WEBVTT 00:00:00.000 --> 00:00:02.660 We all have a folder of messy research, or maybe, 00:00:02.700 --> 00:00:05.280 you know, a hundred open browser tats. Oh, absolutely. 00:00:05.580 --> 00:00:08.519 It is a total nightmare. Right. What if a tool 00:00:08.519 --> 00:00:11.439 didn't just store that mess? What if it actually 00:00:11.439 --> 00:00:13.800 reasoned through it for you? That is the ultimate 00:00:13.800 --> 00:00:16.120 dream for anyone learning. You just want something 00:00:16.120 --> 00:00:19.179 to automatically connect the dots. Welcome to 00:00:19.179 --> 00:00:22.239 our latest deep dive. We are unpacking a truly 00:00:22.239 --> 00:00:25.219 massive update today. We are looking at the new 00:00:25.219 --> 00:00:29.089 agentic AI. It's inside Google Notebook LM. This 00:00:29.089 --> 00:00:31.410 fundamentally shifts how we process scattered 00:00:31.410 --> 00:00:34.369 information. It literally turns a static tool 00:00:34.369 --> 00:00:37.109 into an active partner. We'll explore its brand 00:00:37.109 --> 00:00:39.450 new artificial brain today. We're going to look 00:00:39.450 --> 00:00:41.929 at its secure cloud computer. We'll also see 00:00:41.929 --> 00:00:44.170 how it actively conducts web research. Yeah, 00:00:44.210 --> 00:00:46.090 it operates completely differently from those 00:00:46.090 --> 00:00:47.729 older versions. And we'll explore its ability 00:00:47.729 --> 00:00:50.149 to output files. It can build visual charts and 00:00:50.149 --> 00:00:52.630 complete slide decks. Which is just wild. It 00:00:52.630 --> 00:00:54.710 really is. Then we'll dig into the hard performance 00:00:54.710 --> 00:00:58.179 numbers. These stats prove why this changes everything 00:00:58.179 --> 00:01:00.960 for you. It reshapes workflows for students, 00:01:01.140 --> 00:01:04.239 coders, and businesses. The jump in pure factual 00:01:04.239 --> 00:01:07.040 accuracy is honestly incredible. To understand 00:01:07.040 --> 00:01:10.500 what this tool can actually do now, we first 00:01:10.500 --> 00:01:12.480 have to look under the hood. We really need to 00:01:12.480 --> 00:01:14.879 understand what it fundamentally is. Right. It 00:01:14.879 --> 00:01:17.900 has totally graduated from a simple PDF summarizer. 00:01:18.060 --> 00:01:20.760 Exactly. The old version was basically a document 00:01:20.760 --> 00:01:23.459 chatbot. You would upload a text file, and it 00:01:23.459 --> 00:01:25.950 waited. It just sat there until you typed a prompt. 00:01:26.129 --> 00:01:28.329 Yeah, it was a totally passive experience. Now, 00:01:28.390 --> 00:01:32.269 Notebook LM is powered by Gemini 3 .5 Flash. 00:01:32.530 --> 00:01:35.049 Yeah. It also runs on something called anti -gravity. 00:01:35.260 --> 00:01:37.840 Antigrammody is under the hood tech making the 00:01:37.840 --> 00:01:40.519 AI process faster. It gives the system a lot 00:01:40.519 --> 00:01:43.480 more cognitive runway. Exactly. The chat experience 00:01:43.480 --> 00:01:46.640 feels much more thoughtful now. It's incredibly 00:01:46.640 --> 00:01:49.260 deliberate. It's extremely useful for complex, 00:01:49.379 --> 00:01:51.640 layered questions. It doesn't just hand you a 00:01:51.640 --> 00:01:53.739 quick summary anymore. No, it reads across your 00:01:53.739 --> 00:01:56.319 longest, most complex documents. It actively 00:01:56.319 --> 00:01:58.340 searches for the most important hidden points. 00:01:58.599 --> 00:02:01.359 It can take a massive, complicated research question 00:02:01.359 --> 00:02:03.579 and it breaks it down into small, manageable 00:02:03.579 --> 00:02:06.359 steps. Right, and that is a massive philosophical 00:02:06.359 --> 00:02:10.360 shift. It really is. A normal chatbot needs you 00:02:10.360 --> 00:02:12.280 to hold its hand. You literally have to guide 00:02:12.280 --> 00:02:15.460 every single logical step. This new version acts 00:02:15.460 --> 00:02:18.560 more like a real assistant. It anticipates what 00:02:18.560 --> 00:02:20.639 you actually need to know next. It helps you 00:02:20.639 --> 00:02:23.099 figure out the next logical move. And that brings 00:02:23.099 --> 00:02:27.460 us to the absolute biggest shift. Every single 00:02:27.460 --> 00:02:30.939 notebook now gets its own secure cloud computer. 00:02:31.719 --> 00:02:34.699 Beat. I want to pause and think about that. It's 00:02:34.699 --> 00:02:37.120 the technical feature making this update so huge. 00:02:37.300 --> 00:02:40.139 It is essentially spinning up a temporary virtual 00:02:40.139 --> 00:02:43.159 environment. A workspace loaded with over 100 00:02:43.159 --> 00:02:46.379 curated software skills. Yeah. This means Notebook 00:02:46.379 --> 00:02:49.120 LM can do much more than read. If you give it 00:02:49.120 --> 00:02:51.539 raw data, it writes Python. It runs that code 00:02:51.539 --> 00:02:53.659 inside a locked digital laboratory. Exactly. 00:02:53.780 --> 00:02:55.419 It does all of this right inside your notebook. 00:02:55.520 --> 00:02:57.979 It moves from a passive state to an active state. 00:02:58.349 --> 00:03:00.870 It actually executes tasks instead of just predicting 00:03:00.870 --> 00:03:04.189 text. You hand Notebook LM a highly complex multi 00:03:04.189 --> 00:03:06.430 -step task. Then you just let it move from raw 00:03:06.430 --> 00:03:08.189 sources. It goes straight to finished results 00:03:08.189 --> 00:03:11.610 for you. It handles the messy, tedious execution 00:03:11.610 --> 00:03:15.270 for you. Now this update is rolling out in structured 00:03:15.270 --> 00:03:19.289 stages. It's going first to Google AI Ultra subscribers. 00:03:19.889 --> 00:03:21.949 They're making sure the infrastructure can handle 00:03:21.949 --> 00:03:25.150 the load. But it's expected to expand to other 00:03:25.150 --> 00:03:27.780 plans soon. I think about the upgrade kind of 00:03:27.780 --> 00:03:30.800 like this. The old version was like a passive 00:03:30.800 --> 00:03:32.900 filing cabinet. You put papers in and pulled 00:03:32.900 --> 00:03:35.560 papers out. This new version is a completely 00:03:35.560 --> 00:03:38.400 different paradigm. It's like hiring a dedicated 00:03:38.400 --> 00:03:41.000 research assistant, one who arrives with their 00:03:41.000 --> 00:03:43.639 own fully loaded laptop. A laptop packed with 00:03:43.639 --> 00:03:45.439 tools they already know. You just need to point 00:03:45.439 --> 00:03:47.460 them at an interesting problem. Does giving it 00:03:47.460 --> 00:03:49.539 an independent cloud computer mean something 00:03:49.539 --> 00:03:52.979 risky? Like, can it run code completely unsupervised? 00:03:53.080 --> 00:03:55.180 No, it operates inside a strictly locked down 00:03:55.180 --> 00:03:58.099 sandbox. It can only use its skills on your specific 00:03:58.099 --> 00:04:00.400 research. It can't browse the broader web and 00:04:00.400 --> 00:04:03.099 trigger external actions. So it remains an active 00:04:03.099 --> 00:04:06.300 assistant, but securely boxed in. Exactly. That 00:04:06.300 --> 00:04:08.919 design choice keeps the entire system highly 00:04:08.919 --> 00:04:11.539 predictable. So it has this secure sandbox to 00:04:11.539 --> 00:04:14.870 play in. The next logical question is how it 00:04:14.870 --> 00:04:18.009 fetches materials. How does it tackle a really 00:04:18.009 --> 00:04:20.310 messy project? That brings us to the headline 00:04:20.310 --> 00:04:22.990 feature here. It's a system they call agentic 00:04:22.990 --> 00:04:25.009 web research. This is where the workflow gets 00:04:25.009 --> 00:04:27.870 incredibly interesting. You no longer need perfect 00:04:27.870 --> 00:04:31.110 sources to get started. Right. Before, you manually 00:04:31.110 --> 00:04:33.089 collected all the links yourself. You downloaded 00:04:33.089 --> 00:04:35.490 PDFs and cleaned up your own notes. You had to 00:04:35.490 --> 00:04:37.649 organize everything before asking any meaningful 00:04:37.649 --> 00:04:40.810 questions. You basically carry the entire burden 00:04:40.810 --> 00:04:44.089 of the legwork. Only then could Notebook LM actually 00:04:44.089 --> 00:04:46.529 be useful to you. But now the starting process 00:04:46.529 --> 00:04:49.310 is drastically more intuitive. It embraces the 00:04:49.310 --> 00:04:52.449 natural chaos of early academic research. You 00:04:52.449 --> 00:04:54.870 could begin with a very loose, half -baked idea. 00:04:55.129 --> 00:04:57.009 Yeah, you might just type a really vague prompt 00:04:57.009 --> 00:04:59.970 like, I want to understand how AI search changes 00:04:59.970 --> 00:05:03.550 SEO. From there, Notebook LM actively goes out 00:05:03.550 --> 00:05:06.209 and finds sources. It acts like a digital scout. 00:05:06.569 --> 00:05:08.470 looking for relevant information. It suggests 00:05:08.470 --> 00:05:10.930 specific articles you should add to your project. 00:05:11.170 --> 00:05:13.509 You know, I still wrestle with props drift myself. 00:05:14.610 --> 00:05:17.509 Beat. I often get lost in the messy middle of 00:05:17.509 --> 00:05:20.610 research. I end up with 50 open tabs and total 00:05:20.610 --> 00:05:23.430 confusion. Oh, we all fall into that trap eventually. 00:05:23.649 --> 00:05:26.350 Most research doesn't begin with a clean, perfectly 00:05:26.350 --> 00:05:28.829 organized folder. Right. It usually starts with 00:05:28.829 --> 00:05:31.970 intense confusion and scattered ideas. But Notebook 00:05:31.970 --> 00:05:35.370 LM can now search the live web for you. It finds 00:05:35.370 --> 00:05:37.709 high quality sources and gathers them all together. 00:05:37.949 --> 00:05:40.089 It naturally connects those new sources to your 00:05:40.089 --> 00:05:43.550 existing files. But this process requires a very 00:05:43.550 --> 00:05:46.120 strong safety net. We really need to talk about 00:05:46.120 --> 00:05:48.439 the control layer. Yes, the control layer is 00:05:48.439 --> 00:05:51.379 a vital part here. It prevents the AI from just 00:05:51.379 --> 00:05:53.699 flooding your notebook. It won't dump a hundred 00:05:53.699 --> 00:05:56.540 random links on your desk. No, it always surfaces 00:05:56.540 --> 00:06:00.079 its suggested sources to you first. Then it explicitly 00:06:00.079 --> 00:06:02.759 asks for your permission before adding anything. 00:06:03.019 --> 00:06:05.100 It only integrates a source after you actually 00:06:05.100 --> 00:06:07.699 approve it. That makes the workflow feel incredibly 00:06:07.699 --> 00:06:10.660 safe and manageable. You still decide what ultimately 00:06:10.660 --> 00:06:13.069 belongs in your notebook. But you don't have 00:06:13.069 --> 00:06:15.430 to painstakingly hunt for every single source. 00:06:15.610 --> 00:06:19.029 It also has the ability to reason across different 00:06:19.029 --> 00:06:21.889 topics. This is huge. It synthesizes information 00:06:21.889 --> 00:06:24.930 from highly conflicting viewpoints. A standard 00:06:24.930 --> 00:06:27.430 document chatbot usually answers from just one 00:06:27.430 --> 00:06:31.250 file. But real research means comparing competing 00:06:31.250 --> 00:06:34.250 ideas across many sources. Exactly. You can ask 00:06:34.250 --> 00:06:36.709 it to compare three different reports. You can 00:06:36.709 --> 00:06:39.300 ask it to find the biggest disagreements. It 00:06:39.300 --> 00:06:41.339 can even look for primary sources in other languages. 00:06:41.519 --> 00:06:43.759 Yeah, it translates the relevant parts and adds 00:06:43.759 --> 00:06:45.699 them in. It helps you understand a topic from 00:06:45.699 --> 00:06:47.439 multiple different angles. It doesn't just give 00:06:47.439 --> 00:06:49.920 you one neat, oversimplified summary. No, it 00:06:49.920 --> 00:06:52.860 thrives in the nuance and contradiction of real 00:06:52.860 --> 00:06:57.180 data. To sex silence. How does this control layer 00:06:57.180 --> 00:07:00.420 actually prevent the AI from pulling in low -quality 00:07:00.420 --> 00:07:02.939 internet junk? It uses a strict permission -first 00:07:02.939 --> 00:07:05.399 approval process for every edition. It presents 00:07:05.399 --> 00:07:07.620 a summary of the source before adding it. If 00:07:07.620 --> 00:07:09.759 the link looks like unreliable clickbait, you 00:07:09.759 --> 00:07:12.980 just click reject. It suggests the sources, but 00:07:12.980 --> 00:07:16.079 you hold the library keys. Mid -roll sponsor 00:07:16.079 --> 00:07:19.579 read. We've been exploring the new agentic web 00:07:19.579 --> 00:07:22.319 research features. Gathering and reasoning through 00:07:22.319 --> 00:07:24.540 the research is certainly great, but research 00:07:24.540 --> 00:07:26.839 is totally useless if it stays in a chat box. 00:07:27.100 --> 00:07:29.579 You absolutely need tangible deliverables to 00:07:29.579 --> 00:07:31.819 share with your team. A brilliant insight doesn't 00:07:31.819 --> 00:07:34.240 matter if nobody else sees it. Right. Notebook 00:07:34.240 --> 00:07:37.959 LM now generates real, fully formatted, downloadable 00:07:37.959 --> 00:07:40.040 files. It is not just returning text answers 00:07:40.040 --> 00:07:42.459 inside the chat window. This changes the daily 00:07:42.459 --> 00:07:45.259 professional workflow in a massive way. Real 00:07:45.259 --> 00:07:47.899 work usually doesn't end with a simple paragraph 00:07:47.899 --> 00:07:50.199 summary. Exactly. You eventually need a formal 00:07:50.199 --> 00:07:53.120 report or a slide deck. Well, you can now instantly 00:07:53.120 --> 00:07:56.240 download PDFs and Word docs. It creates markdown 00:07:56.240 --> 00:07:59.800 files, Excel sheets, and CSV files. It even generates 00:07:59.800 --> 00:08:03.019 full PowerPoint presentations, JSON data, and 00:08:03.019 --> 00:08:04.860 images. You don't have to copy and paste text 00:08:04.860 --> 00:08:07.579 manually anymore. You just ask Notebook LM to 00:08:07.579 --> 00:08:10.100 format the exact file type. It also handles raw 00:08:10.100 --> 00:08:12.360 numbers and heavy data sets effortlessly. This 00:08:12.360 --> 00:08:14.339 part is especially helpful if you work with analytics. 00:08:14.920 --> 00:08:17.660 Analyzing spreadsheets manually is incredibly 00:08:17.660 --> 00:08:21.120 tedious and error prone. So true. You can upload 00:08:21.120 --> 00:08:24.040 your raw campaign data or sales numbers. You 00:08:24.040 --> 00:08:26.579 ask it to analyze what actually drove the results. 00:08:26.899 --> 00:08:29.220 You can ask it to build clean, accurate visual 00:08:29.220 --> 00:08:31.879 charts. It writes the code to process that data 00:08:31.879 --> 00:08:34.159 in the background. That's much better than staring 00:08:34.159 --> 00:08:37.039 at rows of numbers yourself. Trying to spot hidden 00:08:37.039 --> 00:08:39.720 statistical patterns alone is incredibly frustrating. 00:08:39.980 --> 00:08:43.320 It's like stacking Lego blocks of data. This 00:08:43.320 --> 00:08:46.120 AI doesn't just sort those blocks by their color. 00:08:46.440 --> 00:08:49.200 It acts as the master builder, assembling the 00:08:49.200 --> 00:08:51.039 pieces together. You get a fully built house 00:08:51.039 --> 00:08:53.120 you can actually live in. It builds the house 00:08:53.120 --> 00:08:55.320 and hands you the front door keys. You just get 00:08:55.320 --> 00:08:58.450 to enjoy the final organized result. If I ask 00:08:58.450 --> 00:09:00.509 for a PowerPoint, am I getting a finished deck 00:09:00.509 --> 00:09:03.070 or just an outline? You receive a fully formatted, 00:09:03.090 --> 00:09:05.350 downloadable PowerPoint file containing slides. 00:09:05.669 --> 00:09:08.269 It includes the structured text, bullet points, 00:09:08.450 --> 00:09:10.850 and basic visual layouts. You might tweak the 00:09:10.850 --> 00:09:13.009 design slightly, but the heavy lifting is finished. 00:09:13.309 --> 00:09:15.330 You get a fully formatted deck ready for final 00:09:15.330 --> 00:09:18.330 polish. Yeah, it completely eliminates the terror 00:09:18.330 --> 00:09:21.649 of the blank white page. So it promises the absolute 00:09:21.649 --> 00:09:24.330 world to us right now. It offers agentic research. 00:09:24.809 --> 00:09:27.750 visual charts, and secure cloud computers. Let's 00:09:27.750 --> 00:09:29.389 look at the actual performance data to verify 00:09:29.389 --> 00:09:32.269 this. It's incredibly easy to make big promises 00:09:32.269 --> 00:09:35.049 in a press release. The actual performance numbers 00:09:35.049 --> 00:09:37.470 are what really matter here. The new system beats 00:09:37.470 --> 00:09:40.370 the old version consistently. It wins over 65 00:09:40.370 --> 00:09:43.269 % of the time in key areas. Those core areas 00:09:43.269 --> 00:09:46.009 include factual accuracy and advanced multilingual 00:09:46.009 --> 00:09:48.490 support. They also include document creation 00:09:48.490 --> 00:09:51.549 and deep, complex research tasks. Let's look 00:09:51.549 --> 00:09:57.250 at a few specific numbers from the update. That 00:09:57.250 --> 00:10:00.629 is a massive jump in its raw comprehension capability. 00:10:00.870 --> 00:10:03.210 It means it doesn't lose the plot in a massive 00:10:03.210 --> 00:10:11.029 book. Whoa. Beat. Imagine scaling to a billion 00:10:11.029 --> 00:10:13.529 queries across the web. Imagine extracting the 00:10:13.529 --> 00:10:15.690 exact right data points without crashing once. 00:10:15.950 --> 00:10:18.070 The scale of that technical improvement is just 00:10:18.070 --> 00:10:20.669 truly mind boggling. I want to ground this in 00:10:20.669 --> 00:10:22.889 reality for a second. Beat. When you look at 00:10:22.889 --> 00:10:25.730 professionals in the field today, how are they 00:10:25.730 --> 00:10:28.980 actually deploying this tool? right now. Researchers 00:10:28.980 --> 00:10:31.639 are using it to conduct massive literature reviews 00:10:31.639 --> 00:10:34.620 quickly. They upload several complex scientific 00:10:34.620 --> 00:10:37.379 papers into the digital workspace. And then they 00:10:37.379 --> 00:10:40.240 ask the AI to map conflicting methodologies. 00:10:40.379 --> 00:10:43.080 Exactly. It spots the subtle gaps and exports 00:10:43.080 --> 00:10:46.120 a clean summary table. That saves a researcher 00:10:46.120 --> 00:10:49.299 weeks of incredibly tedious manual reading. Tech 00:10:49.299 --> 00:10:51.039 teams are also using it in a very interesting 00:10:51.039 --> 00:10:53.679 way. They treat it as a dedicated, lightweight 00:10:53.679 --> 00:10:56.669 coding assistant. It acts as the research layer 00:10:56.669 --> 00:10:59.889 before actual coding begins. Developers have 00:10:59.889 --> 00:11:03.350 to constantly read completely new API documentation. 00:11:03.809 --> 00:11:05.929 It's usually incredibly dense and very hard to 00:11:05.929 --> 00:11:08.490 parse quickly. Right. So Notebook LM reads through 00:11:08.490 --> 00:11:11.350 those highly technical docs for them. It condenses 00:11:11.350 --> 00:11:13.610 the complex roles into a simple markdown file. 00:11:13.830 --> 00:11:16.070 It then feeds that organized file into various 00:11:16.070 --> 00:11:19.210 coding agents. A coding agent is an AI tool that 00:11:19.210 --> 00:11:21.490 writes software code. Those autonomous agents 00:11:21.490 --> 00:11:24.330 work much better with clean, organized instruction. 00:11:25.320 --> 00:11:28.299 Notebook LM structures that messy research before 00:11:28.299 --> 00:11:30.740 the agent starts typing. Small businesses are 00:11:30.740 --> 00:11:33.299 using this exact process for faster decisions. 00:11:33.600 --> 00:11:37.820 They turn messy, raw, and spend data into clear 00:11:37.820 --> 00:11:40.279 action plans. They dump their weekly marketing 00:11:40.279 --> 00:11:42.840 spreadsheets directly into the notebook. The 00:11:42.840 --> 00:11:45.460 AI analyzes the trends and generates a visual 00:11:45.460 --> 00:11:47.720 performance chart. They can quickly decide whether 00:11:47.720 --> 00:11:50.370 to scale a campaign or not. It completely removes 00:11:50.370 --> 00:11:52.470 the friction from the boring middle part. Yeah, 00:11:52.570 --> 00:11:54.529 you have the raw information sitting right there 00:11:54.529 --> 00:11:56.789 before you, but you still need to understand 00:11:56.789 --> 00:11:59.529 and properly organize it. It bridges the frustrating 00:11:59.529 --> 00:12:02.389 gap between raw data and actionable insight. 00:12:02.490 --> 00:12:05.090 It really does. Two sec silence. Is this trying 00:12:05.090 --> 00:12:07.809 to replace coding agents eventually or just making 00:12:07.809 --> 00:12:10.139 those existing agents smarter? It's absolutely 00:12:10.139 --> 00:12:11.980 just making those existing agents much smarter. 00:12:12.159 --> 00:12:14.279 It acts as the critical context layer before 00:12:14.279 --> 00:12:17.279 the coding begins. It organizes the messy research 00:12:17.279 --> 00:12:19.740 so the agent can build accurately. It prepares 00:12:19.740 --> 00:12:22.139 the architectural blueprint before the agent 00:12:22.139 --> 00:12:24.220 builds software. It's a powerful partnership 00:12:24.220 --> 00:12:26.480 that makes the whole ecosystem much stronger. 00:12:26.960 --> 00:12:29.360 We have covered a massive amount of ground today. 00:12:29.519 --> 00:12:32.279 Let's quickly recap the big idea behind this 00:12:32.279 --> 00:12:36.220 major update. Notebook LM has completely transformed 00:12:36.220 --> 00:12:39.179 its core identity and practical purpose. It used 00:12:39.179 --> 00:12:42.340 to be a passive storage box for static PDFs. 00:12:42.360 --> 00:12:46.080 Now it's a highly active, fully agentic, digital 00:12:46.080 --> 00:12:48.799 workspace. It actually carries you through the 00:12:48.799 --> 00:12:51.000 messy middle of a project. It takes you from 00:12:51.000 --> 00:12:54.059 a blank page right to the finish line. It actively 00:12:54.059 --> 00:12:57.279 finds sources and reasons across incredibly complex 00:12:57.279 --> 00:13:00.820 documents. It analyzes raw, unstructured data 00:13:00.820 --> 00:13:04.460 and creates beautiful, accurate charts. It exports 00:13:04.460 --> 00:13:07.460 finished, downloadable files for your daily technical 00:13:07.460 --> 00:13:10.039 workflows. It truly makes the grueling research 00:13:10.039 --> 00:13:12.799 process feel incredibly seamless. It gives you 00:13:12.799 --> 00:13:15.340 your valuable time back to focus on actual thinking. 00:13:15.519 --> 00:13:17.620 That is the real value here. I want to leave 00:13:17.620 --> 00:13:19.299 you with a provocative final thought. We need 00:13:19.299 --> 00:13:21.299 to talk about the new trust layer here. The strict 00:13:21.299 --> 00:13:23.299 source attribution feature is incredibly important 00:13:23.299 --> 00:13:25.240 to mention. It solves one of the biggest problems 00:13:25.240 --> 00:13:28.340 with modern AI tools. The system provides exact 00:13:28.340 --> 00:13:31.399 source attribution for every single chart. It 00:13:31.399 --> 00:13:33.779 does this for every custom report and visual 00:13:33.779 --> 00:13:36.899 slide deck. This completely eliminates the terrifying 00:13:36.899 --> 00:13:39.940 black box of generative AI. You can trace exactly 00:13:39.940 --> 00:13:42.159 where every single fact came from. It links directly 00:13:42.159 --> 00:13:45.500 back to the original text or data point. So if 00:13:45.500 --> 00:13:48.639 Notebook LM already exports slides with perfect 00:13:48.639 --> 00:13:51.860 sourcing attached, what happens when it integrates 00:13:51.860 --> 00:13:54.840 with high -level video generation next? That 00:13:54.840 --> 00:13:57.240 is a truly fascinating direction to think about 00:13:57.240 --> 00:14:00.120 for a second. The technology to generate dynamic 00:14:00.120 --> 00:14:03.220 video from text is advancing incredibly fast. 00:14:03.580 --> 00:14:06.159 Could your messy, scattered notes become a fully 00:14:06.159 --> 00:14:09.480 sourced mini -documentary? Could you generate 00:14:09.480 --> 00:14:11.980 that entire documentary in just a few short minutes? 00:14:12.200 --> 00:14:14.679 The jump from static slides to dynamic video 00:14:14.679 --> 00:14:17.120 is definitely coming. And having guaranteed factual 00:14:17.120 --> 00:14:19.899 accuracy will make it incredibly powerful. Thank 00:14:19.899 --> 00:14:22.120 you for joining us on this deep dive today. I 00:14:22.120 --> 00:14:24.019 highly encourage you to test these new features 00:14:24.019 --> 00:14:25.919 yourself. Yeah, definitely go try it out. Take 00:14:25.919 --> 00:14:27.919 your messiest, most complicated project right 00:14:27.919 --> 00:14:29.879 now. Throw it into Notebook LM and see what it 00:14:29.879 --> 00:14:32.139 actually builds. You'll be absolutely amazed 00:14:32.139 --> 00:14:34.580 at what it uncovers for you. It is going to change 00:14:34.580 --> 00:14:36.779 how you approach complex learning completely. 00:14:37.399 --> 00:14:40.000 The days of staring blankly at a messy folder 00:14:40.000 --> 00:14:42.679 are over. Let the active assistant do the heavy 00:14:42.679 --> 00:14:44.620 lifting for you. Otiro Music.