and you could see it's already working on generating 187 00:12:14,550 --> 00:12:19,050 Josh: these automatic um kind of images that we could use perhaps to post on twitter 188 00:12:19,050 --> 00:12:23,370 Josh: to get people excited i mean it shows master the llm power shift it understands 189 00:12:23,370 --> 00:12:28,650 Josh: the topics that we're talking about the trends um opus the deep dive context mastery This just. 190 00:12:28,650 --> 00:12:33,410 Ejaaz: Jogged my memory. This reminds me of the live demo we went through with Nano Banana. 191 00:12:33,670 --> 00:12:37,110 Ejaaz: And I believe this is using Nano Banana on the back end, right? 192 00:12:37,390 --> 00:12:39,250 Ejaaz: Google's image generation model. 193 00:12:39,510 --> 00:12:44,170 Josh: I believe so. So we're using Nano Banana Pro. And we're going to test that by 194 00:12:44,170 --> 00:12:47,670 Josh: using it for a photo shoot. So what I did is I stole a little hat. 195 00:12:47,970 --> 00:12:50,130 Josh: And I'm going to drop the hat in here. 196 00:12:50,840 --> 00:12:53,700 Josh: And we can have it do a photo shoot in real 197 00:12:53,700 --> 00:12:57,240 Josh: time that allows us to generate our 198 00:12:57,240 --> 00:13:00,420 Josh: own limitless branded imagery so 199 00:13:00,420 --> 00:13:03,400 Josh: we'll load that up we'll change the format to 200 00:13:03,400 --> 00:13:07,220 Josh: be let's say square and then we could choose the photo shoot template so maybe 201 00:13:07,220 --> 00:13:12,080 Josh: we'll choose flat lay in use and then we'll do two of these product shots here 202 00:13:12,080 --> 00:13:15,500 Josh: and we'll say it looks good and generate and now while that generates we can 203 00:13:15,500 --> 00:13:18,220 Josh: talk about yeah maybe the things that are powering this so one is nano banana 204 00:13:18,220 --> 00:13:23,100 Josh: pro but two is i mean this is using gemini 3.1 pro as well which is the new model, 205 00:13:23,860 --> 00:13:25,300 Josh: brand new just came out. 206 00:13:25,300 --> 00:13:28,540 Ejaaz: Last week a lot of a lot of new launches um 207 00:13:28,540 --> 00:13:31,520 Ejaaz: i was also looking at the costing for a typical product 208 00:13:31,520 --> 00:13:34,540 Ejaaz: photo shoot josh it's between 500 bucks to 209 00:13:34,540 --> 00:13:37,320 Ejaaz: 5 000 bucks on the high-end scale and a lot of the images 210 00:13:37,320 --> 00:13:40,080 Ejaaz: that we're seeing here are on that high-end scale it takes 211 00:13:40,080 --> 00:13:43,140 Ejaaz: a while to kind of like edit in post-production and make 212 00:13:43,140 --> 00:13:46,640 Ejaaz: things like really crystal clear so that's super impressive um 213 00:13:46,640 --> 00:13:50,100 Ejaaz: what i'm also impressed by is this is like available to quite a few uh different 214 00:13:50,100 --> 00:13:53,480 Ejaaz: regions so typically when google launches their ai products it's just for the 215 00:13:53,480 --> 00:13:56,860 Ejaaz: u.s only but this is available in the u.s canada australia new zealand so think 216 00:13:56,860 --> 00:14:01,800 Ejaaz: of like the entire swathes of people that now get access to a tool that cost 217 00:14:01,800 --> 00:14:05,960 Ejaaz: them cents instead of these thousands of dollars it's just so cool yeah 218 00:14:05,960 --> 00:14:09,380 Josh: It's um it's amazing and i think it's a testament to the direction that they're 219 00:14:09,380 --> 00:14:13,980 Josh: going towards which is just kind of complete and total value creation for anyone out there. 220 00:14:14,120 --> 00:14:17,900 Josh: Like they're working on, they have their coding agent, they have their IDE, 221 00:14:18,200 --> 00:14:24,500 Josh: they have their video generation with VO3, they have audio generation now, they have brands and... 222 00:14:25,360 --> 00:14:28,880 Josh: Brand DNA and I guess marketing and understanding of that. And there really 223 00:14:28,880 --> 00:14:31,140 Josh: is no industry that's safe from these tools. 224 00:14:31,200 --> 00:14:34,660 Josh: And as these tools continue to get better, they're just going to become way 225 00:14:34,660 --> 00:14:38,840 Josh: more proficient and way more optimal as it relates to kind of not replacing 226 00:14:38,840 --> 00:14:42,760 Josh: these jobs, but enhancing and augmenting people's abilities that work in these spaces. 227 00:14:42,980 --> 00:14:48,200 Ejaaz: But let me ask you this question, Josh. Do you feel that way about all these products in unison 228 00:14:48,980 --> 00:14:54,940 Ejaaz: today because my take is it's great individually but i want an entire suite 229 00:14:54,940 --> 00:14:59,240 Ejaaz: that can manage all of this for me and they're working towards that right with google ai suite 230 00:14:59,240 --> 00:15:02,280 Josh: Yeah there's going to be a single comprehensive tool 231 00:15:02,280 --> 00:15:05,180 Josh: that has all of these things and it's exciting 232 00:15:05,180 --> 00:15:08,140 Josh: so we have the outputs here which um i didn't load the 233 00:15:08,140 --> 00:15:10,980 Josh: logo onto and i didn't ask to load the logo onto so perhaps 234 00:15:10,980 --> 00:15:13,620 Josh: that's why our logo isn't actually on there but i mean 235 00:15:13,620 --> 00:15:16,460 Josh: to your point you can see the bankless one that showed you earlier looks very 236 00:15:16,460 --> 00:15:19,420 Josh: similar to this one yep and the model actually looks 237 00:15:19,420 --> 00:15:22,500 Josh: like kind of similar to the last one and you can see there are 238 00:15:22,500 --> 00:15:25,720 Josh: some i guess kind of restraints and limitations 239 00:15:25,720 --> 00:15:30,340 Josh: as it relates to the quality of the outputs but to have something this good 240 00:15:30,340 --> 00:15:34,660 Josh: this quickly this easily it's really impressive and it's a fun tool to use um 241 00:15:34,660 --> 00:15:38,700 Josh: and like i mentioned earlier under the hood gemini 3.1 it just came out last 242 00:15:38,700 --> 00:15:43,540 Josh: week um there's a lot of good and bad news about it so maybe we should talk about that next. 243 00:15:43,540 --> 00:15:49,960 Ejaaz: Yes okay let's get back to being suited and booted uh gentlemen we have to get 244 00:15:49,960 --> 00:15:56,020 Ejaaz: into the llms um google dropped a brand new model and on paper 245 00:15:56,680 --> 00:16:01,040 Ejaaz: it's pretty damn impressive. What I've got in showing on the screen right here 246 00:16:01,040 --> 00:16:03,540 Ejaaz: is the ARC AGI2 benchmark. 247 00:16:03,840 --> 00:16:08,400 Ejaaz: Now, just for context here, there was an ARC AGI1, but the models became so 248 00:16:08,400 --> 00:16:12,220 Ejaaz: good that they needed to create a brand new test to make sure that these models 249 00:16:12,220 --> 00:16:14,300 Ejaaz: were actually getting more intelligent. 250 00:16:14,500 --> 00:16:17,040 Ejaaz: So ARC AGI2 is a benchmark that was set. 251 00:16:17,280 --> 00:16:22,300 Ejaaz: And I just want to focus on a little difference here. And by little, I mean a lot. 252 00:16:22,660 --> 00:16:23,760 Josh: This is more than a double. 253 00:16:23,980 --> 00:16:29,260 Ejaaz: That's more than a double. It's a 46 percentage point increase from Gemini 3 Pro. 254 00:16:29,420 --> 00:16:33,260 Ejaaz: And I want to point out that this is a 0.1 version update. 255 00:16:33,500 --> 00:16:38,460 Ejaaz: We've gone from 3 to 3.1. This isn't even Gemini 4 yet. And we've seen such 256 00:16:38,460 --> 00:16:42,420 Ejaaz: a major leap in intelligence and reasoning. And the reason... 257 00:16:42,420 --> 00:16:47,440 Josh: I just looked this up, actually, just real quick. It came out in November 18th of 2025. 258 00:16:47,800 --> 00:16:48,480 Ejaaz: So this is 259 00:16:48,480 --> 00:16:49,760 Josh: Less than four months. 260 00:16:49,920 --> 00:16:54,940 Ejaaz: The iteration cycles for these model updates are getting astounding, to be honest. 261 00:16:55,380 --> 00:17:01,060 Ejaaz: We've spoken about Claude Opus 4.6 and ChatGPT 5.3 Codex. 262 00:17:01,200 --> 00:17:05,140 Ejaaz: These are the new coding models from Anthropic and OpenAI, respectively. 263 00:17:05,840 --> 00:17:09,620 Ejaaz: Those released within a week, sorry, within an hour of each other. 264 00:17:09,840 --> 00:17:13,600 Ejaaz: And their previous model version updates were three weeks prior to that. 265 00:17:13,840 --> 00:17:17,400 Ejaaz: So the fact that these models are improving at a rapid rate doesn't surprise me. 266 00:17:17,600 --> 00:17:21,660 Ejaaz: The secret source behind that is these models are most likely working on themselves, 267 00:17:21,920 --> 00:17:24,560 Ejaaz: i.e. they're reviewing their own code and updating themselves, 268 00:17:24,740 --> 00:17:28,520 Ejaaz: which is a scary topic, which we'll get into another time. But back to Gemini 3.1 Pro. 269 00:17:29,160 --> 00:17:36,020 Ejaaz: It excels in two major leaps. Number one, what you're seeing on the screen here, AGI 2, reasoning. 270 00:17:36,300 --> 00:17:40,920 Ejaaz: So it understands what you're saying. It goes deep into the weeds of your prompt 271 00:17:40,920 --> 00:17:43,060 Ejaaz: and gives you a really well thought out answer. 272 00:17:43,240 --> 00:17:47,420 Ejaaz: Again, on paper. The other thing that it's really good at is coding. 273 00:17:47,620 --> 00:17:53,980 Ejaaz: It's insanely good at generating visuals specifically using code and understanding 274 00:17:53,980 --> 00:17:56,580 Ejaaz: the physics of simulations that you create. 275 00:17:56,970 --> 00:18:01,970 Ejaaz: But I have to point out that there's been a bunch of bad feedback from people 276 00:18:01,970 --> 00:18:03,250 Ejaaz: actually testing these things. 277 00:18:03,410 --> 00:18:06,630 Ejaaz: So what I've just explained to you has been on paper, but in practice, 278 00:18:06,950 --> 00:18:11,870 Ejaaz: people observe that the reasoning actually kind of defeats its own self. 279 00:18:12,090 --> 00:18:15,310 Ejaaz: It starts getting into these thinking loops where it starts doubting itself, 280 00:18:15,750 --> 00:18:19,390 Ejaaz: questioning its own answers, and you end up waiting 10 minutes for what would 281 00:18:19,390 --> 00:18:21,670 Ejaaz: otherwise be a very simple answer. 282 00:18:21,930 --> 00:18:25,910 Ejaaz: And then other people's experience is when they use the web app version to access 283 00:18:25,910 --> 00:18:30,610 Ejaaz: Gemini 3.1, it's actually a much more reduced model that just agrees with them. 284 00:18:30,730 --> 00:18:36,390 Ejaaz: So it kind of sounds a lot like GPT-4.0, which is known for being very agreeable, 285 00:18:36,570 --> 00:18:40,890 Ejaaz: which OpenAir actually decommissioned last week. So there's a lot of mixed reviews for this. 286 00:18:41,090 --> 00:18:45,590 Ejaaz: There are a few cool examples from the big man himself, Jeff Dean. 287 00:18:46,450 --> 00:18:52,930 Ejaaz: He's showing us an example here of Gemini 3 Pro on the left and Gemini 3.1 Pro on the right. 288 00:18:53,110 --> 00:18:58,290 Ejaaz: As you can see, the SVG generations are just a stark difference, right? 289 00:18:58,730 --> 00:19:03,430 Ejaaz: The physics is super cool. The animations, look at the backgrounds on these things. It's so cool. 290 00:19:04,470 --> 00:19:08,070 Ejaaz: This is a really cool demo of someone using it to do urban planning. 291 00:19:08,170 --> 00:19:10,590 Ejaaz: I'm going to mute it so I can show you the start over here. 292 00:19:11,050 --> 00:19:15,670 Ejaaz: Um, it's just super cool how we can map track and figure out spatial awareness. 293 00:19:15,850 --> 00:19:19,970 Ejaaz: It's a really spatially astute model, which is just super cool for creating 294 00:19:19,970 --> 00:19:22,250 Ejaaz: kind of games or any kind of simulated demos. 295 00:19:22,410 --> 00:19:25,790 Ejaaz: But yeah, the, the actual practical feedback from people hasn't been that great. 296 00:19:26,050 --> 00:19:29,830 Josh: It's funny to see the evolution of these models in between major releases, 297 00:19:29,830 --> 00:19:32,490 Josh: because I mean, what happens during these incremental releases, 298 00:19:32,510 --> 00:19:35,670 Josh: they're not doing this huge pre-training loop where they're, 299 00:19:35,670 --> 00:19:36,910 Josh: they're building an entirely new model. 300 00:19:37,050 --> 00:19:39,770 Josh: They're taking the existing core model and they're kind of building on top of it. 301 00:19:39,990 --> 00:19:44,930 Josh: And what you find is that these incremental models become spiky in some different areas. 302 00:19:45,070 --> 00:19:48,770 Josh: So some are much better at some things, but the trade-off is that it gets worse. 303 00:19:48,950 --> 00:19:54,810 Josh: I mean, a good example we had is GPT 4.5, I think it was, which was trained 304 00:19:54,810 --> 00:19:58,230 Josh: to be an excellent writer, but it was actually so bad at everything else and 305 00:19:58,230 --> 00:19:59,830 Josh: so expensive that they had to depreciate it. 306 00:20:00,070 --> 00:20:03,710 Josh: And I think what you see throughout these kind of incremental evolutions, 307 00:20:03,890 --> 00:20:07,950 Josh: 3.1 being one of them, is that they're trying to improve them and they will 308 00:20:07,950 --> 00:20:11,690 Josh: improve them in some areas, but there are some unexpected areas in which the 309 00:20:11,690 --> 00:20:13,410 Josh: net quality actually declines. 310 00:20:13,770 --> 00:20:16,650 Josh: So each one has its own personality. It has its own skill set. 311 00:20:16,990 --> 00:20:21,370 Josh: Net net they're better because they've been refined and they've been distilled 312 00:20:21,370 --> 00:20:26,790 Josh: down into just a highly refined model but there are going to be those like downsides 313 00:20:26,790 --> 00:20:29,710 Josh: that we see and i think that's just the natural part of these incremental releases 314 00:20:29,710 --> 00:20:35,070 Josh: is kind of seeing how these these models evolve over time from that huge core base model my. 315 00:20:35,070 --> 00:20:38,330 Ejaaz: Major lesson from this uh i think which is one that we've learned a while ago 316 00:20:38,330 --> 00:20:41,890 Ejaaz: is just don't trust the benchmarks just trust your own experience and obviously 317 00:20:41,890 --> 00:20:46,470 Ejaaz: different people use these models for different things um do i think this is the end of Google? 318 00:20:46,650 --> 00:20:49,710 Ejaaz: No, but they did go on a gargantuan winning streak. 319 00:20:49,930 --> 00:20:52,010 Ejaaz: And this may have subdued them just slightly. 320 00:20:52,450 --> 00:20:56,710 Ejaaz: The good news is I hear that Gemini 4 is cooking up an absolute beast. 321 00:20:57,190 --> 00:20:59,990 Ejaaz: And what I like is that they're integrating. Yeah, I'm integrating. 322 00:21:00,350 --> 00:21:03,590 Ejaaz: What I like is that Google's integrating a lot of these models into a singular 323 00:21:03,590 --> 00:21:04,970 Ejaaz: interface. I mentioned this earlier. 324 00:21:05,150 --> 00:21:08,250 Ejaaz: The one thing that annoys me is I need to go one place for Nano Banana. 325 00:21:08,250 --> 00:21:11,230 Ejaaz: I need to go another place for VO3, their video generation model. 326 00:21:11,330 --> 00:21:13,910 Ejaaz: And then I need to go another place to access Gemini 3.1. 327 00:21:14,150 --> 00:21:17,110 Ejaaz: Sometimes you can access it through the API, sometimes it's through the web 328 00:21:17,110 --> 00:21:20,130 Ejaaz: app, sometimes randomly it's for free through Google search. 329 00:21:20,390 --> 00:21:24,110 Ejaaz: And so I'm like, I just want one place to go to. I'll give you my money. 330 00:21:24,290 --> 00:21:26,810 Ejaaz: Just let me combine all these things into one experience. 331 00:21:26,990 --> 00:21:31,250 Ejaaz: And that's all I want for like a single price, a single Netflix subscription. 332 00:21:31,630 --> 00:21:34,310 Ejaaz: And they're moving towards that phase, which I think is great. 333 00:21:34,450 --> 00:21:39,010 Ejaaz: And so my guess is maybe they kind of dropped the ball here because they're 334 00:21:39,010 --> 00:21:40,410 Ejaaz: focusing on all that other stuff. 335 00:21:40,770 --> 00:21:45,130 Ejaaz: The other part I want to mention is Google's distribution is just insane. 336 00:21:45,410 --> 00:21:50,870 Ejaaz: Just through the other two features, Liria and Pomelli, they've been able to 337 00:21:50,870 --> 00:21:54,610 Ejaaz: distribute these to tens of millions of people on day one. 338 00:21:54,710 --> 00:21:57,070 Ejaaz: And that feedback loop cannot be understated. 339 00:21:57,230 --> 00:22:00,150 Ejaaz: So whereas Google, you know, may have this initial reaction, 340 00:22:00,490 --> 00:22:02,390 Ejaaz: they have been the quickest to bounce back. 341 00:22:02,510 --> 00:22:06,590 Ejaaz: Don't forget where Google was two years ago when they had the worst AI model. 342 00:22:06,710 --> 00:22:10,910 Ejaaz: And now they're like fighting for the throne pretty much. So I wouldn't count them out. It's cool. 343 00:22:11,190 --> 00:22:15,130 Josh: Yeah, it feels like the Adobe Creative Cloud Suite on steroids for Google and 344 00:22:15,130 --> 00:22:17,010 Josh: for creators. And I think that's something... 345 00:22:17,090 --> 00:22:20,090 Josh: I can get really excited about. And what you saw with these demos today is it's 346 00:22:20,090 --> 00:22:23,210 Josh: real technology that's here today and it's improving very, very quickly. 347 00:22:23,210 --> 00:22:26,950 Josh: So I'm excited for 3.2, 3.3, and then 4.0 whenever that comes, 348 00:22:27,030 --> 00:22:28,370 Josh: which I'm sure is going to shock the world. 349 00:22:28,490 --> 00:22:31,670 Josh: But that wraps everything up for our fun little demo day. 350 00:22:31,930 --> 00:22:35,070 Josh: I would encourage you to go and try these things. I mean, a lot of them are 351 00:22:35,070 --> 00:22:36,770 Josh: just available for free, which is pretty cool. 352 00:22:36,910 --> 00:22:39,570 Josh: You can just go and test them out, create a song, send it to your friend, 353 00:22:39,790 --> 00:22:41,330 Josh: roast them, whatever it may be. 354 00:22:41,450 --> 00:22:45,830 Josh: It's a fun way to kind of help creators, help marketers. 355 00:22:45,830 --> 00:22:48,930 Josh: These tools are awesome and they're coming so quickly so if 356 00:22:48,930 --> 00:22:52,550 Josh: you enjoyed please don't forget share it with a friend subscribe 357 00:22:52,550 --> 00:22:55,750 Josh: to our newsletter it comes out twice a week every single week last week we had 358 00:22:55,750 --> 00:22:59,530 Josh: our biggest week ever we never had more views than last week so the mission 359 00:22:59,530 --> 00:23:02,850 Josh: is working and it's thanks to you guys for sharing it with your friends and 360 00:23:02,850 --> 00:23:06,410 Josh: liking it and commenting rating five stars on your favorite podcast platform 361 00:23:06,410 --> 00:23:09,430 Josh: each has any final notes before we sign off here i. 362 00:23:09,430 --> 00:23:14,050 Ejaaz: Have one small piece of feedback sorry feedback i have a small piece of homework 363 00:23:14,050 --> 00:23:18,150 Ejaaz: for those of you who are up to it um i'm desperate to hear your music creations 364 00:23:18,150 --> 00:23:23,690 Ejaaz: um or even your product photo shoots so try and find josh and i uh on twitter 365 00:23:23,690 --> 00:23:27,450 Ejaaz: we'll link to our handles or uh profile pages below 366 00:23:28,110 --> 00:23:32,410 Ejaaz: and dm us i want to hear some of your creations because i'm genuinely uh interested 367 00:23:32,410 --> 00:23:36,690 Ejaaz: in how creative people can get um and maybe it's a breakup text or maybe it's 368 00:23:36,690 --> 00:23:40,370 Ejaaz: telling your girlfriend to uh i don't know make something for you by the time 369 00:23:40,370 --> 00:23:42,810 Ejaaz: you get home i don't know i'm gonna i'm gonna try it out later and see what 370 00:23:42,810 --> 00:23:45,410 Ejaaz: it's like. But yeah, let us know your questions. I want to see what you guys are up to. 371 00:23:46,210 --> 00:23:48,830 Josh: Awesome. Well, thanks so much for watching. And yeah, we'll see you guys in the next one.