WEBVTT 00:00:00.000 --> 00:00:02.399 Imagine for a second, you take a messy folder 00:00:02.399 --> 00:00:05.879 on your desktop, a few rough product photos, 00:00:06.179 --> 00:00:09.240 some scattered customer insights, you drop it 00:00:09.240 --> 00:00:11.560 all into an AI, and minutes later, you have a 00:00:11.560 --> 00:00:14.320 fully operational marketing engine. It's wild. 00:00:14.660 --> 00:00:17.100 Blogs are written, polished ad images are rendered, 00:00:17.320 --> 00:00:20.039 social campaigns are fully mapped out. The sheer 00:00:20.039 --> 00:00:22.839 speed of that transformation, it fundamentally 00:00:22.839 --> 00:00:25.460 changes how we think about creative work entirely. 00:00:25.879 --> 00:00:28.949 Welcome to the Deep Dive. I am very glad you 00:00:28.949 --> 00:00:31.070 are here with us today. We are exploring something 00:00:31.070 --> 00:00:33.810 that feels like a massive shift. Yeah, it really 00:00:33.810 --> 00:00:36.369 is. We're looking at Google's new generative 00:00:36.369 --> 00:00:39.090 AI marketing stack. We're mapping out a very 00:00:39.090 --> 00:00:40.770 specific journey today. We're going to follow 00:00:40.770 --> 00:00:43.170 a demo brand. Right. Let's call them Healthy 00:00:43.170 --> 00:00:46.689 Crunch. They make high -protein snack bars. Got 00:00:46.689 --> 00:00:49.469 it. We'll watch this imaginary brand move through 00:00:49.469 --> 00:00:52.070 an interconnected ecosystem of tools. Notebook 00:00:52.070 --> 00:01:00.179 LM, Gemini, Nano Banana, Gems. Omni, Flow, and 00:01:00.179 --> 00:01:02.960 finally, Pomeli. That is quite a long list of 00:01:02.960 --> 00:01:05.540 tools. Yeah. But the underlying theme here is 00:01:05.540 --> 00:01:07.760 really important for you to understand. The magic 00:01:07.760 --> 00:01:10.159 is not just that a tool can generate an image. 00:01:10.239 --> 00:01:12.659 The magic is how these different systems talk 00:01:12.659 --> 00:01:14.420 to each other. Yeah, they communicate. You never 00:01:14.420 --> 00:01:16.959 actually have to start from a blank screen. That 00:01:16.959 --> 00:01:19.540 connectivity is the real breakthrough here. We're 00:01:19.540 --> 00:01:22.579 moving away from isolated parlor tricks. This 00:01:22.579 --> 00:01:25.299 is a connected assembly line for brand creation. 00:01:25.760 --> 00:01:27.340 So let's start at the beginning of that line. 00:01:27.480 --> 00:01:30.260 Notebook, LM, and Gemini. Right. This is where 00:01:30.260 --> 00:01:32.780 we build the brand memory. If you have ever used 00:01:32.780 --> 00:01:35.420 AI for marketing, you know it usually fails. 00:01:35.439 --> 00:01:37.760 Oh, almost always at first. It fails because 00:01:37.760 --> 00:01:40.319 the tool does not actually know your brand. You 00:01:40.319 --> 00:01:43.319 have to build its brain first. That foundational 00:01:43.319 --> 00:01:46.299 knowledge is the missing piece for most people. 00:01:46.480 --> 00:01:48.560 If you just open a fresh window and ask for a 00:01:48.560 --> 00:01:51.519 blog post, then ask for an Instagram caption, 00:01:51.700 --> 00:01:54.750 then a newsletter. Right. Each output might look 00:01:54.750 --> 00:01:57.650 okay on its own, but put them side by side, they 00:01:57.650 --> 00:02:00.510 feel disjointed, they lack a cohesive soul. I'd 00:02:00.510 --> 00:02:02.810 still wrestle with prompt drift myself. Yeah. 00:02:02.870 --> 00:02:05.930 Yeah. I will start a project, and the first few 00:02:05.930 --> 00:02:09.870 outputs sound perfect. But slowly, the AI just 00:02:09.870 --> 00:02:12.469 forgets your brand voice halfway through. It's 00:02:12.469 --> 00:02:15.430 so common. It drifts away into this generic corporate 00:02:15.430 --> 00:02:19.030 tone. It is deeply frustrating. That drift happens 00:02:19.030 --> 00:02:21.889 because standard chat interfaces have a rolling 00:02:21.889 --> 00:02:24.870 memory limit. As you add new instructions, older 00:02:24.870 --> 00:02:27.969 context just gets pushed out. Notebook LM solves 00:02:27.969 --> 00:02:30.750 this structural flaw. I also. Well, instead of 00:02:30.750 --> 00:02:33.590 a blank chat, Notebook LM serves as an anchored 00:02:33.590 --> 00:02:36.569 project space. I see. For healthy crunch, you 00:02:36.569 --> 00:02:39.810 upload your actual assets. You drop in product 00:02:39.810 --> 00:02:43.900 photos. your brand guidelines, real customer 00:02:43.900 --> 00:02:47.099 reviews. It is like giving the AI an employee 00:02:47.099 --> 00:02:49.699 handbook before its first day. That analogy works 00:02:49.699 --> 00:02:51.879 perfectly on a technical level. You are essentially 00:02:51.879 --> 00:02:54.539 creating a localized private database. Right. 00:02:54.639 --> 00:02:57.500 Once that hub is established, Gemini taps into 00:02:57.500 --> 00:02:59.479 it. It reads only your approved documents. That 00:02:59.479 --> 00:03:01.780 makes sense. Then it generates blog briefs. It 00:03:01.780 --> 00:03:04.039 writes the article using Canvas mode. It even 00:03:04.039 --> 00:03:05.979 creates feature banners that match your physical 00:03:05.979 --> 00:03:07.879 packaging. And because it is part of the Google 00:03:07.879 --> 00:03:10.580 ecosystem, everything exports seamlessly. Exactly. 00:03:10.599 --> 00:03:12.620 It pushes right into Google Docs so your human 00:03:12.620 --> 00:03:15.379 team can review the drafts. The beauty is that 00:03:15.379 --> 00:03:18.430 the notebook remains intact. Tomorrow... You 00:03:18.430 --> 00:03:20.889 can open that same project space. You can ask 00:03:20.889 --> 00:03:23.090 for a retail line sheet or a lunchbox planner 00:03:23.090 --> 00:03:27.430 PDF. The AI does not need to be retrained. The 00:03:27.430 --> 00:03:30.009 foundational context is permanently locked in 00:03:30.009 --> 00:03:32.810 place. So how strict are these context window 00:03:32.810 --> 00:03:35.789 limits when we are initially loading up the notebook? 00:03:35.949 --> 00:03:38.530 The capacity is massive now. It handles thousands 00:03:38.530 --> 00:03:43.469 of pages. But curation. is still critical. Stuffing 00:03:43.469 --> 00:03:46.289 the notebook with irrelevant data actually dilutes 00:03:46.289 --> 00:03:48.629 the core identity. You only want the absolute 00:03:48.629 --> 00:03:50.889 best examples of your brand. So give it boundaries 00:03:50.889 --> 00:03:53.449 and the AI stays completely on brand. Perfectly 00:03:53.449 --> 00:03:56.270 said. So we have the text foundation locked down. 00:03:56.370 --> 00:03:58.810 Right. But marketing requires visual impact. 00:03:59.110 --> 00:04:01.830 We need to transition from text rules to a visual 00:04:01.830 --> 00:04:04.969 sandbox. This brings us to Nano Banana. Nano 00:04:04.969 --> 00:04:07.550 Banana. It is an interesting name for a powerful 00:04:07.550 --> 00:04:10.530 tool. It's catchy. It runs directly through Gemini's 00:04:10.530 --> 00:04:12.610 image mode. And what stands out to me is that 00:04:12.610 --> 00:04:15.310 you do not need a perfect text prompt. The reliance 00:04:15.310 --> 00:04:18.029 on complex prompting is feeding fast. With Nano 00:04:18.029 --> 00:04:20.610 Banana, you just upload a raw photo of the healthy 00:04:20.610 --> 00:04:23.170 crunch bar. You select a visual style from a 00:04:23.170 --> 00:04:26.089 menu. Like a preset. Exactly. Maybe a clean studio 00:04:26.089 --> 00:04:29.029 style. Within moments, it generates professional 00:04:29.029 --> 00:04:32.250 ad shots. I do want to push back gently on this 00:04:32.250 --> 00:04:34.709 idea, though. Sure. Is this technology actually 00:04:34.709 --> 00:04:38.829 meant to replace human designers entirely? Or 00:04:38.829 --> 00:04:40.790 is it just designed to get the team to a better 00:04:40.790 --> 00:04:43.310 starting line? It is absolutely about getting 00:04:43.310 --> 00:04:46.610 to the starting line faster. A human still dictates 00:04:46.610 --> 00:04:50.470 the taste and the strategy. But NanoBanana gives 00:04:50.470 --> 00:04:54.050 you incredible granular control. It has a masking 00:04:54.050 --> 00:04:56.170 and sketch feature. How does that work? Let's 00:04:56.170 --> 00:04:58.870 say the generated banner looks stunning. But 00:04:58.870 --> 00:05:00.829 there's a distracting shadow in the corner. Okay. 00:05:00.970 --> 00:05:03.949 You just circle it with your mouse. The AI repaints 00:05:03.949 --> 00:05:06.389 that specific area without changing the rest 00:05:06.389 --> 00:05:08.110 of the image. You can also use it for strategic 00:05:08.110 --> 00:05:10.410 planning, right? Yes. You can upload a photo 00:05:10.410 --> 00:05:13.250 and ask the system for three A -B testing prompt 00:05:13.250 --> 00:05:15.769 ideas. You do this before you ever generate the 00:05:15.769 --> 00:05:18.170 final image. That particular workflow saves hours 00:05:18.170 --> 00:05:20.750 of frustration. You get the AI to brainstorm 00:05:20.750 --> 00:05:23.709 the angles first. Once you pick the best concept, 00:05:23.949 --> 00:05:27.410 you generate the image. Wow. Mantle Banana also 00:05:27.410 --> 00:05:29.870 solves a massive headache for social media managers. 00:05:30.110 --> 00:05:33.550 It handles spatial resizing brilliantly. Explain 00:05:33.550 --> 00:05:35.769 how that resizing actually works under the hood. 00:05:36.370 --> 00:05:38.430 Historically, if you took a square Instagram 00:05:38.430 --> 00:05:41.910 ad and made it vertical for TikTok. You just 00:05:41.910 --> 00:05:44.110 stretch the pixels. Exactly. And it looked terrible. 00:05:44.370 --> 00:05:46.750 Nano Banana uses a technique called outpainting. 00:05:46.829 --> 00:05:50.310 Okay. It analyzes the existing image and hallucinates 00:05:50.310 --> 00:05:52.189 the missing top and bottom sections. That is 00:05:52.189 --> 00:05:54.730 fascinating. It literally paints new studio lighting 00:05:54.730 --> 00:05:56.569 and background elements to fill the vertical 00:05:56.569 --> 00:05:59.850 space. The snack bar stays perfectly proportioned 00:05:59.850 --> 00:06:02.389 in the center. How do we avoid wasting our API 00:06:02.389 --> 00:06:05.410 usage limits on all these random visual experiments? 00:06:06.009 --> 00:06:08.110 By flipping the traditional workflow entirely. 00:06:08.779 --> 00:06:11.100 You generate the text -based conceptual prompts 00:06:11.100 --> 00:06:14.500 first. You review those ideas, select the strongest 00:06:14.500 --> 00:06:17.560 one, and only render that final choice. Draft 00:06:17.560 --> 00:06:20.459 first, refine, then polish only the best creative 00:06:20.459 --> 00:06:22.839 direction. Exactly. That disciplined approach 00:06:22.839 --> 00:06:25.120 prevents digital burnout. But even with great 00:06:25.120 --> 00:06:28.079 tools, prompting the same style over and over 00:06:28.079 --> 00:06:31.100 becomes tedious. We need to turn this manual 00:06:31.100 --> 00:06:34.129 workflow into a repeatable system. Which brings 00:06:34.129 --> 00:06:37.610 us to the concept of gems. Yes. Gems are fascinating. 00:06:37.889 --> 00:06:40.750 They exist to lock in the workflow. They solve 00:06:40.750 --> 00:06:43.550 the human bottleneck of repetitive prompting. 00:06:43.610 --> 00:06:45.970 They serve as the architectural bridge between 00:06:45.970 --> 00:06:49.550 individual skill and team capability. A gem is 00:06:49.550 --> 00:06:52.529 fundamentally a custom AI agent. For anyone listening 00:06:52.529 --> 00:06:55.370 who might be confused by that term, an AI agent 00:06:55.370 --> 00:06:58.490 is a custom AI setup that follows specific rules 00:06:58.490 --> 00:07:01.430 for repeat tasks. That structural definition 00:07:01.430 --> 00:07:03.579 is vital. Yeah. Let's look at... how a marketing 00:07:03.579 --> 00:07:06.019 team actually operates today. Okay. You might 00:07:06.019 --> 00:07:08.459 have one senior art director who knows exactly 00:07:08.459 --> 00:07:10.639 how to coax the perfect lighting out of an AI. 00:07:10.899 --> 00:07:13.240 Right. But the rest of the team struggles. With 00:07:13.240 --> 00:07:16.000 gems, that art director can build a studio ad 00:07:16.000 --> 00:07:18.759 shot gem. They preload it with the exact packaging 00:07:18.759 --> 00:07:20.779 dimensions. Yep. They add the specific brand 00:07:20.779 --> 00:07:23.540 hex codes. They set nano banana as the default 00:07:23.540 --> 00:07:26.180 rendering engine. Then they save it. Now, that 00:07:26.180 --> 00:07:28.319 complex chain of commands is hidden behind a 00:07:28.319 --> 00:07:31.220 single button. That is brilliant. The next time 00:07:31.220 --> 00:07:33.879 a junior copywriter needs an image, they just 00:07:33.879 --> 00:07:36.779 upload a raw photo into that specific gem. It 00:07:36.779 --> 00:07:39.860 is like saving a custom preset on a synthesizer. 00:07:39.980 --> 00:07:42.639 You lock in the magic so anyone can play it. 00:07:42.699 --> 00:07:45.680 That musical analogy captures the workflow perfectly. 00:07:45.899 --> 00:07:49.519 It guarantees a consistent premium output regardless 00:07:49.519 --> 00:07:52.300 of who is driving the machine. Right. It completely 00:07:52.300 --> 00:07:55.500 eliminates the uneven quality that usually plagues 00:07:55.500 --> 00:07:58.750 AI -generated marketing content. You no longer 00:07:58.750 --> 00:08:01.790 rely on one person's prompting skills. How does 00:08:01.790 --> 00:08:03.970 locking in these models fundamentally change 00:08:03.970 --> 00:08:07.170 the dynamics of a modern marketing team? It democratizes 00:08:07.170 --> 00:08:10.050 high -level execution across the board. The junior 00:08:10.050 --> 00:08:12.829 staff can now generate visual assets with the 00:08:12.829 --> 00:08:15.370 exact same technical fidelity as the senior leadership. 00:08:15.769 --> 00:08:18.889 Gems turn perfect individual prompts into a shared, 00:08:18.910 --> 00:08:22.220 foolproof team system. Exactly. We have established 00:08:22.220 --> 00:08:25.220 a robust system for static imagery, but static 00:08:25.220 --> 00:08:27.379 assets are rarely enough to carry a modern campaign. 00:08:28.019 --> 00:08:29.819 Audiences want to see the wrapper tear open. 00:08:29.879 --> 00:08:31.879 They want to see the chocolate drizzle. We need 00:08:31.879 --> 00:08:33.740 to make these healthy crunch snacks move. That 00:08:33.740 --> 00:08:35.799 desire for motion brings us to the Omni model. 00:08:36.139 --> 00:08:38.759 Omni is the dedicated video generation model 00:08:38.759 --> 00:08:41.980 living inside the Gemini ecosystem. Okay. It's 00:08:41.980 --> 00:08:44.360 designed to create short, highly polished clips. 00:08:45.120 --> 00:08:47.860 Currently, these clips run up to 10 seconds long. 00:08:48.220 --> 00:08:50.919 You feed it the simple assets we already created, 00:08:51.059 --> 00:08:54.139 a product photo from Nano Banana, a character 00:08:54.139 --> 00:08:57.039 image we developed in our gems. Then it generates 00:08:57.039 --> 00:08:59.639 motion. You can generate dynamic recipe reels 00:08:59.639 --> 00:09:02.620 or quick social ads. The spatial control here 00:09:02.620 --> 00:09:05.419 is what makes it usable for real brands. Oh, 00:09:05.419 --> 00:09:08.179 so? You can keep your digital model holding the 00:09:08.179 --> 00:09:12.000 exact healthy crunch snack bar. But you can change 00:09:12.000 --> 00:09:14.159 her environment completely. Oh, wow. You can 00:09:14.159 --> 00:09:16.399 move her from a bright morning kitchen into a 00:09:16.399 --> 00:09:18.899 moody grocery store aisle. I have to question 00:09:18.899 --> 00:09:21.120 the core limitation here, though. Paint 10 seconds? 00:09:21.360 --> 00:09:23.779 Feels awfully short. Can you really tell a meaningful 00:09:23.779 --> 00:09:26.440 brand story in just 10 seconds? It is a severe 00:09:26.440 --> 00:09:29.480 constraint. But consumer attention spans on social 00:09:29.480 --> 00:09:32.039 media often demand immediate impact anyway. That 00:09:32.039 --> 00:09:35.139 is fair. To maximize those 10 seconds, you need 00:09:35.139 --> 00:09:38.039 extreme precision. You should always use a dedicated 00:09:38.039 --> 00:09:40.879 storyboard gem first. This helps you plan your 00:09:40.879 --> 00:09:43.740 specific camera angles and lighting cues before 00:09:43.740 --> 00:09:46.240 you ask Omni to render anything. The other technique 00:09:46.240 --> 00:09:48.899 that stood out to me was video referencing. Video 00:09:48.899 --> 00:09:50.879 referencing solves one of the hardest problems 00:09:50.879 --> 00:09:54.659 in AI video. Which is? Language is inherently 00:09:54.659 --> 00:09:57.899 terrible at describing motion. If you type, pan 00:09:57.899 --> 00:10:01.460 the camera slowly while she eats, the AI interprets 00:10:01.460 --> 00:10:04.419 slowly in unpredictable ways. But if you upload 00:10:04.419 --> 00:10:07.149 a 10 -second reference clip... A video you shot 00:10:07.149 --> 00:10:10.330 on your phone. Exactly. Omni analyzes the actual 00:10:10.330 --> 00:10:13.389 pixels. It copies the exact pacing, the specific 00:10:13.389 --> 00:10:15.669 camera drift, and the editing rhythm. It strips 00:10:15.669 --> 00:10:17.889 the visual data and applies your brand assets 00:10:17.889 --> 00:10:20.529 over that mathematical framework. Why is that 00:10:20.529 --> 00:10:23.450 video referencing step so vital for the Omni 00:10:23.450 --> 00:10:26.129 model to function well? Because without a reference... 00:10:26.399 --> 00:10:29.100 The AI has to invent physics and timing from 00:10:29.100 --> 00:10:32.279 scratch. The reference video provides an undeniable 00:10:32.279 --> 00:10:34.559 mathematical template for the movement. Reference 00:10:34.559 --> 00:10:37.000 clips guide the pacing so the AI isn't just guessing. 00:10:37.200 --> 00:10:39.360 Nailed it. Now we have our 10 -second clips. 00:10:39.519 --> 00:10:42.039 Right. But your earlier pushback remains valid. 00:10:42.299 --> 00:10:45.639 A single short clip is not a full campaign. We 00:10:45.639 --> 00:10:47.559 need to sequence these moments together. This 00:10:47.559 --> 00:10:50.679 leads us directly to Google Flow. Flow feels 00:10:50.679 --> 00:10:53.480 like the director's chair. It is where all these 00:10:53.480 --> 00:10:56.799 fragmented pieces finally assemble. Flow is essentially 00:10:56.799 --> 00:10:59.759 a non -linear video workspace built entirely 00:10:59.759 --> 00:11:02.679 around the Omni model. Okay. It gives you a traditional 00:11:02.679 --> 00:11:04.980 editing timeline. It allows you to stitch those 00:11:04.980 --> 00:11:07.980 short clips into a cohesive narrative. The most 00:11:07.980 --> 00:11:10.340 impressive feature here is the reusable character 00:11:10.340 --> 00:11:12.980 system. Yes. You can create a digital personality. 00:11:13.320 --> 00:11:16.019 Let's call her the healthy crunch host. You meticulously 00:11:16.019 --> 00:11:18.779 design her look. You lock in her wardrobe choices. 00:11:19.100 --> 00:11:22.340 You define her personality quirks. You even synthesize 00:11:22.340 --> 00:11:24.960 a custom voice for her. Wow. Flow saves this 00:11:24.960 --> 00:11:27.860 entire profile as a distinct digital asset. And 00:11:27.860 --> 00:11:30.799 then you simply tag her in future scenes. maintain 00:11:30.799 --> 00:11:33.519 total brand consistency across multiple videos. 00:11:33.720 --> 00:11:36.320 Exactly. You connect a host intro to a macro 00:11:36.320 --> 00:11:39.039 product shot, then you link that to a lifestyle 00:11:39.039 --> 00:11:41.679 lunchbox scene. Flow handles the transition logic 00:11:41.679 --> 00:11:44.860 between those nodes. It also features a powerful 00:11:44.860 --> 00:11:47.440 agent mode. What does that do? You can ask the 00:11:47.440 --> 00:11:49.539 flow agent to generate three distinct variations 00:11:49.539 --> 00:11:52.000 of the entire sequence. It alters the camera 00:11:52.000 --> 00:11:55.480 move slightly for easy A -B testing. Whoa. Two 00:11:55.480 --> 00:11:58.659 sec silence. Imagine generating endless video 00:11:58.659 --> 00:12:01.720 variations with the exact same digital host perfectly 00:12:01.720 --> 00:12:04.779 on brand every time. The scale of that capability 00:12:04.779 --> 00:12:07.159 is staggering. It feels like science fiction, 00:12:07.340 --> 00:12:09.879 but it is driven by deeply structured logic. 00:12:10.059 --> 00:12:12.600 You define the timeline. You anchor the character 00:12:12.600 --> 00:12:15.519 parameters. You inject the product data. Flow 00:12:15.519 --> 00:12:18.379 handles the complex rendering math. How exactly 00:12:18.379 --> 00:12:20.600 does Flow handle a complex storyboard compared 00:12:20.600 --> 00:12:23.639 to just standard text prompting? Standard prompting 00:12:23.639 --> 00:12:26.080 forces the AI to remember everything simultaneously, 00:12:26.320 --> 00:12:28.779 which it struggles with. Flow creates a visual 00:12:28.779 --> 00:12:31.700 timeline, mapping scenes chronologically, and 00:12:31.700 --> 00:12:33.759 linking characters precisely across the entire 00:12:33.759 --> 00:12:36.200 sequence. Flow gives you the timeline and character 00:12:36.200 --> 00:12:39.240 memory that Gemini lacks. Exactly right. So we 00:12:39.240 --> 00:12:41.620 have reached the final stage. We have our perfectly 00:12:41.620 --> 00:12:44.200 toned blogs. We have our static imagery and our 00:12:44.200 --> 00:12:47.039 timeline edited videos. Yep. All the assets are 00:12:47.039 --> 00:12:49.600 ready. But manually logging into five different 00:12:49.600 --> 00:12:52.639 platforms to push these assets live is exhausting. 00:12:52.980 --> 00:12:55.600 The worst part of the job. We need a distribution 00:12:55.600 --> 00:12:58.899 engine to finish the job. That brings us to Google 00:12:58.899 --> 00:13:02.220 Pameli, the final mile of the journey. Pameli 00:13:02.220 --> 00:13:04.399 is designed to automate the heavy lifting of 00:13:04.399 --> 00:13:06.820 distribution. Okay. The workflow starts simply. 00:13:06.980 --> 00:13:10.440 You just enter your existing website URL into 00:13:10.440 --> 00:13:13.019 the system. And from that URL, Pameli builds 00:13:13.019 --> 00:13:16.889 what it calls a business DNA. actively scrapes 00:13:16.889 --> 00:13:19.049 your site architecture it extracts your exact 00:13:19.049 --> 00:13:22.549 hex codes for brand colors that is incredibly 00:13:22.549 --> 00:13:25.289 useful it pulls your typography it reads your 00:13:25.289 --> 00:13:27.870 product catalog and absorbs your existing messaging 00:13:27.870 --> 00:13:30.769 style it builds a comprehensive profile of who 00:13:30.769 --> 00:13:33.710 you are online once that dna is established you 00:13:33.710 --> 00:13:36.370 can access pre -built campaign modules right 00:13:36.370 --> 00:13:38.710 let's say you want to run a promotion for world 00:13:38.710 --> 00:13:41.519 chocolate day you click that module And Pameli 00:13:41.519 --> 00:13:43.519 instantly generates the surrounding architecture. 00:13:43.779 --> 00:13:45.960 Just like that. Just like that. It writes the 00:13:45.960 --> 00:13:48.600 email headers. It crafts the social descriptions. 00:13:48.899 --> 00:13:51.559 It even designs the call to action buttons using 00:13:51.559 --> 00:13:53.899 your exact brand colors. You can also just use 00:13:53.899 --> 00:13:56.799 the chat interface. You talk to Pameli, request 00:13:56.799 --> 00:13:59.919 a specific campaign brief, and it auto -generates 00:13:59.919 --> 00:14:02.639 the matching social assets. It can run rapid 00:14:02.639 --> 00:14:05.460 AI photo shoots using your existing catalog items. 00:14:05.639 --> 00:14:08.220 Wow. It takes it a step further by building simple 00:14:08.220 --> 00:14:10.960 test web pages. You can validate new campaign 00:14:10.960 --> 00:14:13.639 ideas or test different messaging before committing 00:14:13.639 --> 00:14:16.379 your main engineering team to a full site update. 00:14:16.809 --> 00:14:19.169 I am curious about the reality of this automation, 00:14:19.409 --> 00:14:22.190 though. Sure. How much human review is genuinely 00:14:22.190 --> 00:14:24.990 still required before hitting the publish button 00:14:24.990 --> 00:14:27.690 on a Pomele -generated test page? A significant 00:14:27.690 --> 00:14:30.820 amount of review is still necessary. Pameli acts 00:14:30.820 --> 00:14:33.559 like an incredibly fast construction crew. It 00:14:33.559 --> 00:14:36.919 builds the house rapidly. But a human absolutely 00:14:36.919 --> 00:14:39.320 needs to walk through the rooms, check the wiring, 00:14:39.480 --> 00:14:41.840 and inspect the paint before inviting actual 00:14:41.840 --> 00:14:44.279 customers inside. Why does Pameli specifically 00:14:44.279 --> 00:14:47.039 need the website URL before it does anything 00:14:47.039 --> 00:14:50.039 else? To anchor its generations in verifiable 00:14:50.039 --> 00:14:53.399 truth. Scraping the live site allows it to pull 00:14:53.399 --> 00:14:55.899 your actual live parameters rather than relying 00:14:55.899 --> 00:14:58.500 on some outdated brand document. It scrapes your 00:14:58.500 --> 00:15:01.179 business DNA so you avoid uploading files manually. 00:15:01.500 --> 00:15:03.720 That's the core of it. Yeah. And that automation 00:15:03.720 --> 00:15:07.019 wraps up the entire ecosystem. If we step back 00:15:07.019 --> 00:15:08.620 and look at the grand arc we just discussed. 00:15:08.779 --> 00:15:11.789 Yeah. Google's generative AI strategy is clearly 00:15:11.789 --> 00:15:14.909 no longer about isolated tricks. It is a fully 00:15:14.909 --> 00:15:17.809 integrated assembly line. It is a complete end 00:15:17.809 --> 00:15:20.789 -to -end system. We watched Notebook LM set the 00:15:20.789 --> 00:15:23.470 foundational memory. We saw Nano Banana operate 00:15:23.470 --> 00:15:26.970 as the visual sandbox. We used GEMS to standardize 00:15:26.970 --> 00:15:29.549 those creative prompts across an entire team. 00:15:30.090 --> 00:15:32.850 Omni and Flow handle the complex physics of motion 00:15:32.850 --> 00:15:35.169 and timeline editing. And Pamele pushed the final 00:15:35.169 --> 00:15:37.649 assets out to the real world. Exactly. The mechanical 00:15:37.649 --> 00:15:39.889 heavy lifting is essentially solved. Marketers 00:15:39.889 --> 00:15:41.889 are no longer starting from scratch. They are 00:15:41.889 --> 00:15:44.889 now editors and directors. It is a profound shift 00:15:44.889 --> 00:15:48.490 in how we approach creative output. Beat. We 00:15:48.490 --> 00:15:50.289 have fundamentally lowered the cost of generating 00:15:50.289 --> 00:15:53.120 high quality assets to near zero. Absolutely 00:15:53.120 --> 00:15:56.220 zero friction. If AI standardizes this perfect 00:15:56.220 --> 00:15:59.639 workflow and guarantees premium visual output 00:15:59.639 --> 00:16:02.919 for literally every company on Earth. Yeah. Does 00:16:02.919 --> 00:16:05.580 brand survival now rely entirely on having a 00:16:05.580 --> 00:16:08.659 fundamentally unique human perspective or a radically 00:16:08.659 --> 00:16:11.240 better product in the real world? When everyone 00:16:11.240 --> 00:16:13.600 can look perfect, perfection becomes the baseline. 00:16:13.840 --> 00:16:16.139 That's the ultimate question brands have to answer 00:16:16.139 --> 00:16:18.350 now. Thank you for joining us on this deep dive. 00:16:18.610 --> 00:16:20.850 It has been a fascinating journey into the mechanics 00:16:20.850 --> 00:16:23.029 of modern creation. We deeply appreciate your 00:16:23.029 --> 00:16:23.730 time. Take care.