WEBVTT 00:00:00.000 --> 00:00:03.160 I was sitting in my car this morning just staring 00:00:03.160 --> 00:00:05.179 at the dashboard and I had this weird thought 00:00:05.179 --> 00:00:07.080 it feels like most of us are sitting in the driver's 00:00:07.080 --> 00:00:10.039 seat of a Ferrari this machine engineered for 00:00:10.039 --> 00:00:12.259 you know absolute peak performance but we're 00:00:12.259 --> 00:00:14.980 driving it around a golf course at five miles 00:00:14.980 --> 00:00:17.539 an hour we have this massive engine under the 00:00:17.539 --> 00:00:20.660 hood but we're just we're afraid to do anything 00:00:20.660 --> 00:00:23.339 more than tap the accelerator that is wow that's 00:00:23.850 --> 00:00:25.649 painfully accurate, especially if you look at 00:00:25.649 --> 00:00:27.870 how the average person is using Gemini 3 .0 right 00:00:27.870 --> 00:00:30.149 now. We treat it like a search engine from, what, 00:00:30.190 --> 00:00:33.469 2010. Ask a question, get an answer, close the 00:00:33.469 --> 00:00:36.130 tab. But the reality is this thing is designed 00:00:36.130 --> 00:00:37.929 to run your life while you sleep. It's not a 00:00:37.929 --> 00:00:39.929 chatbot anymore. It's a workflow engine. Welcome 00:00:39.929 --> 00:00:42.490 to the deep dive. I want to slow things down 00:00:42.490 --> 00:00:44.789 today. We usually rush through the news, but 00:00:44.789 --> 00:00:46.490 today we have a piece of source material that 00:00:46.490 --> 00:00:48.850 I think requires us to actually, you know, pause 00:00:48.850 --> 00:00:52.289 and think. We are looking at a guide by Max Ann 00:00:52.289 --> 00:00:56.270 titled, You're Using Gemini 3 .0 Wrong. It was 00:00:56.270 --> 00:00:58.409 published just yesterday, February 8th, 2026. 00:00:58.890 --> 00:01:01.009 And the premise is exactly what you just said. 00:01:01.280 --> 00:01:03.899 The AI landscape has shifted right under our 00:01:03.899 --> 00:01:06.180 feet, and most of us just completely missed it. 00:01:06.280 --> 00:01:08.500 It really has. And that shift is subtle, but 00:01:08.500 --> 00:01:10.939 it's profound. We're not just generating text 00:01:10.939 --> 00:01:12.760 anymore. We are building deliverables. We're 00:01:12.760 --> 00:01:15.260 building systems. Right. So our mission today 00:01:15.260 --> 00:01:18.560 is to unpack seven specific hidden features that 00:01:18.560 --> 00:01:20.640 are buried, and I mean deep in the settings menu 00:01:20.640 --> 00:01:23.260 of Gemini 3 .0. These are features that, according 00:01:23.260 --> 00:01:25.319 to Ahn, change the whole game. But before we 00:01:25.319 --> 00:01:26.900 get to the how -to, I want to talk about the 00:01:26.900 --> 00:01:30.239 friction. Ann calls it the amnesia problem. This 00:01:30.239 --> 00:01:32.099 is the core argument of the whole guide, really. 00:01:32.260 --> 00:01:34.099 It definitely resonated with me, but I have to 00:01:34.099 --> 00:01:36.659 play devil's advocate for a second. Isn't amnesia 00:01:36.659 --> 00:01:38.900 just a feature of software? I mean, I open Word, 00:01:39.040 --> 00:01:41.579 it's blank. I open Excel, it's blank. Why is 00:01:41.579 --> 00:01:43.760 this a problem for AI? Because Word and Excel 00:01:43.760 --> 00:01:47.000 are tools you use to output your thoughts. The 00:01:47.000 --> 00:01:49.500 AI is supposed to be a partner that contributes 00:01:49.500 --> 00:01:52.569 thoughts. Think about it this way. If you hired 00:01:52.569 --> 00:01:55.209 a new executive assistant, a human being, and 00:01:55.209 --> 00:01:57.090 every single morning when you walked into the 00:01:57.090 --> 00:01:59.549 office, you had to reintroduce yourself. Hi, 00:01:59.730 --> 00:02:01.750 I'm Dave. I wrote a podcast. Here's our style 00:02:01.750 --> 00:02:03.670 guide. Here's who our sponsors are. And then 00:02:03.670 --> 00:02:05.230 the next day you do it all over again. Okay, 00:02:05.290 --> 00:02:07.450 yeah. I'd fire them by lunch. Exactly. Yeah. 00:02:07.530 --> 00:02:09.650 You'd never get any deep work done because you're 00:02:09.650 --> 00:02:12.389 spending 90 % of your energy just on onboarding. 00:02:12.949 --> 00:02:15.229 But that is the default setting for Gemini. It 00:02:15.229 --> 00:02:17.719 starts from zero. Every time. It has no memory 00:02:17.719 --> 00:02:20.159 of your style, your context, your projects. So 00:02:20.159 --> 00:02:23.159 real work, complex work, it exposes those gaps 00:02:23.159 --> 00:02:25.840 immediately. You feel that friction because you're 00:02:25.840 --> 00:02:28.680 just constantly repeating context. So if the 00:02:28.680 --> 00:02:30.979 intelligence, I mean, if the intelligence isn't 00:02:30.979 --> 00:02:34.219 the problem, but the memory is, what is the architectural 00:02:34.219 --> 00:02:36.990 fix here? Because I don't want to just give it 00:02:36.990 --> 00:02:39.610 all my data and hope for the best. The fix is 00:02:39.610 --> 00:02:41.270 that you have to stop treating it like a blank 00:02:41.270 --> 00:02:44.169 slate. You have to enable the persistent productivity 00:02:44.169 --> 00:02:47.009 system settings. You build a system of record 00:02:47.009 --> 00:02:49.650 inside the AI. Okay, let's get into the mechanics 00:02:49.650 --> 00:02:52.069 of that. This leads us to the first big feature, 00:02:52.270 --> 00:02:56.189 max and highlights, custom gems. Now, I'll be 00:02:56.189 --> 00:02:58.310 honest, gems sounds a bit like a mobile game 00:02:58.310 --> 00:03:01.009 currency to me. It does, doesn't it? But stick 00:03:01.009 --> 00:03:03.509 with me on this one. So in this context, it's 00:03:03.509 --> 00:03:05.819 something very different. Right. Think of a gem 00:03:05.819 --> 00:03:08.599 as a cloned version of the AI that you have trained 00:03:08.599 --> 00:03:11.479 for one very specific purpose. You aren't just 00:03:11.479 --> 00:03:13.919 opening a chat. You are defining three things. 00:03:14.099 --> 00:03:17.219 A roll. a task and an output can you walk us 00:03:17.219 --> 00:03:19.580 through a concrete example the guide mentions 00:03:19.580 --> 00:03:21.620 a meeting summary assistant but that sounds pretty 00:03:21.620 --> 00:03:23.840 basic i can already ask it to summarize meetings 00:03:23.840 --> 00:03:26.939 you can but how often do you get a summary that's 00:03:26.939 --> 00:03:29.240 actually you know usable usually it's too long 00:03:29.240 --> 00:03:32.400 or completely misses the point so with the gem 00:03:32.400 --> 00:03:34.500 you don't just paste notes and say summarize 00:03:34.500 --> 00:03:38.340 this you build the gem once you tell it roll 00:03:39.069 --> 00:03:41.550 You are my executive assistant. Task. When I 00:03:41.550 --> 00:03:44.250 paste meeting notes, extract key decisions, action 00:03:44.250 --> 00:03:47.629 items with owners, and any open questions. Output. 00:03:47.909 --> 00:03:51.210 Keep it under 300 words. Use bullet points. Okay, 00:03:51.270 --> 00:03:53.150 so it's a saved prompt. I feel like I've seen 00:03:53.150 --> 00:03:54.849 that before. It's more than a prompt. This is 00:03:54.849 --> 00:03:57.430 the part that people miss. You can upload reference 00:03:57.430 --> 00:04:00.750 files to the gem. Up to 10 documents. What kind 00:04:00.750 --> 00:04:04.110 of documents? Like a PDF of my company handbook 00:04:04.110 --> 00:04:05.949 or something? You could, but that's kind of the 00:04:05.949 --> 00:04:08.189 boring way to do it. The guide suggests something 00:04:08.189 --> 00:04:10.990 much smarter. Don't just upload a generic style 00:04:10.990 --> 00:04:13.310 guide. Upload the last five emails you wrote 00:04:13.310 --> 00:04:15.610 that you were actually proud of. Upload a project 00:04:15.610 --> 00:04:18.069 brief that got approved immediately. That's interesting. 00:04:18.170 --> 00:04:20.230 So you're giving it training data on your best 00:04:20.230 --> 00:04:22.529 work. Exactly. You're saying, don't just write 00:04:22.529 --> 00:04:24.850 like a professional. Write like me. on a good 00:04:24.850 --> 00:04:28.490 day so it mimics your specific voice your formatting 00:04:28.490 --> 00:04:31.329 even your shorthand That's a huge difference. 00:04:31.490 --> 00:04:34.610 Usually I find myself typing the same three paragraphs 00:04:34.610 --> 00:04:37.389 of instructions. You know, don't use emojis. 00:04:37.389 --> 00:04:40.610 Keep it brief before I even paste the content 00:04:40.610 --> 00:04:42.430 I want processed. Right. And that's that friction 00:04:42.430 --> 00:04:45.029 we talked about. With a gem, you click the preset 00:04:45.029 --> 00:04:47.730 in your sidebar, paste the notes, and it just 00:04:47.730 --> 00:04:50.170 executes. It already knows the rules before you 00:04:50.170 --> 00:04:52.529 type a single word. So effectively it becomes 00:04:52.529 --> 00:04:55.269 a clickable preset that houses your specific 00:04:55.269 --> 00:04:57.949 institutional knowledge. Precisely. All right. 00:04:58.079 --> 00:05:01.220 Let's pivot. So gems solve the issue of repeating 00:05:01.220 --> 00:05:04.220 instructions, but that doesn't solve the issue 00:05:04.220 --> 00:05:07.180 of the AI being isolated from your actual data. 00:05:07.339 --> 00:05:08.980 It still doesn't know what I'm working on unless 00:05:08.980 --> 00:05:12.620 I paste it in. Right. By default, Gemini is sitting 00:05:12.620 --> 00:05:15.180 in a clean room. It can't see your Gmail, can't 00:05:15.180 --> 00:05:17.420 see your drive. It doesn't know you have a dentist 00:05:17.420 --> 00:05:20.920 appointment at 2 p .m. It's smart, but it's blind 00:05:20.920 --> 00:05:23.060 to your actual life. And this brings us to feature 00:05:23.060 --> 00:05:26.310 number two, connecting Google Workspace. There's 00:05:26.310 --> 00:05:28.310 a toggle in the settings to change that. Yes, 00:05:28.470 --> 00:05:30.990 under Connected Apps. Once you flip that switch, 00:05:31.230 --> 00:05:34.029 you unlock the ability to use the at symbol to 00:05:34.029 --> 00:05:36.050 direct its attention. I have to admit, when I 00:05:36.050 --> 00:05:37.970 read this part of the guide, I was a little skeptical. 00:05:38.269 --> 00:05:40.730 It felt a bit gimmicky. But then I tried the 00:05:40.730 --> 00:05:44.399 example. at Gmail draft a reply to Sarah about 00:05:44.399 --> 00:05:47.439 the Q1 report. And what happened? Well, it worked. 00:05:47.660 --> 00:05:50.439 But more importantly, it didn't just write a 00:05:50.439 --> 00:05:53.920 generic Dear Sarah email. It actually found the 00:05:53.920 --> 00:05:57.120 thread. That's the whoa moment. It reads the 00:05:57.120 --> 00:05:58.939 thread to understand the context of what Sarah 00:05:58.939 --> 00:06:01.540 asked. Yeah. It looks at your previous replies 00:06:01.540 --> 00:06:03.560 to understand the tone you usually take with 00:06:03.560 --> 00:06:06.300 her. And then it drafts the response. It's not 00:06:06.300 --> 00:06:08.720 just writing text. It's performing an action 00:06:08.720 --> 00:06:11.610 based on history. And it's not just email. The 00:06:11.610 --> 00:06:14.589 guide lists at Drive to find files, at Calendar 00:06:14.589 --> 00:06:16.889 to check schedules, and even at Workspace to 00:06:16.889 --> 00:06:18.769 search everything at once. Think about the panic 00:06:18.769 --> 00:06:20.610 search. You know that feeling, right? You're 00:06:20.610 --> 00:06:22.470 on a Zoom call. Someone asks, hey, where did 00:06:22.470 --> 00:06:24.430 we land on that budget proposal from last Thursday? 00:06:24.750 --> 00:06:27.050 Oh, God. And you start sweating. You open three 00:06:27.050 --> 00:06:29.689 tabs. Is it in Drive? Is it in Slack? Is it in 00:06:29.689 --> 00:06:32.410 an email? Exactly. Instead of panic searching, 00:06:32.610 --> 00:06:35.689 you just type into the chat. At Workspace, find 00:06:35.689 --> 00:06:38.519 the deck from my Thursday client call. And it 00:06:38.519 --> 00:06:40.720 pulls it up. It just eliminates that cognitive 00:06:40.720 --> 00:06:42.980 load of, where did I put that thing? Does this 00:06:42.980 --> 00:06:45.759 fundamentally change the search behavior of the 00:06:45.759 --> 00:06:48.980 user? Yes, completely. It eliminates tab switching 00:06:48.980 --> 00:06:52.019 by bringing the data directly into the chat interface. 00:06:52.519 --> 00:06:54.620 Let's talk about the third feature, which I think 00:06:54.620 --> 00:06:56.899 is my favorite concept in the whole guide, scheduled 00:06:56.899 --> 00:07:00.879 briefings. Max Ann describes this as work happening 00:07:00.879 --> 00:07:03.899 while you sleep. This is where we move from being 00:07:03.899 --> 00:07:08.000 reactive to proactive. Usually you have to initiate 00:07:08.000 --> 00:07:10.480 the interaction with the AI. You ask, it answers. 00:07:10.860 --> 00:07:13.279 Schedule briefings flips that. You write a prompt, 00:07:13.379 --> 00:07:15.360 but you attach a time trigger to it. So give 00:07:15.360 --> 00:07:17.439 me the mechanics. How does that look in practice? 00:07:17.600 --> 00:07:19.579 Do I need to know how to code or something? No, 00:07:19.579 --> 00:07:21.199 and that's the beauty of it. It's surprisingly 00:07:21.199 --> 00:07:23.980 natural language. You literally type every Monday 00:07:23.980 --> 00:07:27.139 at 7 a .m., summarize my calendar, and highlight 00:07:27.139 --> 00:07:29.079 meetings that need preparation. I have to be 00:07:29.079 --> 00:07:31.899 honest here. I still wrestle with prompt drift 00:07:31.899 --> 00:07:34.769 myself. I have all these good intentions to ask 00:07:34.769 --> 00:07:37.550 for a summary or check my tasks, but I get busy 00:07:38.120 --> 00:07:40.620 The day starts and I forget to ask until I'm 00:07:40.620 --> 00:07:42.500 already late for the meeting. That is exactly 00:07:42.500 --> 00:07:44.839 what this solves. It automates the asking. Yeah. 00:07:44.920 --> 00:07:47.079 You can set it up for daily briefings on unread 00:07:47.079 --> 00:07:50.740 emails from your boss or a Friday 5 p .m. summary 00:07:50.740 --> 00:07:53.040 of everything you completed that week so you 00:07:53.040 --> 00:07:55.699 can log your hours. It shifts the dynamic completely. 00:07:55.879 --> 00:07:57.819 You aren't chasing the information. The information 00:07:57.819 --> 00:08:00.180 is waiting for you when you wake up. Precisely. 00:08:00.180 --> 00:08:02.480 It's the difference between pulling data and 00:08:02.480 --> 00:08:04.800 having data pushed to you. And I assume there's 00:08:04.800 --> 00:08:07.240 a dashboard for this. Or does it just run for 00:08:07.240 --> 00:08:09.240 hours? and I can never stop it. No, thankfully. 00:08:09.379 --> 00:08:12.439 There's a specific scheduled actions panel in 00:08:12.439 --> 00:08:14.339 the settings where you can pause or edit them. 00:08:14.620 --> 00:08:17.680 Okay, feature number four. This connects back 00:08:17.680 --> 00:08:19.800 to the memory issue we started with, but on a 00:08:19.800 --> 00:08:21.800 macro level. It's called personal instructions. 00:08:22.180 --> 00:08:24.759 Think of this as global memory. So we talked 00:08:24.759 --> 00:08:28.100 about custom gems, which are for specific tasks, 00:08:28.199 --> 00:08:30.980 like a summary assistant. Personal instructions 00:08:30.980 --> 00:08:33.860 are for you as a user. Across every interaction, 00:08:34.019 --> 00:08:36.299 every chat, every day. So this is where I tell 00:08:36.299 --> 00:08:39.120 it things like I'm a podcaster or I prefer American 00:08:39.120 --> 00:08:41.980 English spelling. Exactly. Instead of explaining 00:08:41.980 --> 00:08:44.700 I like bullet points or keep it professional, 00:08:44.840 --> 00:08:47.639 in every single chat, you save it once in the 00:08:47.639 --> 00:08:50.759 settings. The source lists examples like use 00:08:50.759 --> 00:08:53.080 a direct professional tone, keep replies under 00:08:53.080 --> 00:08:56.299 200 words, or I work in B2B sauce marketing. 00:08:56.519 --> 00:08:59.529 But hold on. People change. My job changes. If 00:08:59.529 --> 00:09:01.330 I set this up to say I'm a marketing manager 00:09:01.330 --> 00:09:03.590 and then I switch to being a, I don't know, a 00:09:03.590 --> 00:09:06.549 yoga instructor, isn't this going to screw up 00:09:06.549 --> 00:09:08.710 all my results? That's a valid fear. But the 00:09:08.710 --> 00:09:10.730 guide mentions a really important command here, 00:09:10.809 --> 00:09:13.370 the forget function. It's granular. You can tell 00:09:13.370 --> 00:09:16.149 Gemini, forget my memory about formatting preferences 00:09:16.149 --> 00:09:18.710 or forget that I work in marketing. And it wipes 00:09:18.710 --> 00:09:20.909 just that specific part while keeping the rest. 00:09:21.360 --> 00:09:23.240 That's smart. It lets the system evolve with 00:09:23.240 --> 00:09:25.360 you. What's the psychological benefit of this 00:09:25.360 --> 00:09:28.100 feature beyond just saving keystrokes? It stops 00:09:28.100 --> 00:09:30.679 the feeling of talking to a stranger. The tool 00:09:30.679 --> 00:09:33.580 adapts to your identity. We're going to take 00:09:33.580 --> 00:09:35.320 a quick break, but when we come back, we're going 00:09:35.320 --> 00:09:38.480 to look at the flip side of memory, when you 00:09:38.480 --> 00:09:41.179 don't want the AI to remember you, and a feature 00:09:41.179 --> 00:09:43.220 for the heavy researchers out there that bridges 00:09:43.220 --> 00:09:45.500 the gap between your notes and the internet. 00:09:46.299 --> 00:09:49.480 Stay with us. Mid -roll sponsor break placeholder. 00:09:50.439 --> 00:09:53.399 Welcome back to the Deep Dive. We're unpacking 00:09:53.399 --> 00:09:56.320 the hidden features of Gemini 3 .0 based on Max 00:09:56.320 --> 00:09:58.360 Anne's new guide. We just spent a lot of time 00:09:58.360 --> 00:10:00.539 talking about memory, context, and letting the 00:10:00.539 --> 00:10:03.419 AI know everything about you. But the guide brings 00:10:03.419 --> 00:10:05.539 up a really important counterpoint in feature 00:10:05.539 --> 00:10:08.779 number five, privacy. Right, because sometimes 00:10:08.779 --> 00:10:11.720 memory is a liability. It is. Sometimes I just 00:10:11.720 --> 00:10:14.480 want to ask a dumb question. Or I'm brainstorming 00:10:14.480 --> 00:10:16.720 a surprise party for my wife, and I don't want 00:10:16.720 --> 00:10:18.799 party planning to suddenly become part of my 00:10:18.799 --> 00:10:21.000 professional work profile. Exactly. Or you're 00:10:21.000 --> 00:10:22.539 researching a medical symptom. You don't want 00:10:22.539 --> 00:10:24.779 that influencing your future business drafts. 00:10:24.779 --> 00:10:28.179 The guide calls this temporary chat mode. It's 00:10:28.179 --> 00:10:30.220 essentially incognito mode for your workflow. 00:10:30.419 --> 00:10:33.960 Yes. When you toggle this on, the history is 00:10:33.960 --> 00:10:37.159 deleted after 72 hours. And more importantly, 00:10:37.320 --> 00:10:39.960 that data isn't used to train the model and it 00:10:39.960 --> 00:10:42.139 doesn't get saved to your long -term memory settings. 00:10:42.419 --> 00:10:44.980 It's a safety valve. You can ask the dumb questions 00:10:44.980 --> 00:10:47.480 or the sensitive questions without polluting 00:10:47.480 --> 00:10:49.840 your long -term profile. Exactly. It keeps your 00:10:49.840 --> 00:10:52.340 global memory clean. Now, feature number six 00:10:52.340 --> 00:10:54.580 is a heavy hitter. This is for the people who 00:10:54.580 --> 00:10:57.399 are doing deep work. The guide talks about the 00:10:57.399 --> 00:11:01.019 integration of Notebook LM. Now, I know Notebook 00:11:01.019 --> 00:11:03.200 LM as a separate app. I've used it to organize 00:11:03.200 --> 00:11:06.039 research papers. Is this different? It's the 00:11:06.039 --> 00:11:07.899 same brain, but now it's integrated directly 00:11:07.899 --> 00:11:10.340 into the interface. It's designed for when you 00:11:10.340 --> 00:11:12.620 need to synthesize a massive amount of information. 00:11:13.259 --> 00:11:15.340 The guide says you can attach a notebook with 00:11:15.340 --> 00:11:18.940 up to 300 sources, PDFs, videos, Google Docs, 00:11:18.940 --> 00:11:21.000 entire websites. That's a library. That's not 00:11:21.000 --> 00:11:23.340 just a file. That's a massive archive. It is. 00:11:23.460 --> 00:11:26.000 But the magic in Gemini 3 .0 is the combination. 00:11:26.360 --> 00:11:28.440 In the old world, you had your static notes in 00:11:28.440 --> 00:11:30.840 one app and live search engine in another. Here, 00:11:30.919 --> 00:11:32.899 you have both. So you can ask a question that 00:11:32.899 --> 00:11:35.879 bridges both worlds. Yes. The example in the 00:11:35.879 --> 00:11:38.779 guide is perfect, you can ask. What are the latest 00:11:38.779 --> 00:11:41.740 trends in AI video? That's a live web search. 00:11:42.019 --> 00:11:44.440 And then in the same prompt, add, and how do 00:11:44.440 --> 00:11:46.679 they compare to the research in my attached notebook? 00:11:47.700 --> 00:11:50.139 And that's the static knowledge base. That is 00:11:50.139 --> 00:11:52.700 incredibly powerful. You're cross -referencing 00:11:52.700 --> 00:11:55.019 the entire internet against your private archives 00:11:55.019 --> 00:11:58.179 in seconds. It creates a synthesis that neither 00:11:58.179 --> 00:12:00.820 tool could do on its own. The live web doesn't 00:12:00.820 --> 00:12:02.440 know your notes, and your notes don't know what 00:12:02.440 --> 00:12:04.750 happened five minutes ago. Putting them together 00:12:04.750 --> 00:12:07.370 is where the insight happens. So it essentially 00:12:07.370 --> 00:12:09.370 bridges the gap between your private archive 00:12:09.370 --> 00:12:12.669 and the public internet. Exactly. Deep structured 00:12:12.669 --> 00:12:15.470 research meets real -time data in a single answer. 00:12:15.649 --> 00:12:18.190 We are down to the final feature, number seven, 00:12:18.370 --> 00:12:21.590 and the bonus. This is about moving beyond text. 00:12:21.710 --> 00:12:25.490 We're talking about canvas and formulas. I feel 00:12:25.490 --> 00:12:27.409 like canvas is a term that gets thrown around 00:12:27.409 --> 00:12:29.509 a lot in tech. What does it mean here? Canvas 00:12:29.509 --> 00:12:32.600 is a shift in interface. Usually you chat, you 00:12:32.600 --> 00:12:35.820 get a column of text. It's linear. But if you're 00:12:35.820 --> 00:12:37.620 building a presentation or a coding project, 00:12:37.980 --> 00:12:41.059 linear text is terrible. Canvas opens a dedicated 00:12:41.059 --> 00:12:43.740 window on the side to build actual artifacts. 00:12:44.139 --> 00:12:47.470 The guide mentions asking it to... create a 10 00:12:47.470 --> 00:12:49.690 -slide presentation. And it doesn't just outline 00:12:49.690 --> 00:12:52.370 the slides in text. It builds the slides. It 00:12:52.370 --> 00:12:54.429 handles the formatting, the structure, the flow. 00:12:54.610 --> 00:12:57.049 And because it's a dedicated window, you can 00:12:57.049 --> 00:12:59.070 edit the text directly, just like you would in 00:12:59.070 --> 00:13:01.250 a document editor, while still chatting with 00:13:01.250 --> 00:13:03.470 the AI to make big changes. So I can highlight 00:13:03.470 --> 00:13:05.429 slide three and just say, make this punchier. 00:13:05.590 --> 00:13:08.360 Exactly. It's collaborative editing, not just 00:13:08.360 --> 00:13:11.039 command and response. It also mentions building 00:13:11.039 --> 00:13:14.340 web apps or interactive quizzes from dense PDFs. 00:13:14.720 --> 00:13:18.259 Turn this policy document into a quiz and it 00:13:18.259 --> 00:13:20.220 builds a playable interface. It's moving from 00:13:20.220 --> 00:13:22.240 tell me to make this. Which brings us to the 00:13:22.240 --> 00:13:25.360 bonus feature AI formulas in Google Sheets. Oh 00:13:25.360 --> 00:13:27.620 man, this one hits home. I live in spreadsheets 00:13:27.620 --> 00:13:30.580 and they are usually a mess. We all do. This 00:13:30.580 --> 00:13:33.299 feature is a superpower for messy data. It's 00:13:33.299 --> 00:13:36.360 the AI function. Walk me through a painful scenario. 00:13:36.899 --> 00:13:38.860 Let's say I just got back from a trade show. 00:13:39.159 --> 00:13:42.899 Okay, so you have a CSV export of 500 people. 00:13:43.100 --> 00:13:45.299 The names are mixed with the job titles. Some 00:13:45.299 --> 00:13:47.740 are in all caps, some are lowercase. And the 00:13:47.740 --> 00:13:50.379 interests column is just a paragraph of text. 00:13:50.940 --> 00:13:53.580 Usually cleaning that up is two hours of manual 00:13:53.580 --> 00:13:56.740 copy pasting or writing complex nested formulas 00:13:56.740 --> 00:13:58.980 that break if you breathe on them wrong. Right, 00:13:59.039 --> 00:14:00.840 and I usually just give up and do it manually. 00:14:01.059 --> 00:14:04.029 With this, you just type AI. Extract the job 00:14:04.029 --> 00:14:07.049 title and format as title case A2 and drag it 00:14:07.049 --> 00:14:09.929 down the column. It processes every row individually 00:14:09.929 --> 00:14:13.049 using the LLM. That is wild. It treats every 00:14:13.049 --> 00:14:15.990 cell like a tiny prompt. Exactly. You can use 00:14:15.990 --> 00:14:18.330 it to translate, to extract sentiment from customer 00:14:18.330 --> 00:14:21.809 reviews, to reformat dates, anything. But there's 00:14:21.809 --> 00:14:23.889 a catch, right? It can't do a million rows instantly. 00:14:24.169 --> 00:14:26.210 Correct. The guide notes that it currently processes 00:14:26.210 --> 00:14:29.090 only the first 200 cells at a time due to usage 00:14:29.090 --> 00:14:32.350 caps. Still, for a lot of tasks, 200 rows of 00:14:32.350 --> 00:14:34.340 automation... thinking is a lot better than zero. 00:14:34.519 --> 00:14:36.940 Absolutely. It turns a two -hour headache into 00:14:36.940 --> 00:14:38.960 a two -minute drag and drop. So let's zoom out. 00:14:39.039 --> 00:14:41.879 We've covered seven features, custom gems, workspace 00:14:41.879 --> 00:14:44.299 connections, scheduling, personal instructions, 00:14:44.620 --> 00:14:49.820 privacy, Newtbook LM, and Canvas. What is the 00:14:49.820 --> 00:14:52.240 big idea here? If you take one thing away from 00:14:52.240 --> 00:14:55.259 this deep dive, what should it be? The big idea 00:14:55.259 --> 00:14:58.080 is the shift in mindset. We are moving from a 00:14:58.080 --> 00:15:00.460 search engine mindset to a productivity system 00:15:00.460 --> 00:15:03.710 mindset. The search engine is passive. You go 00:15:03.710 --> 00:15:06.049 to it when you have a question. The system is 00:15:06.049 --> 00:15:08.149 active. It surrounds you. It feels like the goal 00:15:08.149 --> 00:15:10.429 is to set up a system that remembers how you 00:15:10.429 --> 00:15:12.230 work so you don't have to repeat yourself. That's 00:15:12.230 --> 00:15:14.730 it. Automation in context. Yeah. If you were 00:15:14.730 --> 00:15:18.309 repeating yourself to your AI in 2026, you were 00:15:18.309 --> 00:15:20.809 using it wrong. The friction isn't the tool's 00:15:20.809 --> 00:15:23.110 intelligence. It's the user's default settings. 00:15:23.710 --> 00:15:25.590 Maxanne ends the guide with a bit of a challenge. 00:15:25.629 --> 00:15:28.269 She says, stop reading. Start setting this up. 00:15:28.559 --> 00:15:30.799 It's good advice. It feels a little unfair how 00:15:30.799 --> 00:15:33.139 much time this saves once it's running, but you 00:15:33.139 --> 00:15:34.919 have to do the initial lift. You have to go into 00:15:34.919 --> 00:15:36.460 the settings, turn on the workspace connection, 00:15:36.659 --> 00:15:38.899 and write your first gem. It's the difference 00:15:38.899 --> 00:15:41.419 between driving the Ferrari in first gear and 00:15:41.419 --> 00:15:44.740 finally shifting up. Well said. Thanks for listening. 00:15:44.899 --> 00:15:47.460 Go fix your settings. We'll see you in the next 00:15:47.460 --> 00:15:48.980 deep dive. Thanks, everyone.