WEBVTT 00:00:00.000 --> 00:00:02.540 What if you could easily clone your own voice? 00:00:02.940 --> 00:00:06.200 And build a 3D model of your actual body. Plus, 00:00:06.580 --> 00:00:08.140 automate your workday before your coffee gets 00:00:08.140 --> 00:00:10.660 cold. It definitely sounds like pure science 00:00:10.660 --> 00:00:13.279 fiction right now. But that's exactly the reality 00:00:13.279 --> 00:00:15.679 we're unpacking today. Welcome to this NeapDive. 00:00:15.759 --> 00:00:17.379 We're really glad you joined us. Yeah, we have 00:00:17.379 --> 00:00:19.839 a very specific mission for you today. We're 00:00:19.839 --> 00:00:23.879 exploring 10 highly practical AI workflow hacks. 00:00:24.000 --> 00:00:26.559 They fall into three very distinct categories. 00:00:26.800 --> 00:00:30.160 Right. We have visual cloning. creative generation, 00:00:30.960 --> 00:00:33.420 and daily productivity automation. These tools 00:00:33.420 --> 00:00:36.020 fundamentally alter how we interact with technology. 00:00:36.240 --> 00:00:39.299 They do, but there's a crucial insight to establish 00:00:39.299 --> 00:00:41.740 early on. Okay, let's unpack this. The creative 00:00:41.740 --> 00:00:45.119 hacks are visually stunning and really fun. Absolutely. 00:00:45.340 --> 00:00:47.859 But the real -time savings come from pure automation. 00:00:48.320 --> 00:00:50.859 Reclaiming your schedule saves you hours, not 00:00:50.859 --> 00:00:52.859 just minutes. That's where the leverage actually 00:00:52.859 --> 00:00:55.259 lives for you. We also have to remember the golden 00:00:55.259 --> 00:00:58.409 rule here. Right. Better input always equals 00:00:58.409 --> 00:01:00.729 better output. The underlying model doesn't matter 00:01:00.729 --> 00:01:03.090 if your instructions are vague. The precision 00:01:03.090 --> 00:01:05.969 of your prompt dictates the final result. Let's 00:01:05.969 --> 00:01:07.769 start with moving you into the digital world. 00:01:08.530 --> 00:01:11.329 Before AI does the heavy lifting, we must recreate 00:01:11.329 --> 00:01:13.810 you. It's kind of like stacking Lego blocks of 00:01:13.810 --> 00:01:16.150 your own identity. What's fascinating here is 00:01:16.150 --> 00:01:19.170 how simple it is now. We begin by extracting 00:01:19.170 --> 00:01:22.430 a 3D model from a 2D selfie. You upload a standard 00:01:22.430 --> 00:01:25.340 photo to Chad GPT or Gemini. You prompt it to 00:01:25.340 --> 00:01:28.519 create a collectible 3D action figurine. You 00:01:28.519 --> 00:01:30.439 specify it needs to stand on a white background. 00:01:30.719 --> 00:01:33.959 Yeah, and ChatGPT gives you a stylized 2D concept 00:01:33.959 --> 00:01:36.920 image back. AI models usually struggle to guess 00:01:36.920 --> 00:01:39.780 depth from flat pictures. But bringing that generated 00:01:39.780 --> 00:01:43.099 image into a tool called Tripo3D changes everything. 00:01:43.420 --> 00:01:45.620 Tripo3D actually specializes in interpreting 00:01:45.620 --> 00:01:48.329 depth from shading and geometry. It converts 00:01:48.329 --> 00:01:51.329 that flat pixel image into a printable 3D mesh. 00:01:51.569 --> 00:01:53.650 It uses neural networks to infer the missing 00:01:53.650 --> 00:01:56.109 spatial data. You can download the file or use 00:01:56.109 --> 00:01:58.010 their integrated printing service. There's a 00:01:58.010 --> 00:02:00.250 really clever trick to improve this process though. 00:02:00.629 --> 00:02:02.769 A standard selfie usually just shows your upper 00:02:02.769 --> 00:02:06.370 body. Yeah, and that starves the AI of necessary 00:02:06.370 --> 00:02:08.590 visual context. It doesn't know how to anchor 00:02:08.590 --> 00:02:11.610 the figure in 3D space. So you ask ChatGPT to 00:02:11.610 --> 00:02:14.330 extend the image first. You prompt it to generate 00:02:14.330 --> 00:02:17.129 the full body head to toe. That gives Tripo3D 00:02:17.129 --> 00:02:20.430 a complete spatial bounding box to analyze. The 00:02:20.430 --> 00:02:23.310 result is a much cleaner, structurally sound 00:02:23.310 --> 00:02:25.789 3D model. Better visual data yields a better 00:02:25.789 --> 00:02:29.090 physical object. Beat. Let's move from the physical 00:02:29.090 --> 00:02:31.550 shape to the voice. Right. So we use a platform 00:02:31.550 --> 00:02:34.990 called Eleven Labs for voice synthesis. The technology 00:02:34.990 --> 00:02:38.050 behind this is genuinely remarkable. They only 00:02:38.050 --> 00:02:40.949 need a 10 second audio clip to start modeling. 00:02:41.189 --> 00:02:43.860 That sounds incredibly fast. Maybe a little too 00:02:43.860 --> 00:02:46.919 fast. It is. Providing 30 to 60 seconds of audio 00:02:46.919 --> 00:02:49.659 is noticeably better. A voice isn't just pitch 00:02:49.659 --> 00:02:52.939 or basic volume. It's breath patterns, pacing, 00:02:53.259 --> 00:02:55.919 and subtle vocal fry. More audio gives the AI 00:02:55.919 --> 00:02:58.800 a deeper map of those micro -expressions. Once 00:02:58.800 --> 00:03:01.080 cloned, you just type out a script. The system 00:03:01.080 --> 00:03:03.159 reads it back in your exact natural cadence. 00:03:03.319 --> 00:03:04.900 And here's where the technology crosses into 00:03:04.900 --> 00:03:08.189 magic. 11Labs can translate your text into entirely 00:03:08.189 --> 00:03:10.870 different languages. Yeah, it uses your unique 00:03:10.870 --> 00:03:14.330 vocal signature to speak that new language. So 00:03:14.330 --> 00:03:17.129 my voice can speak flawless Japanese without 00:03:17.129 --> 00:03:19.889 me learning it. That's a wild concept to wrap 00:03:19.889 --> 00:03:22.270 your head around. It really is. But we can push 00:03:22.270 --> 00:03:24.569 the digital cloning even further. People throw 00:03:24.569 --> 00:03:27.280 the phrase digital twin around constantly. which 00:03:27.280 --> 00:03:29.319 is just a virtual copy of you that speaks and 00:03:29.319 --> 00:03:31.840 moves naturally. We use a platform called HeyGen 00:03:31.840 --> 00:03:34.819 to build this avatar. You feed it two to five 00:03:34.819 --> 00:03:37.969 minutes of clean video footage. HeyGen maps your 00:03:37.969 --> 00:03:40.550 facial landmarks and tracks your micro -expressions. 00:03:40.830 --> 00:03:43.150 It doesn't just copy your face, it studies your 00:03:43.150 --> 00:03:45.550 physical mannerisms. You give the avatar a script, 00:03:45.610 --> 00:03:47.629 and it delivers it perfectly. You completely 00:03:47.629 --> 00:03:50.129 bypass the camera, the lighting, and the retakes. 00:03:50.210 --> 00:03:52.430 It also translates your speech with automatic 00:03:52.430 --> 00:03:54.990 lip syncing. Right. It actually alters the virtual 00:03:54.990 --> 00:03:57.590 jawline and cheek movements. It matches the new 00:03:57.590 --> 00:03:59.969 foams of whatever language it's speaking. To 00:03:59.969 --> 00:04:03.490 sex silence. This raises a really important question 00:04:03.490 --> 00:04:06.569 for me. If a machine can replicate our exact 00:04:06.569 --> 00:04:10.270 voice and face in minutes, what happens to the 00:04:10.270 --> 00:04:13.740 value of genuine in -person communication? I 00:04:13.740 --> 00:04:16.079 think we'll experience a massive cultural shift. 00:04:16.639 --> 00:04:18.939 When artificial communication becomes incredibly 00:04:18.939 --> 00:04:22.100 cheap and easy to produce, people will inevitably 00:04:22.100 --> 00:04:25.680 crave real, unedited human interaction even more. 00:04:26.259 --> 00:04:28.939 The real world becomes the premium experience. 00:04:29.319 --> 00:04:31.560 So authenticity becomes a premium feature, not 00:04:31.560 --> 00:04:34.339 just the default standard. Precisely. We've successfully 00:04:34.339 --> 00:04:36.920 digitized your identity now. Moving forward, 00:04:37.240 --> 00:04:39.439 how do we generate the actual content? We want 00:04:39.439 --> 00:04:41.899 to share ideas without endless manual labor. 00:04:42.199 --> 00:04:44.139 This is where we shift into creative generation. 00:04:44.300 --> 00:04:47.519 We're moving from identity replication to asset 00:04:47.519 --> 00:04:50.120 creation. Let's talk about generating music with 00:04:50.120 --> 00:04:53.160 an AI called Suno. It builds a complete song 00:04:53.160 --> 00:04:55.519 from a single text prompt. It generates vocals, 00:04:55.920 --> 00:04:58.279 a melody, and the full instrumentation. It works 00:04:58.279 --> 00:05:01.000 a lot like a text generator, actually. It predicts 00:05:01.000 --> 00:05:03.759 the next audio waveform token instead of a word. 00:05:03.939 --> 00:05:07.180 You describe the mood, the genre, or the specific 00:05:07.180 --> 00:05:10.079 story. Suno structures the verse, the chorus, 00:05:10.139 --> 00:05:12.040 and the bridge automatically. If you're feeling 00:05:12.040 --> 00:05:14.199 uninspired, they have a dice button. It throws 00:05:14.199 --> 00:05:16.220 creative prompt ideas at you to break the block. 00:05:16.560 --> 00:05:19.120 Say you want a laid -back lo -fi hip -hop track. 00:05:19.399 --> 00:05:21.480 You ask for soft piano and rain sounds in the 00:05:21.480 --> 00:05:24.560 background. The subject is working late at night, 00:05:24.899 --> 00:05:28.660 tired, but focused. Suno parses that intent and 00:05:28.660 --> 00:05:31.220 delivers a finished track. There's also a brilliant 00:05:31.220 --> 00:05:33.620 way to repurpose your current work here. You 00:05:33.620 --> 00:05:36.720 paste a blog post or an essay directly into Suno. 00:05:36.810 --> 00:05:39.509 You ask it to adapt that written content into 00:05:39.509 --> 00:05:42.689 lyrics. It's a shockingly fast way to create 00:05:42.689 --> 00:05:46.329 engaging audio formats. Beat. But let's connect 00:05:46.329 --> 00:05:48.449 this back to video content. We mentioned HeyGen 00:05:48.449 --> 00:05:51.029 earlier for building personal avatars. But it 00:05:51.029 --> 00:05:53.129 also translates your existing pre -recorded videos 00:05:53.129 --> 00:05:55.129 perfectly. You just upload your file and pick 00:05:55.129 --> 00:05:57.550 the target language. HeyGen translates the audio 00:05:57.550 --> 00:06:00.269 and reconstructs the mouth movements. As we discussed, 00:06:00.389 --> 00:06:03.120 it maps the facial landmarks to new phones. The 00:06:03.120 --> 00:06:04.860 speaker generally looks like they're natively 00:06:04.860 --> 00:06:07.379 speaking Spanish. This is massive for creators 00:06:07.379 --> 00:06:09.879 and global educators. You can reach entirely 00:06:09.879 --> 00:06:12.279 new markets without reshooting a single frame. 00:06:12.439 --> 00:06:14.519 But the output relies entirely on the quality 00:06:14.519 --> 00:06:17.519 of the input. If your original video has terrible 00:06:17.519 --> 00:06:21.180 echoey audio, the translation suffers. Good source 00:06:21.180 --> 00:06:23.699 material gives the AI clean data to manipulate. 00:06:24.160 --> 00:06:27.379 Now let's shift our focus to parsing dense visual 00:06:27.379 --> 00:06:30.220 data. We have a tool called Notebook LM. for 00:06:30.220 --> 00:06:33.660 handling dry research. It turns complex PDFs 00:06:33.660 --> 00:06:36.600 into scannable visual infographics. You upload 00:06:36.600 --> 00:06:39.779 your sources into their secure environment. Notebook 00:06:39.779 --> 00:06:42.560 LM uses a process called retrieval augmented 00:06:42.560 --> 00:06:45.139 generation. Which is an AI that only uses the 00:06:45.139 --> 00:06:47.759 specific documents you give it. Exactly. It anchors 00:06:47.759 --> 00:06:50.519 its understanding strictly to your uploaded documents. 00:06:50.620 --> 00:06:52.699 It maps out the relationships between different 00:06:52.699 --> 00:06:54.540 data points automatically. You don't have to 00:06:54.540 --> 00:06:56.439 manually extract the key statistics yourself. 00:06:56.600 --> 00:06:58.819 You can choose from professional or editorial 00:06:58.819 --> 00:07:01.439 layout styles. If the presets don't work, you 00:07:01.439 --> 00:07:03.839 just type a custom prompt. You tell it to simplify 00:07:03.839 --> 00:07:06.259 the concepts for absolute beginners. You ask 00:07:06.259 --> 00:07:08.279 it to highlight only the most critical financial 00:07:08.279 --> 00:07:10.779 metrics. This is an absolute game changer for 00:07:10.779 --> 00:07:13.480 team summaries. Complex data becomes instantly 00:07:13.480 --> 00:07:16.339 readable at a quick glance. Whoa, imagine scaling 00:07:16.339 --> 00:07:19.300 to a billion queries. Analyzing enterprise level 00:07:19.300 --> 00:07:22.060 databases like that is staggering to think about. 00:07:23.210 --> 00:07:25.649 Let's move on to generating full slide decks. 00:07:26.069 --> 00:07:28.389 We use a platform called Gamma for this workflow. 00:07:28.750 --> 00:07:31.269 You provide a topic, upload a document, or paste 00:07:31.269 --> 00:07:35.329 text. Gamma parses the context and builds a structured 00:07:35.329 --> 00:07:38.069 markdown outline. Then it applies sophisticated 00:07:38.069 --> 00:07:40.949 design systems to render the slides. It creates 00:07:40.949 --> 00:07:43.389 the layout, writes the copy, and sources the 00:07:43.389 --> 00:07:46.350 imagery. It does all of this in under 60 seconds. 00:07:46.569 --> 00:07:48.810 I have to push back on this a little bit. Can 00:07:48.810 --> 00:07:53.170 a 60 second slide deck really capture deep, nuanced 00:07:53.170 --> 00:07:57.050 research? That sounds like a recipe for generic, 00:07:57.290 --> 00:07:59.290 soulless fluff. That's a fair concern. It creates 00:07:59.290 --> 00:08:01.670 a structural baseline, not the final, polished 00:08:01.670 --> 00:08:04.310 masterpiece. It solves the blank page problem 00:08:04.310 --> 00:08:06.430 immediately. You still have to step in and guide 00:08:06.430 --> 00:08:09.670 the AI. Gamma has a built -in AI editor for quick 00:08:09.670 --> 00:08:12.230 revisions. You describe your necessary updates 00:08:12.230 --> 00:08:14.610 using simple, plain English commands. Yeah, which 00:08:14.610 --> 00:08:17.029 makes editing incredibly fast. I still wrestle 00:08:17.029 --> 00:08:19.769 with prompt drift myself. Which is when the AI 00:08:19.769 --> 00:08:21.870 slowly forgets your original typed instructions. 00:08:22.069 --> 00:08:23.910 That definitely happens as the context window 00:08:23.910 --> 00:08:26.290 gets crowded. But Gammae handles direct slide 00:08:26.290 --> 00:08:28.550 edits surprisingly well. You can even present 00:08:28.550 --> 00:08:31.430 directly from the browser window. So AI provides 00:08:31.430 --> 00:08:34.690 the baseline, but humans provide the final polish. 00:08:35.210 --> 00:08:37.690 Exactly. We've explored personal avatars and 00:08:37.690 --> 00:08:40.610 creative content generation. Now we arrive at 00:08:40.610 --> 00:08:42.789 the most crucial category of the day. This is 00:08:42.789 --> 00:08:44.450 where we connect everything to the bigger picture. 00:08:44.690 --> 00:08:46.990 We're going to look at reclaiming your actual 00:08:46.990 --> 00:08:50.860 time. sponsor. Mid -roll sponsor read goes here, 00:08:50.860 --> 00:08:53.100 provided separately. Okay, let's unpack this 00:08:53.100 --> 00:08:56.039 final category of automation tools. Creating 00:08:56.039 --> 00:08:58.639 digital avatars and generating music is visually 00:08:58.639 --> 00:09:01.600 impressive, but the underlying mechanics of automation 00:09:01.600 --> 00:09:04.460 are far more impactful. The most profound time 00:09:04.460 --> 00:09:07.360 savings hide in the incredibly boring tasks. 00:09:07.779 --> 00:09:10.340 We're talking about automating your daily administrative 00:09:10.340 --> 00:09:12.460 grind. Let's start with a fundamentally different 00:09:12.460 --> 00:09:14.679 approach to media editing. We use a platform 00:09:14.679 --> 00:09:17.379 called Descript for video and audio. Traditional 00:09:17.379 --> 00:09:20.220 editing is purely spatial. You manually cut blocks 00:09:20.220 --> 00:09:23.139 of time on a visual timeline. Descript completely 00:09:23.139 --> 00:09:25.559 flips that paradigm on its head. It forces an 00:09:25.559 --> 00:09:28.019 alignment between audio waveforms and text characters. 00:09:28.419 --> 00:09:30.399 It converts your uploaded video into a written 00:09:30.399 --> 00:09:32.740 text transcript. What's fascinating here is how 00:09:32.740 --> 00:09:35.720 it treats video like a document. To edit the 00:09:35.720 --> 00:09:38.720 media, you literally just edit the text document. 00:09:39.179 --> 00:09:41.840 You highlight a messy sentence and press the 00:09:41.840 --> 00:09:44.620 delete key. That specific section is instantly 00:09:44.620 --> 00:09:47.269 removed from the actual video. This feels like 00:09:47.269 --> 00:09:50.049 treating reality like a word processor. You delete 00:09:50.049 --> 00:09:53.090 a printed word and time just skips forward. It 00:09:53.090 --> 00:09:55.870 democratizes editing for people who hate complex 00:09:55.870 --> 00:09:59.230 timelines. Descript also has an incredibly powerful 00:09:59.230 --> 00:10:02.590 feature called the Underlord. The Underlord automatically 00:10:02.590 --> 00:10:05.750 scans your file for awkward pauses. It identifies 00:10:05.750 --> 00:10:08.230 filler words and rambling tangents in seconds. 00:10:08.370 --> 00:10:10.389 You can remove all of them with a single click. 00:10:10.509 --> 00:10:12.909 You don't have to scrub through hours of raw 00:10:12.909 --> 00:10:15.429 footage manually. They also include an audio 00:10:15.429 --> 00:10:18.399 repair tool called Studio Sound. It isolates 00:10:18.399 --> 00:10:21.000 your voice and digitally removes background noise. 00:10:21.340 --> 00:10:23.399 It essentially regenerates the frequencies of 00:10:23.399 --> 00:10:25.799 your spoken words. It makes a bad microphone 00:10:25.799 --> 00:10:27.720 sound like a professional studio environment. 00:10:27.840 --> 00:10:29.940 It takes one click and saves you from frustrating 00:10:29.940 --> 00:10:32.700 re -records. Now let's explore delegating your 00:10:32.700 --> 00:10:35.279 schedule to an AI. A lot of people are intimidated 00:10:35.279 --> 00:10:37.700 by the concept of agents, but if we simplify 00:10:37.700 --> 00:10:40.970 it, it's very approachable. An AI agent is just 00:10:40.970 --> 00:10:43.490 a smart assistant that completes tasks across 00:10:43.490 --> 00:10:46.490 different apps. We're looking at the ChatGPP 00:10:46.490 --> 00:10:49.309 agent functionality specifically. It acts as 00:10:49.309 --> 00:10:52.090 an orchestrator for your daily software. It uses 00:10:52.090 --> 00:10:54.929 API calls to talk to your existing applications. 00:10:55.190 --> 00:10:58.509 Tools like Google Calendar, Gmail, and your team's 00:10:58.509 --> 00:11:00.990 Slack channels. You write out a plain English 00:11:00.990 --> 00:11:03.950 intent for the agent. For example, you want it 00:11:03.950 --> 00:11:06.779 to organize your chaotic morning. The agent converts 00:11:06.779 --> 00:11:10.100 your intent into specific data requests. It pulls 00:11:10.100 --> 00:11:12.419 your unread emails and cross -references your 00:11:12.419 --> 00:11:14.899 daily meetings. You ask it to flag anything that 00:11:14.899 --> 00:11:17.279 is strictly urgent. You tell it to find empty 00:11:17.279 --> 00:11:20.220 blocks for deep focused work. It analyzes the 00:11:20.220 --> 00:11:23.080 overlap and generates a prioritized daily plan. 00:11:23.419 --> 00:11:25.379 You can even set this to run automatically every 00:11:25.379 --> 00:11:27.600 morning. But there's a very important trap to 00:11:27.600 --> 00:11:30.620 avoid here. Start by connecting just one single 00:11:30.620 --> 00:11:32.639 application at first. Right, that's crucial. 00:11:33.100 --> 00:11:35.100 Linking your calendar alone provides incredibly 00:11:35.100 --> 00:11:37.899 clean, reliable output. If you connect 10 apps 00:11:37.899 --> 00:11:40.500 immediately, the agent hallucinates and breaks. 00:11:40.879 --> 00:11:42.940 Focused data streams prevent the system from 00:11:42.940 --> 00:11:45.360 getting completely confused. Let's move to our 00:11:45.360 --> 00:11:47.559 final workflow hack for the day. We're looking 00:11:47.559 --> 00:11:50.440 at turning flat spreadsheets into interactive 00:11:50.440 --> 00:11:53.659 dashboards. We use an AI model called Claude. 00:11:53.840 --> 00:11:57.259 for this process. You upload a massive confusing 00:11:57.259 --> 00:12:00.980 CSV file into Claude. You describe the specific 00:12:00.980 --> 00:12:03.700 trends you want to visualize. Claude isn't just 00:12:03.700 --> 00:12:06.600 drawing a static picture of your data. It actually 00:12:06.600 --> 00:12:09.519 writes and executes React code in the background. 00:12:09.679 --> 00:12:11.720 It builds a lightweight functional web application 00:12:11.720 --> 00:12:15.539 just for you. The output is a highly visual interactive 00:12:15.539 --> 00:12:18.840 dashboard interface. Claude publishes this artifact 00:12:18.840 --> 00:12:21.159 and gives you a simple web link. You send that 00:12:21.159 --> 00:12:23.629 link directly to your team or client. They never 00:12:23.629 --> 00:12:25.789 have to open the original intimidating spreadsheet 00:12:25.789 --> 00:12:27.730 file. You can instruct Claw to add functional 00:12:27.730 --> 00:12:30.330 data filters too. Users can sort the generated 00:12:30.330 --> 00:12:32.850 charts by specific dates or categories. It looks 00:12:32.850 --> 00:12:35.230 like you spent days coding a custom analytics 00:12:35.230 --> 00:12:37.830 tool, but you really just typed a paragraph of 00:12:37.830 --> 00:12:42.389 instructions. Two secs silence. Let me ask a 00:12:42.389 --> 00:12:44.190 deeper philosophical question about this. Go 00:12:44.190 --> 00:12:47.149 for it. If we fully automate our daily planning 00:12:47.149 --> 00:12:50.580 and our data analysis, Do we risk losing our 00:12:50.580 --> 00:12:53.399 intuitive grasp on our own metrics? We definitely 00:12:53.399 --> 00:12:55.799 lose that granular friction of managing every 00:12:55.799 --> 00:12:59.179 tiny detail. But friction isn't always valuable. 00:12:59.759 --> 00:13:02.820 By stepping back, we gain a much broader strategic 00:13:02.820 --> 00:13:05.600 view. So we trade micromanagement for higher 00:13:05.600 --> 00:13:08.840 -level strategic clarity. Exactly. We stop drowning 00:13:08.840 --> 00:13:11.200 in the data entry and start analyzing the actual 00:13:11.200 --> 00:13:13.440 insights. We've covered a tremendous amount of 00:13:13.440 --> 00:13:15.929 ground in this deep dive. from generating 3D 00:13:15.929 --> 00:13:19.070 models to deploying custom code via Claude. It's 00:13:19.070 --> 00:13:21.070 easy to feel overwhelmed by the sheer volume 00:13:21.070 --> 00:13:23.710 of tools, but there's a vital philosophy to take 00:13:23.710 --> 00:13:26.289 away from this. Do not attempt to adopt all 10 00:13:26.289 --> 00:13:28.870 of these workflows today. Trying to overhaul 00:13:28.870 --> 00:13:31.789 your entire life at once guarantees failure. 00:13:31.929 --> 00:13:34.269 You have to isolate one single friction point 00:13:34.269 --> 00:13:36.789 in your day. Just pick one specific problem and 00:13:36.789 --> 00:13:39.190 test a free tier. If you hate writing meeting 00:13:39.190 --> 00:13:42.259 recats, build a chat GPT agent. If your audio 00:13:42.259 --> 00:13:44.600 sounds terrible, run it through Descript Studio 00:13:44.600 --> 00:13:46.779 Sound. Once you experience the friction disappearing, 00:13:47.100 --> 00:13:49.899 the rest clicks into place. The underlying logic 00:13:49.899 --> 00:13:52.059 of prompt engineering starts to feel completely 00:13:52.059 --> 00:13:54.799 intuitive. The landscape of our daily work is 00:13:54.799 --> 00:13:57.860 accelerating at lightning speed. It's a thrilling 00:13:57.860 --> 00:14:01.460 time to rethink how we spend our energy. Thank 00:14:01.460 --> 00:14:03.889 you for joining us on this deep dive today. It's 00:14:03.889 --> 00:14:06.490 been a fascinating exploration of what is actually 00:14:06.490 --> 00:14:08.509 possible. But before we let you go, consider 00:14:08.509 --> 00:14:11.309 this final thought. If your personal agent is 00:14:11.309 --> 00:14:13.590 automatically summarizing your incoming emails. 00:14:13.830 --> 00:14:16.049 And your colleague's AI agent is the one actually 00:14:16.049 --> 00:14:18.470 writing them. Are we just creating a world where 00:14:18.470 --> 00:14:21.690 machines talk to machines? Do we just sit back 00:14:21.690 --> 00:14:23.789 and take the credit for the conversation? Think 00:14:23.789 --> 00:14:25.950 about that the next time you auto -generate a 00:14:25.950 --> 00:14:27.830 reply. OTRO music.