WEBVTT 00:00:02.239 --> 00:00:04.780 Welcome to the AI Chat Podcast. I'm your host, 00:00:04.860 --> 00:00:07.179 Jaden Schaefer. Every day I cover cutting -edge 00:00:07.179 --> 00:00:09.320 AI news and talk with the leaders behind it, 00:00:09.339 --> 00:00:11.300 breaking down what it means for your life and 00:00:11.300 --> 00:00:17.530 business. Samsung held acquisition talks with 00:00:17.530 --> 00:00:19.149 Physical Intelligence, the robotics company, 00:00:19.250 --> 00:00:22.109 and this is making the OpenAI robotics push a 00:00:22.109 --> 00:00:24.390 little bit more complicated. Meta has launched 00:00:24.390 --> 00:00:27.230 something called Content Seal AI Watermarking. 00:00:27.230 --> 00:00:29.149 This is three years after they promised to put 00:00:29.149 --> 00:00:32.409 AI labels on content. Synthesia launched Roleplay 00:00:32.409 --> 00:00:34.689 Sessions. Their goal is to let employees talk 00:00:34.689 --> 00:00:37.869 with AI to improve their conversations in a bunch 00:00:37.869 --> 00:00:39.850 of different areas, and they get judged on it. 00:00:40.030 --> 00:00:42.770 Samsung is in talks to back Mistral at a 20 billion 00:00:42.770 --> 00:00:46.429 euro valuation. Next era. Duke and 200 other 00:00:46.429 --> 00:00:49.130 utility companies signed Trump's AI electricity 00:00:49.130 --> 00:00:51.149 pledge. Last week, I just started running my 00:00:51.149 --> 00:00:53.549 first ever Facebook ad campaign for my startup 00:00:53.549 --> 00:00:56.729 AI box dot AI, which up until now has grown completely 00:00:56.729 --> 00:00:59.450 organically. We're ready to invest a little bit 00:00:59.450 --> 00:01:02.210 into paid marketing. The top performing ad I 00:01:02.210 --> 00:01:03.350 wanted to tell you about because I thought it 00:01:03.350 --> 00:01:05.230 was pretty funny. It's basically it looks like 00:01:05.230 --> 00:01:07.930 a news clip. It's like says tech alert. And it's 00:01:07.930 --> 00:01:10.170 me in the kind of big banner image looking shocked. 00:01:10.290 --> 00:01:12.129 And there's like in the top corner, there's like 00:01:12.129 --> 00:01:15.019 a zoom in on like an image and camera lens. And 00:01:15.019 --> 00:01:17.959 the headline on it says, now Claude makes images, 00:01:18.079 --> 00:01:20.579 no extra apps needed. And the caption on it says, 00:01:20.719 --> 00:01:23.359 get Claude to generate images, every top image 00:01:23.359 --> 00:01:25.659 model right inside the chat, no new apps, no 00:01:25.659 --> 00:01:27.739 API keys. Anyways, I thought it was funny. This 00:01:27.739 --> 00:01:31.280 is basically, I've tried 77 different ad types 00:01:31.280 --> 00:01:34.379 and styles and variations. And that is the winner 00:01:34.379 --> 00:01:37.159 of 77. So I figured if I was going to tell you 00:01:37.159 --> 00:01:39.780 guys about the new MCP on AI box that lets you 00:01:39.780 --> 00:01:42.980 generate 80 different AI models inside of Claude. 00:01:43.340 --> 00:01:44.879 going to just tell you guys the top performing 00:01:44.879 --> 00:01:47.140 Facebook ad. And yeah, it's super cool. I've 00:01:47.140 --> 00:01:49.140 been using it all week. Anytime that you want 00:01:49.140 --> 00:01:51.260 to generate an image inside of Claude, you could 00:01:51.260 --> 00:01:53.560 just say Claude generate an image of X, Y, Z. 00:01:53.659 --> 00:01:54.920 And you just basically put in the prompt that 00:01:54.920 --> 00:01:56.840 you give to chat GPT or any other image generator, 00:01:56.959 --> 00:01:59.219 and it will pull the image straight into Claude. 00:01:59.219 --> 00:02:01.319 Claude can then go use it. The thing that I mostly 00:02:01.319 --> 00:02:03.799 am using it for number one, uh, Facebook ads, 00:02:03.900 --> 00:02:05.980 because I've been creating 77 of them this week. 00:02:06.140 --> 00:02:08.500 Also though, I've done it with a bunch of different, 00:02:08.620 --> 00:02:10.819 uh, websites where I'm like, Hey, I generate 00:02:10.819 --> 00:02:12.879 this website and we all know what. cloud website 00:02:12.879 --> 00:02:15.180 looks like it's pretty generic. And so I'll say, 00:02:15.199 --> 00:02:17.460 hey, generate custom graphics for each section 00:02:17.460 --> 00:02:19.780 of this website. Sometimes I'll even say, hey, 00:02:19.860 --> 00:02:21.840 go generate, you know, take a screenshot, I'll 00:02:21.840 --> 00:02:23.139 take a screenshot of what the website looks like. 00:02:23.180 --> 00:02:25.780 And I'm like, regenerate this website, an image 00:02:25.780 --> 00:02:27.360 of this website where the design looks a lot 00:02:27.360 --> 00:02:30.199 better. And it will use an image generator to 00:02:30.199 --> 00:02:31.699 kind of fill it out with custom graphics, make 00:02:31.699 --> 00:02:33.860 it look better. And then I'll say, hey, now go 00:02:33.860 --> 00:02:35.900 and take each of the little custom graphics in 00:02:35.900 --> 00:02:38.219 this image and generate an image for each of 00:02:38.219 --> 00:02:40.539 those and embed it into the website. This is 00:02:40.539 --> 00:02:42.650 basically turned my website. and apps that I've 00:02:42.650 --> 00:02:45.069 done from looking super basic and generic to 00:02:45.069 --> 00:02:47.629 looking super customizable or custom, right? 00:02:47.710 --> 00:02:49.650 So anyways, very cool use case. If you want to 00:02:49.650 --> 00:02:53.210 check it out, it's AIbox .ai slash MCP. I'll 00:02:53.210 --> 00:02:55.210 leave a link in the description. Speaking of 00:02:55.210 --> 00:02:57.210 Cloud and Anthropic, Anthropic held acquisition 00:02:57.210 --> 00:03:00.090 talks with Physical Intelligence this spring. 00:03:00.229 --> 00:03:04.069 It is an $11 billion robotics startup. And the 00:03:04.069 --> 00:03:06.250 deal didn't actually close, but both Anthropic 00:03:06.250 --> 00:03:09.210 and OpenAI are now really racing to try and build 00:03:09.210 --> 00:03:11.610 robotic capabilities. because they're both going 00:03:11.610 --> 00:03:14.370 for the new big IPO. They're both going to, you 00:03:14.370 --> 00:03:16.229 know, debut on the stock market later this year. 00:03:16.310 --> 00:03:18.830 And robotics is such a hot area. It's kind of 00:03:18.830 --> 00:03:20.449 interesting SpaceX, you know, with the Space 00:03:20.449 --> 00:03:22.669 Data Center route, which Sam Altman says is a 00:03:22.669 --> 00:03:24.870 scam. Elon Musk says it's going to happen. But 00:03:24.870 --> 00:03:26.710 whatever way you want to cut it, it was a very 00:03:26.710 --> 00:03:29.189 novel, very new kind of competitive advantage 00:03:29.189 --> 00:03:32.120 that SpaceX had. Open AI and Anthropic, I think 00:03:32.120 --> 00:03:34.080 both want to get into something physical. Open 00:03:34.080 --> 00:03:36.379 AI has done a lot with physical data centers. 00:03:36.460 --> 00:03:38.379 I think that's going to bode really well for 00:03:38.379 --> 00:03:40.240 them. But I think both of them want to get into 00:03:40.240 --> 00:03:42.719 robotics, which is kind of the new phase of AI 00:03:42.719 --> 00:03:45.240 where we're going to have robots that have AI 00:03:45.240 --> 00:03:47.340 embedded into their brains. One thing I wanted 00:03:47.340 --> 00:03:49.159 to clarify is that while physical intelligence 00:03:49.159 --> 00:03:51.280 is a quote unquote robotics company, they don't 00:03:51.280 --> 00:03:53.379 have robots. They've raised a billion dollars 00:03:53.379 --> 00:03:55.819 and they make a very popular AI model that's 00:03:55.819 --> 00:03:57.840 used for controlling a lot of robots. It's used 00:03:57.840 --> 00:04:01.000 in academic research. It's called PI's. 0 .5, 00:04:01.060 --> 00:04:04.439 and that's the symbol pi. OpenAI acquired 17 00:04:04.439 --> 00:04:07.500 companies since 2023, and they just launched 00:04:07.500 --> 00:04:10.000 an official robotics lab in San Francisco. Anthropic 00:04:10.000 --> 00:04:12.879 is testing robot control through their safety 00:04:12.879 --> 00:04:15.439 research group. Both of these companies share 00:04:15.439 --> 00:04:18.180 investors in physical intelligence, including 00:04:18.180 --> 00:04:20.699 Coastal Adventures, Thrive Capital, and Founders 00:04:20.699 --> 00:04:23.259 Fund. So there is definitely some overlap in 00:04:23.259 --> 00:04:26.019 their different robotic strategies. I think if 00:04:26.019 --> 00:04:27.959 you understand the physical world, that's going 00:04:27.959 --> 00:04:29.990 to be super essential. for the next generation 00:04:29.990 --> 00:04:31.970 of these AI systems that OpenAI and Anthropic 00:04:31.970 --> 00:04:34.250 are both coming out with. So a model like this 00:04:34.250 --> 00:04:37.089 that can help robotics would be incredibly valuable, 00:04:37.310 --> 00:04:39.189 not just for the model, but because it has the 00:04:39.189 --> 00:04:41.230 ability to go and collect real world data when 00:04:41.230 --> 00:04:43.329 it's put inside of a robot. You know, we have 00:04:43.329 --> 00:04:45.470 these different companies that are paying, I 00:04:45.470 --> 00:04:47.410 think Swift might be one of them, but it's literally 00:04:47.410 --> 00:04:50.930 just paying people to accept a robot or a person 00:04:50.930 --> 00:04:53.149 in their home to come and clean their home while 00:04:53.149 --> 00:04:55.670 wearing a hat with a camera on it. And it's essentially 00:04:55.670 --> 00:04:57.089 the camera's recording that's going to be used 00:04:57.089 --> 00:04:59.649 for training robots. But if you actually have 00:04:59.649 --> 00:05:01.329 a model that's used for the robots, then you 00:05:01.329 --> 00:05:02.910 can embed that in the robots and then you can 00:05:02.910 --> 00:05:05.430 send them out to do tasks. You can have the robots 00:05:05.430 --> 00:05:07.990 capturing data to help train other robots. I 00:05:07.990 --> 00:05:09.250 mean, there's a lot of different interesting 00:05:09.250 --> 00:05:11.370 use cases. I mean, maybe at that point, if the 00:05:11.370 --> 00:05:13.930 robots are training, they'll train in a bad way. 00:05:13.990 --> 00:05:15.610 So maybe we still want to get the humans for 00:05:15.610 --> 00:05:17.629 the training data right now. But either way, 00:05:17.689 --> 00:05:19.589 this is a really critical company that both of 00:05:19.589 --> 00:05:22.319 them are looking at. Meta has just launched Content 00:05:22.319 --> 00:05:24.759 Seal. They launched it just this month. It is 00:05:24.759 --> 00:05:27.040 an invisible watermark that identifies images 00:05:27.040 --> 00:05:29.639 made with their new Muse model. So three years 00:05:29.639 --> 00:05:31.660 ago, they promised, you know, they're like, hey, 00:05:31.680 --> 00:05:33.279 look, we're going to get these labels on all 00:05:33.279 --> 00:05:34.959 of the different AI generated content that's 00:05:34.959 --> 00:05:37.160 on like Facebook and Instagram and stuff. This 00:05:37.160 --> 00:05:39.860 watermark in particular only works on Muse images. 00:05:40.139 --> 00:05:42.740 So obviously there's millions of AI photos that 00:05:42.740 --> 00:05:45.259 Meta generated since 2023, and those are not 00:05:45.259 --> 00:05:47.699 detectable by the tool. I think there's a bunch 00:05:47.699 --> 00:05:50.399 of different articles online where people are 00:05:50.399 --> 00:05:52.519 kind of begging on Meta for that. I'm going to 00:05:52.519 --> 00:05:54.500 be honest. At the end of the day, I think most 00:05:54.500 --> 00:05:57.779 image generation models with AI, like they're 00:05:57.779 --> 00:06:00.060 getting really good, but they're not perfect. 00:06:00.120 --> 00:06:02.029 And they're pretty. I'd say they're detectable 00:06:02.029 --> 00:06:04.649 now. We're not like past the undetectable phase. 00:06:04.889 --> 00:06:06.930 So as far as like a lot of people are kind of 00:06:06.930 --> 00:06:08.750 begging on Meta saying like, look, they promised 00:06:08.750 --> 00:06:12.230 this and all of these images since 2023 are undetectable. 00:06:12.329 --> 00:06:14.529 And Meta's models in particular, in my opinion, 00:06:14.529 --> 00:06:17.230 are pretty bad. And so I'm not super concerned 00:06:17.230 --> 00:06:19.170 about all of their old image models. This new 00:06:19.170 --> 00:06:21.569 Muse image model is really good. And so I'm happy 00:06:21.569 --> 00:06:24.449 that they're adding it here. Detection only runs 00:06:24.449 --> 00:06:26.629 through a rate limited web tool though. So it's 00:06:26.629 --> 00:06:28.810 not actually inside of Meta AI where users are 00:06:28.810 --> 00:06:31.029 actually seeing all the AI. generated images, 00:06:31.209 --> 00:06:33.430 and Meta hasn't said when or if they're going 00:06:33.430 --> 00:06:36.089 to add detection to their chatbot. ContentSeal 00:06:36.089 --> 00:06:38.689 is basically copying what Google did with SynthID, 00:06:38.949 --> 00:06:41.389 which OpenAI already uses, but Meta built their 00:06:41.389 --> 00:06:43.810 own standard. They didn't really want, I think 00:06:43.810 --> 00:06:45.410 they were all, even it's actually funny. So they 00:06:45.410 --> 00:06:47.529 were actually on the same steering committee 00:06:47.529 --> 00:06:50.149 for C2PA, which is kind of an industry -wide 00:06:50.149 --> 00:06:52.269 standard, but they just went and rolled their 00:06:52.269 --> 00:06:53.689 own thing. I mean, tech companies are going to 00:06:53.689 --> 00:06:55.649 do this. Synthesia, a startup that I have been 00:06:55.649 --> 00:06:58.569 using for many, many years, way before ChatGPT, 00:06:58.670 --> 00:07:01.279 launched a new product. called Roleplay Sessions 00:07:01.279 --> 00:07:03.540 on Wednesday, and basically it lets employees 00:07:03.540 --> 00:07:06.680 practice conversations with AI avatars, and then 00:07:06.680 --> 00:07:09.160 it's going to grade them against a rubric. So 00:07:09.160 --> 00:07:11.980 this is a company based out of the UK, and they're 00:07:11.980 --> 00:07:13.879 basically betting that enterprises are going 00:07:13.879 --> 00:07:16.160 to pay more for proof that training actually 00:07:16.160 --> 00:07:19.459 changes behavior, not just cheaper video clips. 00:07:19.720 --> 00:07:22.319 So a lot of the early customers for this include 00:07:22.319 --> 00:07:24.939 one of Europe's top three firms by market value, 00:07:25.040 --> 00:07:27.899 a top five Fortune 100 firm, and one of the world's 00:07:27.899 --> 00:07:29.519 largest recruitment companies. It's always funny 00:07:29.519 --> 00:07:31.540 to me that like no one wants to put their names 00:07:31.540 --> 00:07:34.120 on like, hey, we're using AI to do everyone's 00:07:34.120 --> 00:07:36.160 training. But obviously some of the biggest companies 00:07:36.160 --> 00:07:39.339 in the world are using it. The product uses OpenAI's 00:07:39.339 --> 00:07:41.699 reasoning engine for the actual grading. So it's 00:07:41.699 --> 00:07:44.699 just plugged into, you know, OpenAI. Synthesia 00:07:44.699 --> 00:07:46.779 keeps its avatars and voice technology proprietary. 00:07:47.540 --> 00:07:49.439 So, you know, they're very competitive there. 00:07:49.500 --> 00:07:51.040 That's kind of what they've always done is had 00:07:51.040 --> 00:07:53.939 these, you know, avatars, these AI generated 00:07:53.939 --> 00:07:56.829 avatars. Voice technology, they do a great job 00:07:56.829 --> 00:07:59.730 there. But then once someone does a test with 00:07:59.730 --> 00:08:01.329 one of these avatars, it's going to spit it over 00:08:01.329 --> 00:08:03.829 to OpenAI to give them a grade. Synthesia plans 00:08:03.829 --> 00:08:06.149 to expand from enterprise only. They're going 00:08:06.149 --> 00:08:08.050 to try to target small businesses and schools 00:08:08.050 --> 00:08:11.250 in the next couple months. The cost to run a 00:08:11.250 --> 00:08:13.269 lot of these AI conversations keeps going down, 00:08:13.310 --> 00:08:15.189 especially when you start using open source models 00:08:15.189 --> 00:08:17.529 or other things once they get good enough. They 00:08:17.529 --> 00:08:18.930 don't have to do everything, but for certain 00:08:18.930 --> 00:08:21.290 tasks, the cost goes down. And so then all of 00:08:21.290 --> 00:08:23.430 a sudden, small businesses and schools are a 00:08:23.430 --> 00:08:25.639 better target audience for them. Right now, I 00:08:25.639 --> 00:08:27.579 think this is positioning Synthesia as kind of 00:08:27.579 --> 00:08:30.259 a performance management platform. So it's not 00:08:30.259 --> 00:08:32.840 just an AI video vendor. It's always interesting 00:08:32.840 --> 00:08:35.259 to me when these AI companies start going more 00:08:35.259 --> 00:08:37.639 deeper into one particular niche or another. 00:08:37.779 --> 00:08:40.080 This is a company that like I've used their AI 00:08:40.080 --> 00:08:42.759 avatars for a million years. I used to use them 00:08:42.759 --> 00:08:45.480 years ago for generating podcasts. But right 00:08:45.480 --> 00:08:46.860 now they're making this big shift and they're 00:08:46.860 --> 00:08:50.159 going way more deep on kind of like skills and 00:08:50.159 --> 00:08:53.460 leadership and some of these measurable improvement. 00:08:53.899 --> 00:08:55.919 areas and where you can, you know, get analytics. 00:08:56.000 --> 00:08:57.840 So you can have someone go and talk through, 00:08:57.899 --> 00:09:00.799 you know, with their avatar on what they do in 00:09:00.799 --> 00:09:02.659 a certain situation and it will evaluate them 00:09:02.659 --> 00:09:04.759 for leadership or it'll, you know, be like a 00:09:04.759 --> 00:09:06.820 sales coach. They're selling the avatar something 00:09:06.820 --> 00:09:09.100 and it will evaluate them on that. So honestly, 00:09:09.240 --> 00:09:11.659 very interesting use case. Samsung is currently 00:09:11.659 --> 00:09:14.700 in talks to invest in the French AI startup Mistral 00:09:14.700 --> 00:09:18.100 at a 20 billion euro valuation that is more than 00:09:18.100 --> 00:09:22.539 triple its value in mid 2024. This would make 00:09:22.539 --> 00:09:25.179 Mistral Europe's only Frontier AI lab, which 00:09:25.179 --> 00:09:27.340 is now competing with OpenAI and Anthropic. It 00:09:27.340 --> 00:09:30.620 would also give Samsung a direct stake in this 00:09:30.620 --> 00:09:32.539 new company that actually powers all of their 00:09:32.539 --> 00:09:35.159 Galaxy phones already. Mistral also just raised 00:09:35.159 --> 00:09:38.259 $5 .8 billion in June. So they raised that from 00:09:38.259 --> 00:09:40.259 General Catalyst that was kind of leading that. 00:09:40.360 --> 00:09:43.120 And now this $20 billion valuation, it would 00:09:43.120 --> 00:09:46.860 be a 3 .4x jump in under 18 months. So pretty 00:09:46.860 --> 00:09:49.639 incredible. Definitely on track with all of the 00:09:49.639 --> 00:09:53.070 US Frontier lab growth rates. already integrates 00:09:53.070 --> 00:09:55.850 Mistra models into their Galaxy device features 00:09:55.850 --> 00:09:58.750 in some specific markets. They're trying to reduce 00:09:58.750 --> 00:10:01.309 their reliance on Google Gemini and offer region 00:10:01.309 --> 00:10:04.009 -specific choices, which is interesting. I think 00:10:04.009 --> 00:10:06.129 a lot of places in Europe, this is what they're 00:10:06.129 --> 00:10:08.529 kind of defaulting to. Mistra was just founded 00:10:08.529 --> 00:10:11.289 in 2023. It was a former meta and deep mind researcher, 00:10:11.570 --> 00:10:14.230 and it focuses on open -weight models and API 00:10:14.230 --> 00:10:16.330 services. Really, they're focusing on kind of 00:10:16.330 --> 00:10:18.929 the European market. They're navigating EU AI 00:10:18.929 --> 00:10:21.629 Act and data sovereignty concerns and all. that 00:10:21.629 --> 00:10:25.330 kind of stuff. Nearly 200 utility and data center 00:10:25.330 --> 00:10:28.110 companies, including NextEra Energy and Duke 00:10:28.110 --> 00:10:31.409 Energy, signed Trump's pledge to shield households 00:10:31.409 --> 00:10:35.190 from AI electricity costs. That is now covering 00:10:35.190 --> 00:10:38.370 80 % of U .S. power delivery. It was really just 00:10:38.370 --> 00:10:40.250 this kind of voluntary commitment, and the goal 00:10:40.250 --> 00:10:43.129 with it was to make AI providers not regular 00:10:43.129 --> 00:10:45.909 people pay for new power plants and grid upgrades. 00:10:46.149 --> 00:10:48.529 I think there's a bunch of different ways to 00:10:48.529 --> 00:10:51.149 think about this. A lot of data centers are getting 00:10:51.149 --> 00:10:53.970 kind of shut down. or the data center projects 00:10:53.970 --> 00:10:56.370 are getting shut down from proceeding because 00:10:56.370 --> 00:10:59.610 it's very unpopular, especially when local governments 00:10:59.610 --> 00:11:02.009 are offering big incentives for data centers 00:11:02.009 --> 00:11:04.029 to get built. And especially when they're offering 00:11:04.029 --> 00:11:06.330 energy incentives. So you know, they're subsidizing 00:11:06.330 --> 00:11:08.370 their energy because subsidized energy just means 00:11:08.370 --> 00:11:10.730 that the taxpayer or whoever's locally living 00:11:10.730 --> 00:11:12.889 there, their energy costs go up. So obviously, 00:11:12.950 --> 00:11:14.490 a lot of people are not happy with this. And 00:11:14.490 --> 00:11:17.110 I think this is trying to not just get around 00:11:17.110 --> 00:11:20.139 to that issue, but make them you know, not make 00:11:20.139 --> 00:11:22.179 these projects more popular because we want the 00:11:22.179 --> 00:11:23.980 advancements in our country. But obviously, people 00:11:23.980 --> 00:11:26.320 don't want to get stuck paying the bill for all 00:11:26.320 --> 00:11:28.799 of this. The original signers from March were 00:11:28.799 --> 00:11:30.820 Google, Meta, Microsoft, Oracle, OpenAI, and 00:11:30.820 --> 00:11:33.600 Amazon. And they've expanded that list to a bunch 00:11:33.600 --> 00:11:35.539 of major utilities and data center companies 00:11:35.539 --> 00:11:38.700 like Equinox and Digital Realty. So hopefully 00:11:38.700 --> 00:11:40.799 this makes a difference and data center providers 00:11:40.799 --> 00:11:43.299 are not going to be passing on the cost to consumers. 00:11:43.399 --> 00:11:45.220 And this actually works out. I think there's 00:11:45.220 --> 00:11:47.120 a lot of controversy on this. Some people think 00:11:47.120 --> 00:11:48.440 it'll work. Some people don't think it will. 00:11:48.889 --> 00:11:52.259 But let's hope that this is a good. bit of legislation 00:11:52.259 --> 00:11:54.440 that goes through. Guys, thank you so much for 00:11:54.440 --> 00:11:55.940 tuning into the podcast today. If you enjoyed 00:11:55.940 --> 00:11:57.980 the episode, make sure to go check out the AI 00:11:57.980 --> 00:12:01.519 Box MCP if you want to get AI image, audio, and 00:12:01.519 --> 00:12:03.519 video straight into Cloud. Or if you're using 00:12:03.519 --> 00:12:05.059 something like ChatGPT, they're about to pull 00:12:05.059 --> 00:12:07.120 the image or the video model store off of it. 00:12:07.179 --> 00:12:10.179 If you want to get Google's VO3 inside of that 00:12:10.179 --> 00:12:13.480 or any AI model inside of any AI model, we have 00:12:13.480 --> 00:12:16.100 80 plus AI models and our MCP works with all 00:12:16.100 --> 00:12:17.919 of them. So go check it out. There's a link in 00:12:17.919 --> 00:12:19.460 the description. I'll catch you guys all in the 00:12:19.460 --> 00:12:20.059 next episode.