Pełna transkrypcja odcinka ReadMultiplex.com: Why I Built GrokBot.Rocks The Compressive Directory Of All Known Public Grok Bots.

neural 00:03:26.219 --> 00:03:28.960 networks. The sources argue that to see exactly 00:03:28.960 --> 00:03:31.280 where we are going with autonomous AI agents, 00:03:31.479 --> 00:03:33.599 you really only have to look at the oldest story 00:03:33.599 --> 00:03:35.719 in the history of tools. We have to go back. 00:03:35.860 --> 00:03:38.580 Way back. We have to examine the transition from 00:03:38.580 --> 00:03:41.819 the desktop to the pocket and look at the exact 00:03:41.819 --> 00:03:44.419 moments when manual sequences were annihilated 00:03:44.419 --> 00:03:47.039 by new technology. Annihilated is a strong word, 00:03:47.060 --> 00:03:49.419 but it's accurate. The first example that comes 00:03:49.419 --> 00:03:51.740 to mind, and one that Rommel's work highlights, 00:03:51.919 --> 00:03:54.080 is the transition from the typewriter to the 00:03:54.080 --> 00:03:56.219 word processor. Oh, that's a classic one. Right. 00:03:56.319 --> 00:03:58.659 When you think about the late 1970s and early 00:03:58.659 --> 00:04:01.879 1980s, when CRT screens first started replacing 00:04:01.879 --> 00:04:04.580 physical paper on a roller. The leap in productivity 00:04:04.580 --> 00:04:07.400 was staggering. It was. I mean, you could backspace 00:04:07.400 --> 00:04:10.379 without using whiteout. Right. No more messy 00:04:10.379 --> 00:04:12.659 correction fluid. You could move an entire paragraph 00:04:12.659 --> 00:04:14.900 from page three to page one with a few keystrokes 00:04:14.900 --> 00:04:17.560 instead of retyping the whole document. Spelling 00:04:17.560 --> 00:04:20.600 errors plummeted, too. Exactly. The final result, 00:04:20.740 --> 00:04:23.579 which is a clean manuscript, arrived drastically 00:04:23.579 --> 00:04:26.660 faster. And yet there was a massive cohort of 00:04:26.660 --> 00:04:28.899 highly skilled writers, journalists, office workers 00:04:28.899 --> 00:04:31.759 who just aggressively refused to switch. And 00:04:31.759 --> 00:04:34.180 we have to ask why. Because it wasn't because 00:04:34.180 --> 00:04:36.019 they didn't want a clean manuscript. Everyone 00:04:36.019 --> 00:04:38.459 wants a clean manuscript. Right. Obviously. It 00:04:38.459 --> 00:04:40.579 was because they missed the tactile feedback 00:04:40.579 --> 00:04:42.879 of the sequence. They missed the physical resistance 00:04:42.879 --> 00:04:46.189 of the keys, the smell of the ink ribbon. The 00:04:46.189 --> 00:04:48.329 loud, percussive sound of the carriage return 00:04:48.329 --> 00:04:51.050 at the end of a line. Exactly. That ding and 00:04:51.050 --> 00:04:53.269 the swipe. They had spent decades internally 00:04:53.269 --> 00:04:56.290 associating the physical friction of typing with 00:04:56.290 --> 00:04:58.850 the intellectual act of, quote unquote, real 00:04:58.850 --> 00:05:01.449 writing. So when the screen removed the physical 00:05:01.449 --> 00:05:04.149 friction, they felt like they weren't truly working 00:05:04.149 --> 00:05:06.990

anymore. They felt like frauds almost. They clung 00:05:06.990 --> 00:05:09.050 to the old sequence because the process was quite 00:05:09.050 --> 00:05:11.790 literally their identity. Wow. And we saw the 00:05:11.790 --> 00:05:14.290 exact same psychological block a decade later. 00:05:14.620 --> 00:05:16.560 Right, with the rise of the internet. Specifically 00:05:16.560 --> 00:05:19.819 in academia and research. Yes. Suddenly, massive 00:05:19.819 --> 00:05:22.879 university libraries and sprawling physical card 00:05:22.879 --> 00:05:26.199 catalogs were just digitized. Overnight, practically. 00:05:26.360 --> 00:05:28.459 Anyone could query a database from a terminal 00:05:28.459 --> 00:05:30.959 and finish in minutes what used to require a 00:05:30.959 --> 00:05:33.459 physical trip across campus. Hauling a stack 00:05:33.459 --> 00:05:36.139 of heavy, dusty indexes to a desk. Yeah, spending 00:05:36.139 --> 00:05:39.379 hours manually cross -referencing authors. Yet, 00:05:39.399 --> 00:05:41.819 as the research points out, a significant number 00:05:41.819 --> 00:05:44.500 of veteran researchers just... Kept walking the 00:05:44.500 --> 00:05:47.240 stacks. They insisted on physically roaming the 00:05:47.240 --> 00:05:49.800 library aisles. And again, the psychology is 00:05:49.800 --> 00:05:52.839 fascinating here. In their minds, the sheer physical 00:05:52.839 --> 00:05:55.920 volume of the books, the tangible flipping of 00:05:55.920 --> 00:05:58.399 the index cards, confirmed their thoroughness. 00:05:58.540 --> 00:06:00.259 And the physical exhaustion at the end of the 00:06:00.259 --> 00:06:03.120 day. Right. If they just typed a query and got 00:06:03.120 --> 00:06:05.100 the answer instantly, they felt they had cheated. 00:06:05.339 --> 00:06:07.860 Like they didn't earn the answer. Exactly. They 00:06:07.860 --> 00:06:10.139 equated the sequence of the search with the rigor 00:06:10.139 --> 00:06:12.399 of the findings. And then, you know, you shrink 00:06:12.399 --> 00:06:15.240 that entire revolution down into the pocket with 00:06:15.240 --> 00:06:18.379 the iPhone in 2007. That was a huge cultural 00:06:18.379 --> 00:06:21.199 shift. I remember this vivid shift in corporate 00:06:21.199 --> 00:06:23.879 culture. Suddenly the calendar, the camera, the 00:06:23.879 --> 00:06:26.920 map, the Rolodex, the physical notepad. All of 00:06:26.920 --> 00:06:29.579 it. They're all absorbed into one glass rectangle. 00:06:30.420 --> 00:06:32.639 Meetings that used to require a phone call, a 00:06:32.639 --> 00:06:35.819 secretary checking a desk calendar, a physical 00:06:35.819 --> 00:06:38.939 paper diary. They became digital entries. Yeah. 00:06:39.230 --> 00:06:41.629 that magically, silently updated across everyone's 00:06:41.629 --> 00:06:43.990 devices instantly. But you still had managers, 00:06:44.350 --> 00:06:47.949 often the most senior ones, who fiercely clung 00:06:47.949 --> 00:06:50.629 to their expensive, leather -bound paper planners. 00:06:50.949 --> 00:06:54.170 Why? I never understood that. Well, because the 00:06:54.170 --> 00:06:56.610 act of writing the time and place down with a 00:06:56.610 --> 00:06:59.129 fountain pen made the meeting feel arranged. 00:06:59.550 --> 00:07:02.129 Oh, I see. The digital update felt too invisible. 00:07:02.350 --> 00:07:05.149 It didn't feel real. It is the exact same dynamic 00:07:05.149 --> 00:07:08.199 playing out in a loop. The sequence improves, 00:07:08.459 --> 00:07:11.259 the mechanical steps between an idea and a finished 00:07:11.259 --> 00:07:14.399 product shrink. But a cohort of otherwise capable 00:07:14.399 --> 00:07:16.740 people stays with the methods they already know. 00:07:16.879 --> 00:07:19.480 They fall in love with the process, not the result. 00:07:19.660 --> 00:07:22.000 Exactly. But I want to push further back into 00:07:22.000 --> 00:07:24.639 the history here because the sources trace this 00:07:24.639 --> 00:07:27.860 battle between sequence versus result way back. 00:07:28.079 --> 00:07:30.959 Long before silicon chips. Right. Long before 00:07:30.959 --> 00:07:33.800 even electricity. The historical timeline Brian 00:07:33.800 --> 00:07:36.180 outlines is incredible, particularly when you 00:07:36.180 --> 00:07:38.519 look at the printing industry. Okay, tell me 00:07:38.519 --> 00:07:41.199 about the printing industry. If we span from 00:07:41.199 --> 00:07:45.500 the 1880s to 1985, we see this exact cycle play 00:07:45.500 --> 00:07:48.860 out in just brutal fashion. For centuries, since 00:07:48.860 --> 00:07:51.259 Gutenberg basically, the sequence of setting 00:07:51.259 --> 00:07:54.120 type was incredibly manual. It was a true craft. 00:07:54.399 --> 00:07:57.680 It meant a highly skilled hand compositor standing 00:07:57.680 --> 00:08:00.600 in front of a massive wooden case, physically 00:08:00.600 --> 00:08:03.800 picking up one tiny backwards metal letter at 00:08:03.800 --> 00:08:06.199 a time and placing it into a composing stick. 00:08:06.439 --> 00:08:08.439 And they had to read upside down and backwards. 00:08:08.680 --> 00:08:11.319 Yes. It was a craft that took years to master. 00:08:11.540 --> 00:08:14.100 Right. And they were covered in lead dust, working 00:08:14.100 --> 00:08:17.069 by gaslight. It was exhausting work. And then 00:08:17.069 --> 00:08:20.509 in the 1880s, Otmar Mergenthaler invents the 00:08:20.509 --> 00:08:23.230 linotype machine. And this fundamentally breaks 00:08:23.230 --> 00:08:26.170 the old sequence. So how did the linotype actually 00:08:26.170 --> 00:08:28.970 work? Because it was a massive deal. To understand 00:08:28.970 --> 00:08:30.990 how disruptive the linotype was, you have to 00:08:30.990 --> 00:08:33.149 understand the mechanism. Instead of picking 00:08:33.149 --> 00:08:36.250 up individual letters, an operator sat at a 90 00:08:36.250 --> 00:08:38.750 -key keyboard. Kind of like a massive typewriter. 00:08:38.970 -->

00:08:41.570 Right. And when they pressed a key, a brass matrix, 00:08:41.830 --> 00:08:44.230 which is basically a mold of a letter, dropped 00:08:44.230 --> 00:08:47.330 from a magazine into a line. Okay. Once the line 00:08:47.330 --> 00:08:49.850 was full, the machine injected molten lead directly 00:08:49.850 --> 00:08:52.809 into the molds, casting an entire solid line 00:08:52.809 --> 00:08:55.809 of type at once. A linotype. A linotype, hence 00:08:55.809 --> 00:08:58.610 the name. That is wild. It effectively increased 00:08:58.610 --> 00:09:00.809 the speed of typesetting by, what, an order of 00:09:00.809 --> 00:09:03.789 magnitude? At least. Newspapers just exploded 00:09:03.789 --> 00:09:05.830 in size. They could print multiple editions a 00:09:05.830 --> 00:09:07.850 day because the sequence got radically shorter. 00:09:08.090 --> 00:09:11.169 Exactly. But the hand compositors were devastated, 00:09:11.370 --> 00:09:13.330 of course, because their manual dexterity was 00:09:13.330 --> 00:09:15.860 suddenly devalued. All those years of learning 00:09:15.860 --> 00:09:18.799 to read backwards just gone. But the cycle doesn't 00:09:18.799 --> 00:09:21.320 stop there. Fast forward exactly a century to 00:09:21.320 --> 00:09:24.299 1985, which is the dawn of the desktop publishing 00:09:24.299 --> 00:09:26.600 revolution. Right. This is the era of the Apple 00:09:26.600 --> 00:09:30.100 LaserWriter and Adobe PostScript. And Aldus PageMaker. 00:09:30.899 --> 00:09:33.580 In fact, Paul Brainerd, the founder of Aldus, 00:09:33.600 --> 00:09:36.100 is the one who actually coined the term desktop 00:09:36.100 --> 00:09:38.980 publishing. I didn't know that. Think about the 00:09:38.980 --> 00:09:42.230 disruption at that specific moment, though. typesetting 00:09:42.230 --> 00:09:45.210 houses that had survived the linotype era, companies 00:09:45.210 --> 00:09:48.110 that had defined this meticulous, expensive, 00:09:48.389 --> 00:09:51.730 heavy machinery craft for generations. They suddenly 00:09:51.730 --> 00:09:54.889 had to watch a single person sitting at a single 00:09:54.889 --> 00:09:58.029 desk in a suburban office produce a completely 00:09:58.029 --> 00:10:00.929 finished, mathematically perfect layout on a 00:10:00.929 --> 00:10:03.570 screen. And just print it out locally. The manual 00:10:03.570 --> 00:10:06.009 setting of type, whether by hand or by molten 00:10:06.009 --> 00:10:09.269 lead, just vanished entirely. But here is the 00:10:09.269 --> 00:10:11.149 crucial takeaway from the source material, and 00:10:11.149 --> 00:10:13.049 I want to make sure we underline this for you 00:10:13.049 --> 00:10:15.509 listening. It's an important point. The skill 00:10:15.509 --> 00:10:18.950 that survived that 1985 revolution was not the 00:10:18.950 --> 00:10:21.090 mechanical ability to set type. No, it wasn't. 00:10:21.169 --> 00:10:24.169 The skill that survived was the eye for a good 00:10:24.169 --> 00:10:27.169 page, the aesthetic judgment. The understanding 00:10:27.169 --> 00:10:30.669 of kerning, hierarchy, and balance. Right. The 00:10:30.669 --> 00:10:33.509 sequence was destroyed. But the judgment remained. 00:10:33.809 --> 00:10:36.350 That distinction is paramount. When the sequence 00:10:36.350 --> 00:10:39.149 is automated, the judgment is elevated. I love 00:10:39.149 --> 00:10:41.700 that. I want to look at another historical example 00:10:41.700 --> 00:10:43.779 from the sources because it introduces a slightly 00:10:43.779 --> 00:10:46.700 different angle on this. The story of the telephone 00:10:46.700 --> 00:10:51.860 dial in 1891. Ah, yes. Allman Strauger. Yes. 00:10:52.279 --> 00:10:54.700 Allman Strauger, who's a Kansas City undertaker. 00:10:54.899 --> 00:10:56.840 Which is just the perfect weird historical detail. 00:10:57.139 --> 00:11:00.519 You could never make that up. An undertaker invents 00:11:00.519 --> 00:11:03.990 the rotary dial. The story of Strauger perfectly 00:11:03.990 --> 00:11:06.549 illustrates how frustration with a manual sequence 00:11:06.549 --> 00:11:09.769 drives innovation. In the late 1880s, if you 00:11:09.769 --> 00:11:11.230 wanted to make a phone call, you didn't dial 00:11:11.230 --> 00:11:13.870 a number. Right. You picked up an earpiece, cranked 00:11:13.870 --> 00:11:16.210 a handle, and spoke to a human operator at a 00:11:16.210 --> 00:11:18.669 central switchboard. You would say, connect me 00:11:18.669 --> 00:11:21.090 to the bakery. And the operator would physically 00:11:21.090 --> 00:11:23.549 take a patch cord and plug it into the bakery's 00:11:23.549 --> 00:11:27.190 jack on a massive wall of circuits. But Strauger, 00:11:27.330 --> 00:11:30.889 our undertaker, had a problem. he noticed a sudden 00:11:30.889 --> 00:11:33.149 drop in his business. A very suspicious drop. 00:11:33.250 --> 00:11:35.909 Very suspicious. The story goes that he discovered 00:11:35.909 --> 00:11:39.110 the local human telephone operator was actually 00:11:39.110 --> 00:11:41.570 the wife of a competing undertaker in town. Talk 00:11:41.570 --> 00:11:44.409 about a conflict of interest. Seriously. So Strouger 00:11:44.409 --> 00:11:46.509 became deeply paranoid, and probably rightfully 00:11:46.509 --> 00:11:49.029 so, that whenever someone called the exchange 00:11:49.029 --> 00:11:51.950 asking for an undertaker, she was secretly routing 00:11:51.950 --> 00:11:54.269 his incoming business calls to her husband instead 00:11:54.269 --> 00:11:56.340 of him. Whether he was right about the conspiracy 00:11:56.340 --> 00:11:58.580 or not is almost beside the point, honestly, 00:11:58.700 --> 00:12:01.100 because his reaction changed the world. He realized 00:12:01.100 --> 00:12:02.899 the human in the middle was the problem. Exactly. 00:12:03.220 --> 00:12:05.299 The manual sequence of the switchboard operator 00:12:05.299 --> 00:12:08.159 was a vulnerability. So he designed and patented 00:12:08.159 --> 00:12:10.720 the first

commercially successful automatic telephone 00:12:10.720 --> 00:12:13.139 exchange. It was a complex electromechanical 00:12:13.139 --> 00:12:15.519 switch driven by a rotary dial on the user's 00:12:15.519 --> 00:12:17.399 phone. Right. The user dials the number, the 00:12:17.399 --> 00:12:19.440 electrical pulses control the switches at the 00:12:19.440 --> 00:12:21.620 exchange, and the connection is made automatically. 00:12:22.340 --> 00:12:25.059 The rotary dial entirely removed the manual sequence 00:12:25.059 --> 00:12:28.000 of asking a human to plug a wire into a board. 00:12:28.179 --> 00:12:31.759 The sequence died. But the result, which is the 00:12:31.759 --> 00:12:34.340 actual conversation between two people, remained. 00:12:34.919 --> 00:12:38.240 In fact, it flourished because the friction was 00:12:38.240 --> 00:12:40.620 gone. The conversation was always the point, 00:12:40.679 --> 00:12:43.720 not the patch cords. And we see this friction 00:12:43.720 --> 00:12:46.899 removed again in 1972 with the shift from the 00:12:46.899 --> 00:12:49.580 physical slide rule to the pocket calculator. 00:12:50.259 --> 00:12:52.460 This is perhaps the most direct parallel to the 00:12:52.460 --> 00:12:55.179 cognitive shift we are experiencing today. I 00:12:55.179 --> 00:12:58.169 agree. For the vast majority of the 20th century, 00:12:58.389 --> 00:13:01.169 engineers, physicists, and architects lived and 00:13:01.169 --> 00:13:04.149 died by the slide rule. As the text notes, the 00:13:04.149 --> 00:13:06.450 slide rule literally went to the moon. It did. 00:13:06.769 --> 00:13:09.330 The engineers at NASA, computing orbital mechanics 00:13:09.330 --> 00:13:11.690 for the Apollo missions, were using slide rules 00:13:11.690 --> 00:13:14.309 made by companies like Kufel and Esser. It was 00:13:14.309 --> 00:13:16.870 a physical instrument requiring intense manual 00:13:16.870 --> 00:13:19.409 manipulation to perform logarithmic calculations. 00:13:20.350 --> 00:13:23.389 But then, in 1972, Hewlett -Packard introduces 00:13:23.389 --> 00:13:27.269 the HP -35. It was the world's first scientific 00:13:27.269 --> 00:13:30.190 pocket calculator. It could perform trigonometric 00:13:30.190 --> 00:13:33.049 and exponential functions with just the press 00:13:33.049 --> 00:13:35.830 of a plastic button. The result was instantaneous. 00:13:36.610 --> 00:13:39.730 And within just a few years, the great American 00:13:39.730 --> 00:13:42.190 slide rule manufacturers essentially collapsed. 00:13:42.570 --> 00:13:44.730 Kefla and Esser produced their last custom slide 00:13:44.730 --> 00:13:48.730 rule in 1976. Just four years later. Okay. I 00:13:48.730 --> 00:13:51.730 hear the historical argument. I really do. But 00:13:51.730 --> 00:13:55.009 I have to pause here. Go for it. Because I can't 00:13:55.009 --> 00:13:57.690 let go of the idea that we are losing the craft. 00:13:57.950 --> 00:14:01.370 The craft of the sequence. Right. When an engineer 00:14:01.370 --> 00:14:04.309 is forced to use a slide rule, the physical friction 00:14:04.309 --> 00:14:07.149 of manipulating the slip forces them to intimately 00:14:07.149 --> 00:14:10.110 understand the underlying math. They have to 00:14:10.110 --> 00:14:12.250 keep the decimal point in their head. They have 00:14:12.250 --> 00:14:14.070 to understand the magnitude of the numbers intuitively. 00:14:14.730 --> 00:14:17.149 If a digital device just spits out the answer, 00:14:17.740 --> 00:14:19.799 Don't we just end up with a generation of operators 00:14:19.799 --> 00:14:21.980 who don't actually understand the bridge they 00:14:21.980 --> 00:14:24.879 are building? It's a common fear. It feels like 00:14:24.879 --> 00:14:28.240 we are comparing a slow, meticulously home -cooked 00:14:28.240 --> 00:14:31.340 meal to a microwave TV dinner. Sure, popping 00:14:31.340 --> 00:14:33.360 a tray in the microwave is faster. The sequence 00:14:33.360 --> 00:14:35.720 is shorter. But is the quality of the result 00:14:35.720 --> 00:14:38.039 actually the same? What's fascinating here is 00:14:38.039 --> 00:14:40.860 that that is the most common counter -argument, 00:14:40.879 --> 00:14:43.600 but the historical record... firmly rejects it. 00:14:43.720 --> 00:14:45.879 And it's important to understand why the microwave 00:14:45.879 --> 00:14:48.700 dinner analogy fails mechanically. Why does it 00:14:48.700 --> 00:14:51.620 fail? Because the microwave automates the recipe. 00:14:51.740 --> 00:14:54.620 It replaces the judgment of the chef with a prepackaged 00:14:54.620 --> 00:14:58.139 mediocre standard. But the HP35 calculator did 00:14:58.139 --> 00:15:00.659 not replace the engineer's judgment. It replaced 00:15:00.659 --> 00:15:03.840 the grunt work. Oh, I see. To refine your analogy, 00:15:04.039 --> 00:15:06.059 the calculator isn't the microwave replacing 00:15:06.059 --> 00:15:09.370 the chef. The calculator is the sous chef. who 00:15:09.370 --> 00:15:11.450 chops onions for four hours, freeing the head 00:15:11.450 --> 00:15:13.769 chef to actually design the menu. That's a great 00:15:13.769 --> 00:15:15.669 way to put it. When the slide rule vanished, 00:15:15.850 --> 00:15:19.169 the calculator didn't abandon the bridge. The 00:15:19.169 --> 00:15:21.389 bridge still got built, and it was actually engineered 00:15:21.389 --> 00:15:24.889 better. Better? How? Because the engineer could 00:15:24.889 --> 00:15:27.429 test a dozen different stress scenarios in the 00:15:27.429 --> 00:15:29.909 time it used to take to calculate just one. So 00:15:29.909 --> 00:15:32.029 the people who made the switch didn't lose their 00:15:32.029 --> 00:15:34.309 judgment. They simply applied their high -level 00:15:34.309 --> 00:15:36.649 judgment to a vastly larger volume of finished 00:15:36.649 --> 00:15:40.190 work. Exactly. Desktop publishing didn't abandon 00:15:40.190 --> 00:15:42.950 the beautiful page. It allowed a single designer

00:15:42.950 --> 00:15:45.149 to iterate on 20 different layouts in a day, 00:15:45.230 --> 00:15:47.429 applying their aesthetic judgment to more work. 00:15:47.590 --> 00:15:49.769 While the holdouts ended up applying their identical 00:15:49.769 --> 00:15:52.309 level of judgment to significantly less work. 00:15:52.549 --> 00:15:55.649 The engineer didn't disappear in 1972. It stopped 00:15:55.649 --> 00:15:58.250 wasting their entire afternoon doing basic arithmetic. 00:15:58.629 --> 00:16:01.690 That reframing is powerful. The engineer didn't 00:16:01.690 --> 00:16:04.070 disappear. They just got to spend more time actually 00:16:04.070 --> 00:16:06.769 engineering. Right. And that bridges us perfectly 00:16:06.769 --> 00:16:09.750 from these physical tools to the exact software 00:16:09.750 --> 00:16:12.690 precedent that Brian Romley argues mirrors today's 00:16:12.690 --> 00:16:14.929 AI revolution. The perfect software parallel. 00:16:15.110 --> 00:16:17.169 To understand the Grokbot, we have to look at 00:16:17.169 --> 00:16:20.009 1979. We have to talk about the birth of the 00:16:20.009 --> 00:16:22.210 electronic spreadsheet. We have to talk about 00:16:22.210 --> 00:16:25.309 VisiCalc. VisiCalc is arguably the most vital 00:16:25.309 --> 00:16:27.149 precedent for understanding the Grokbot era. 00:16:27.549 --> 00:16:30.009 For anybody who doesn't know the history, VisiCalc 00:16:30.009 --> 00:16:32.289 was created by Dan Bricklin and Bob Frankston 00:16:32.289 --> 00:16:34.470 for the Apple II. A legendary piece of software. 00:16:34.909 --> 00:16:36.649 And to truly appreciate it, you have to picture 00:16:36.649 --> 00:16:39.230 what financial modeling looked like before 1979. 00:16:39.710 --> 00:16:42.009 We aren't talking about opening Excel. We are 00:16:42.009 --> 00:16:44.570 talking about massive, physical paper ledger 00:16:44.570 --> 00:16:47.710 sheets spread across giant desks. Agonizing is 00:16:47.710 --> 00:16:50.370 the only word for the old sequence. Let's walk 00:16:50.370 --> 00:16:52.970 through the mechanics of it. Imagine a massive 00:16:52.970 --> 00:16:56.110 spreadsheet drawn on paper tracking a company's 00:16:56.110 --> 00:16:58.370 five -year financial projection. Okay, I'm picturing 00:16:58.370 --> 00:17:01.230 it. Lots of tiny boxes. Every row and column 00:17:01.230 --> 00:17:03.730 interacts. Now imagine the boss walks in and 00:17:03.730 --> 00:17:06.309 says, What if our materials cost goes up by 2 00:17:06.309 --> 00:17:09.450 % next quarter? Panic. Utter panic. Exactly. 00:17:09.569 --> 00:17:11.890 Because changing that one assumption meant you 00:17:11.890 --> 00:17:14.410 had to physically get out a pencil eraser, erase 00:17:14.410 --> 00:17:17.289 the materials cost cell, manually calculate the 00:17:17.289 --> 00:17:20.009 new number, write it down, and then erase and 00:17:20.009 --> 00:17:22.509 manually recalculate every single dependent cell 00:17:22.509 --> 00:17:25.069 all the way down to the bottom line. So you would 00:17:25.069 --> 00:17:27.289 have junior accountants working overnight, covered 00:17:27.289 --> 00:17:30.569 in eraser dust, running 10 key calculators, just 00:17:30.569 --> 00:17:33.269 to see how a single 2 % variable changed the 00:17:33.269 --> 00:17:35.849 profit margin. It was a massive manual sequence. 00:17:36.349 --> 00:17:38.869 And then Dan Bricklin is sitting in a classroom 00:17:38.869 --> 00:17:41.309 at Harvard Business School, watching a professor 00:17:41.309 --> 00:17:44.170 erase a blackboard to recalculate a model, and 00:17:44.170 --> 00:17:47.430 he realizes there has to be a better way. Enter 00:17:47.430 --> 00:17:50.809 VisiCalc. The mechanism of VisiCalc was a revelation. 00:17:51.069 --> 00:17:54.329 It introduced instant recalculation. Instant 00:17:54.329 --> 00:17:57.430 recalculation. You change one single number at 00:17:57.430 --> 00:17:59.589 the top of the digital sheet, and the underlying 00:17:59.589 --> 00:18:02.569 logic tree instantly updates every dependent 00:18:02.569 --> 00:18:04.990 cell on the screen. The result was so profoundly 00:18:04.990 --> 00:18:07.170 transformative that accounting firms and businesses 00:18:07.170 --> 00:18:10.950 were buying $2 ,000 Apple II computers for no 00:18:10.950 --> 00:18:13.690 other reason than to run this $100 piece of software. 00:18:13.849 --> 00:18:15.670 It's fundamentally changed the behavior of an 00:18:15.670 --> 00:18:18.170 entire profession. Accountants stopped being 00:18:18.170 --> 00:18:21.269 what the sources brilliantly call human recalculation 00:18:21.269 --> 00:18:23.769 engines. Human recalculation engines. When the 00:18:23.769 --> 00:18:26.170 manual sequence of erasing and writing was removed, 00:18:26.269 --> 00:18:28.349 they evolved into strategic analysts. Suddenly 00:18:28.349 --> 00:18:30.650 they could ask, what if a hundred times in a 00:18:30.650 --> 00:18:33.009 single afternoon? They could model a dozen different 00:18:33.009 --> 00:18:35.410 business scenarios before lunch. And that mechanism, 00:18:35.589 --> 00:18:38.190 the instant recalculation of a sequence, brings 00:18:38.190 --> 00:18:41.210 us directly to the modern parallel. On October 00:18:41.210 --> 00:18:44.890 29th, 2026, Brian Rommel posted an incredibly 00:18:44.890 --> 00:18:47.670 bold claim. What did he say? He wrote, Grokbot 00:18:47.670 --> 00:18:50.009 is the most important business tool of the last 00:18:50.009 --> 00:18:52.789 hundred years. That's a massive statement. Yeah, 00:18:52.789 --> 00:18:54.329 he's saying it's bigger than the commercial internet, 00:18:54.509 --> 00:18:56.769 bigger than the smartphone. How does he justify 00:18:56.769 --> 00:18:59.710 that? He justifies it based on what the technology 00:18:59.710 --> 00:19:03.160 scales. Think about the progression. The linotype 00:19:03.160 --> 00:19:07.099 scaled typesetting. The calculator scaled arithmetic. 00:19:08.119 --> 00:19:10.640 VisiCalc

scaled financial modeling. Okay, I'm 00:19:10.640 --> 00:19:13.460 tracking. If VisiCalc was revolutionary because 00:19:13.460 --> 00:19:16.500 it instantly recalculated numbers, a Grok bot 00:19:16.500 --> 00:19:19.279 is revolutionary because it instantly recalculates 00:19:19.279 --> 00:19:24.559 the entire day. Recalculating the day. I want 00:19:24.559 --> 00:19:25.980 to sit with that phrasing for a second because 00:19:25.980 --> 00:19:27.700 we all know that feeling. You open your laptop 00:19:27.700 --> 00:19:31.460 at 8 a .m. You look at your calendar. And you 00:19:31.460 --> 00:19:34.339 see an eight -hour stretch of pure sequence work. 00:19:34.559 --> 00:19:37.099 Just endless tasks. You have to triage 50 emails, 00:19:37.420 --> 00:19:39.619 pull data from three different SaaS platforms 00:19:39.619 --> 00:19:42.680 for a weekly status report, summarize a 40 -page 00:19:42.680 --> 00:19:45.240 PDF research brief, prep an agenda for a 2 p 00:19:45.240 --> 00:19:48.539 .m. meeting. It's a slog. It is the modern equivalent 00:19:48.539 --> 00:19:52.339 of erasing ledger cells. It's moving digital 00:19:52.339 --> 00:19:54.140 paper from one side of the desk to the other. 00:19:54.579 --> 00:19:57.720 Precisely. And what is happening right now in 00:19:57.720 --> 00:20:00.180 the Grokbot era... is that an eight -hour stretch 00:20:00.180 --> 00:20:02.980 of that specific type of manual sequence work 00:20:02.980 --> 00:20:05.799 can now finish before a cup of coffee cools. 00:20:06.079 --> 00:20:10.410 Before a cup of coffee cools. Assuming an autonomous 00:20:10.410 --> 00:20:14.069 agent is executing the sequence. Correct. But 00:20:14.069 --> 00:20:17.109 let's clarify the mechanics here, because people 00:20:17.109 --> 00:20:19.789 throw around terms like AI and bot very loosely. 00:20:20.009 --> 00:20:23.670 When we say grokbot, what exactly are we talking 00:20:23.670 --> 00:20:25.269 about technically? We aren't just talking about 00:20:25.269 --> 00:20:27.609 a chat window where you type a question and get 00:20:27.609 --> 00:20:30.180 a text response, right? No, that is a crucial 00:20:30.180 --> 00:20:32.920 distinction. A chatbot is reactive. It waits 00:20:32.920 --> 00:20:35.259 for a prompt, gives an answer, and stops. Right. 00:20:35.359 --> 00:20:38.700 A grokbot, in this context, refers to an autonomous 00:20:38.700 --> 00:20:41.720 AI agent. It is a large language model equipped 00:20:41.720 --> 00:20:44.559 with memory, loop logic, and access to external 00:20:44.559 --> 00:20:47.440 tools. So it can actually do things. Exactly. 00:20:47.519 --> 00:20:50.099 You give it an objective, like compile the weekly 00:20:50.099 --> 00:20:52.420 sales report, and it breaks that objective down 00:20:52.420 --> 00:20:54.559 into a sequence of steps. It thinks through the 00:20:54.559 --> 00:20:57.319 process. It navigates to the CRM, pulls the data, 00:20:57.480 --> 00:20:59.680 opens a spreadsheet, formats the numbers, drafts

00:20:59.680 --> 00:21:01.519 a summary based on the data, and prepares an 00:21:01.519 --> 00:21:04.460 email draft. It operates on a loop. Reason, act, 00:21:04.680 --> 00:21:07.460 observe the result, and adjust. It is executing 00:21:07.460 --> 00:21:10.660 the sequence autonomously. Okay, but as powerful 00:21:10.660 --> 00:21:13.130 as that is... It's very easy to get swept away 00:21:13.130 --> 00:21:14.890 in the hype and assume these bots are just going 00:21:14.890 --> 00:21:17.109 to replace every single human function by next 00:21:17.109 --> 00:21:19.269 Tuesday. Right, which isn't true. The sources 00:21:19.269 --> 00:21:21.609 are actually very careful to highlight the boundaries. 00:21:22.130 --> 00:21:24.269 Let's talk about the so what of these limits, 00:21:24.329 --> 00:21:26.589 because understanding what a Grok bot cannot 00:21:26.589 --> 00:21:29.170 do is just as important as understanding what 00:21:29.170 --> 00:21:32.289 it can. It is essential to be precise about the 00:21:32.289 --> 00:21:35.890 boundaries. First, Grok bots cannot handle unscripted 00:21:35.890 --> 00:21:38.589 physical work in the real world. They are digital 00:21:38.589 --> 00:21:41.049 entities. Sure. But more importantly for the 00:21:41.049 --> 00:21:43.509 knowledge worker, they cannot execute judgment 00:21:43.509 --> 00:21:46.609 that relies on unique, unshared context. What 00:21:46.609 --> 00:21:49.329 do you mean by unshared context? Think about 00:21:49.329 --> 00:21:51.670 being in a meeting. A bot can transcribe the 00:21:51.670 --> 00:21:54.589 meeting perfectly, but it cannot read the room. 00:21:54.750 --> 00:21:57.230 Oh, I see. It doesn't know that the VP of sales 00:21:57.230 --> 00:21:59.869 is wincing because of an unspoken political rivalry 00:21:59.869 --> 00:22:02.289 with the marketing director. It doesn't know 00:22:02.289 --> 00:22:04.690 the unwritten cultural dynamics of your specific 00:22:04.690 --> 00:22:07.980 office. It lacks the human intuition of context. 00:22:08.299 --> 00:22:12.319 Right. And most crucially, a grokbot cannot and 00:22:12.319 --> 00:22:15.359 should not make decisions with real irreversible 00:22:15.359 --> 00:22:18.000 human stakes. Like what? Firing an employee, 00:22:18.299 --> 00:22:21.259 finalizing a massive financial merger, deciding 00:22:21.259 --> 00:22:23.980 the strategic direction of a company. That domain 00:22:23.980 --> 00:22:27.329 remains purely human. Going back to our analogy, 00:22:27.630 --> 00:22:29.930 the autonomous agent is the sous chef doing the 00:22:29.930 --> 00:22:32.950 chopping. The human remains the head chef deciding 00:22:32.950 --> 00:22:35.490 what goes on the menu. The sequence is handed 00:22:35.490 --> 00:22:39.240 off. The judgment stays. So we have the theory. 00:22:39.730 --> 00:22:42.799 We understand the historical pattern. And we 00:22:42.799 --> 00:22:44.740 understand the mechanical definition of the AI 00:22:44.740 --> 00:22:47.740 agent. But how is this actually being deployed 00:22:47.740 --> 00:22:50.519 in the real world right this second? It's happening 00:22:50.519 --> 00:22:52.660 fast. Because if I'm a listener sitting at a 00:22:52.660 --> 00:22:55.319 mid -level management job, this might still sound 00:22:55.319 --> 00:22:57.700 like science fiction. Is this actually happening 00:22:57.700 --> 00:22:59.900 in corporate environments today? It absolutely 00:22:59.900 --> 00:23:04.180 is. Brian Roy has been conducting extensive on 00:23:04.180 --> 00:23:06.759 -the -ground fieldwork on this exact transition. 00:23:07.039 --> 00:23:08.900 What's he saying? He's been spending time embedded 00:23:08.900 --> 00:23:11.279 with clients at large public companies, walking 00:23:11.279 --> 00:23:13.160 them through this operational shift department 00:23:13.160 --> 00:23:16.059 by department. And the paradigm he is introducing 00:23:16.059 --> 00:23:19.380 to them is staggering. Okay. He is showing them 00:23:19.380 --> 00:23:22.259 how to arm single, individual employees with 00:23:22.259 --> 00:23:25.319 an almost unlimited number of Grokbot employees. 00:23:25.680 --> 00:23:28.980 It's the ultimate leverage. It's like every single 00:23:28.980 --> 00:23:30.980 mid -level employee suddenly gets promoted to 00:23:30.980 --> 00:23:33.700 a department head overnight. Yes. And they are 00:23:33.700 --> 00:23:36.720 handed a staff of 100 tireless interns who never 00:23:36.720 --> 00:23:40.119 sleep, never complain, process information instantly, 00:23:40.279 --> 00:23:43.640 and never make a copy -paste error. You can imagine 00:23:43.640 --> 00:23:45.619 what happens to the competitive landscape of 00:23:45.619 --> 00:23:48.299 an industry when that happens. One person armed 00:23:48.299 --> 00:23:51.680 with an AI army is incredibly powerful. Unbeatable. 00:23:51.950 --> 00:23:54.630 But an entire legacy company armed with those 00:23:54.630 --> 00:23:57.210 armies becomes a juggernaut that is almost impossible 00:23:57.210 --> 00:24:00.049 to compete against using traditional manual labor 00:24:00.049 --> 00:24:02.809 models. But wait, let's look at the IT and security 00:24:02.809 --> 00:24:06.630 reality of this. If everyone in a 5 ,000 -person 00:24:06.630 --> 00:24:08.910 company suddenly has hundreds of AI agents running 00:24:08.910 --> 00:24:12.029 around doing tasks, accessing databases and drafting 00:24:12.029 --> 00:24:15.049 emails, how is that not just total unmitigated 00:24:15.049 --> 00:24:18.009 chaos? It could be. I mean, an army of interns 00:24:18.009 --> 00:24:20.289 could do an unbelievable amount of damage if 00:24:20.289 --> 00:24:22.509 they are all firing off emails to clients and 00:24:22.509 --> 00:24:25.150 deleting files without oversight. That is the 00:24:25.150 --> 00:24:28.089 exact bottleneck Brian identified, and it is 00:24:28.089 --> 00:24:31.089 where he introduces a crucial operational concept, 00:24:31.349 --> 00:24:35.109 the filter strategy.

The filter strategy. He 00:24:35.109 --> 00:24:37.410 highlights a legacy public company that is currently 00:24:37.410 --> 00:24:40.789 going all in on deploying this AI fleet, but 00:24:40.789 --> 00:24:43.589 they aren't just unleashing bots blindly onto 00:24:43.589 --> 00:24:46.539 their servers. Good. Every single Grokbot command, 00:24:46.740 --> 00:24:49.519 every action an agent attempts to take runs through 00:24:49.519 --> 00:24:52.660 a centralized audit trail governed by a strict 00:24:52.660 --> 00:24:56.079 traffic light filter system. A red, yellow, and 00:24:56.079 --> 00:24:58.500 green light system? How does that work mechanically? 00:24:58.720 --> 00:25:01.220 In a corporate IT environment, a red light means 00:25:01.220 --> 00:25:04.039 the bot physically lacks the API permission to 00:25:04.039 --> 00:25:06.339 execute a destructive action. Okay, give me an 00:25:06.339 --> 00:25:09.440 example. If a bot decides, logically, that the 00:25:09.440 --> 00:25:11.519 best way to clean up a hard drive is to delete 00:25:11.519 --> 00:25:14.150 a master customer database, The system throws 00:25:14.150 --> 00:25:16.809 a red light. The action is hard blocked. Makes 00:25:16.809 --> 00:25:19.569 sense. What about yellow? Yellow light means 00:25:19.569 --> 00:25:22.250 the bot has prepared the action, but it requires 00:25:22.250 --> 00:25:25.349 human authorization before executing. So it pauses. 00:25:25.690 --> 00:25:28.170 Exactly. For example, a bot can read an incoming 00:25:28.170 --> 00:25:31.430 client email, access the CRM, draft a complex 00:25:31.430 --> 00:25:34.009 response, and attach a finalized contract. But 00:25:34.009 --> 00:25:37.529 it cannot hit send. Accuse the draft. It throws 00:25:37.529 --> 00:25:39.410 a yellow light and waits for the human manager 00:25:39.410 --> 00:25:42.069 to review it, tweak it, and click approve. And 00:25:42.069 --> 00:25:44.319 green light. Greenlight means the bot is pre 00:25:44.319 --> 00:25:46.700 -approved to run autonomously in a safe, constrained 00:25:46.700 --> 00:25:49.559 environment. Sorting incoming vendor invoices 00:25:49.559 --> 00:25:52.339 into a specific SharePoint folder, parsing data 00:25:52.339 --> 00:25:55.220 from PDFs into an internal spreadsheet, these 00:25:55.220 --> 00:25:58.529 run on green. That makes total sense. Brian explicitly 00:25:58.529 --> 00:26:00.769 notes that this filter system is the defining 00:26:00.769 --> 00:26:03.329 difference between an army and a mob. An army 00:26:03.329 --> 00:26:06.190 follows orders. A mob just breaks things. The 00:26:06.190 --> 00:26:08.670 human managers are still very much there, but 00:26:08.670 --> 00:26:10.960 their day -to -day reality has shifted. They 00:26:10.960 --> 00:26:13.240 are now overseeing the work. They're managing 00:26:13.240 --> 00:26:16.200 the filter, reviewing the yellow lights, rather 00:26:16.200 --> 00:26:17.980 than performing the manual sequence themselves. 00:26:18.359 -->

00:26:21.119 We actually have concrete case studies from Brian's 00:26:21.119 --> 00:26:23.779 own fleet of deployed bots to prove out the economics 00:26:23.779 --> 00:26:26.460 of this. Oh, like the print system example? Yes. 00:26:26.539 --> 00:26:30.299 On October 10, 2026, he detailed a scenario where 00:26:30.299 --> 00:26:32.819 a Grok bot was put in full control of a physical 00:26:32.819 --> 00:26:35.240 print system. Right, the package printing operation. 00:26:35.559 --> 00:26:38.240 And this wasn't just generating text. This was 00:26:38.240 --> 00:26:41.009 an agent controlling a physical workflow. Very 00:26:41.009 --> 00:26:43.329 different from writing an email. The bot coordinated 00:26:43.329 --> 00:26:46.230 local AI models to run a massive printing operation 00:26:46.230 --> 00:26:49.630 for three solid hours, flawlessly. And the result 00:26:49.630 --> 00:26:54.710 was a 58 % savings in costs. Huge savings. But 00:26:54.710 --> 00:26:57.509 what exactly was the bot doing that a human operator 00:26:57.509 --> 00:26:59.809 couldn't? It comes down to dynamic optimization. 00:27:00.210 --> 00:27:03.250 A human operator usually batches print jobs based 00:27:03.250 --> 00:27:06.809 on simple rules, maybe by paper size or color. 00:27:06.990 --> 00:27:09.630 Right. But the Grok bot was able to ingest the 00:27:09.630 --> 00:27:12.789 entire queue of hundreds of print jobs and dynamically 00:27:12.789 --> 00:27:15.730 optimize the routing in real time. It optimized 00:27:15.730 --> 00:27:18.430 ink flow, paper routing, and job batching simultaneously, 00:27:18.910 --> 00:27:21.869 reacting to millisecond delays in the printers. 00:27:22.150 --> 00:27:24.789 A human brain simply cannot compute those variables 00:27:24.789 --> 00:27:27.470 fast enough. The bot optimized the sequence, 00:27:27.710 --> 00:27:31.029 delivering the 58 % savings. And then there is 00:27:31.029 --> 00:27:34.730 the example from October 28, 2026, which Brian 00:27:34.730 --> 00:27:38.250 famously refers to as the 3 .27 a .m. disagreement. 00:27:38.710 --> 00:27:42.069 I love this example. This perfectly illustrates 00:27:42.069 --> 00:27:44.349 the yellow light concept and the boundary between 00:27:44.349 --> 00:27:47.170 sequence and judgment. I found this deeply fascinating 00:27:47.170 --> 00:27:49.809 because it makes the AI feel much more collaborative. 00:27:50.230 --> 00:27:53.150 His bots were working autonomously on a complex 00:27:53.150 --> 00:27:55.750 task overnight, and they ran into a conflict 00:27:55.750 --> 00:27:58.549 in the data. They hit a wall. Now, a poorly designed 00:27:58.549 --> 00:28:00.869 automation script would just crash, or worse, 00:28:01.009 --> 00:28:04.009 guess and make a terrible mistake. Which happens 00:28:04.009 --> 00:28:07.410 all the time with legacy software. Instead, the 00:28:07.410 --> 00:28:10.849 bots paused. They texted Brian a single, perfectly 00:28:10.849 --> 00:28:13.450 crafted message in the middle

of the night, presenting 00:28:13.450 --> 00:28:18.049 clear choices. A, B, C, and D. If we connect 00:28:18.049 --> 00:28:20.069 this to the bigger picture of what work looks 00:28:20.069 --> 00:28:23.210 like in the 2020s, this is the ultimate proof 00:28:23.210 --> 00:28:26.589 of the human retaining the judgment. The bots 00:28:26.589 --> 00:28:28.809 did all the grueling sequence work. They assembled 00:28:28.809 --> 00:28:31.210 the facts, cross -referenced the data, structured 00:28:31.210 --> 00:28:34.170 the problem, and isolated the variables. They 00:28:34.170 --> 00:28:36.089 handed the human a multiple choice question. 00:28:36.630 --> 00:28:38.930 Brian didn't have to do the research. He just 00:28:38.930 --> 00:28:41.329 had to wake up, look at his phone, apply his 00:28:41.329 --> 00:28:44.509 unique human judgment, pick B, and go back to 00:28:44.509 --> 00:28:47.170 sleep while the bots resumed the sequence. It 00:28:47.170 --> 00:28:49.569 makes you wonder what the absolute limit of this 00:28:49.569 --> 00:28:52.529 is. If a human only has to answer multiple choice 00:28:52.529 --> 00:28:55.690 questions, how small can a company get and still 00:28:55.690 --> 00:28:58.730 operate like a massive corporation? Brian tested 00:28:58.730 --> 00:29:01.170 the absolute limit of that exact question with 00:29:01.170 --> 00:29:04.769 his zero human company concept. This is where 00:29:04.769 --> 00:29:07.470 things get truly mind bending. He built a test 00:29:07.470 --> 00:29:10.130 company run entirely by AI platforms. Entirely. 00:29:10.150 --> 00:29:12.970 He used discarded data from a long bankrupt company, 00:29:13.089 --> 00:29:15.750 essentially their old emails, operational manuals 00:29:15.750 --> 00:29:17.990 and customer data, and fed it into a network 00:29:17.990 --> 00:29:20.569 of autonomous agents. What was the goal? The 00:29:20.569 --> 00:29:22.329 goal was to see if the company could function, 00:29:22.410 --> 00:29:25.190 interact and perform tasks with almost zero human 00:29:25.190 --> 00:29:28.130 oversight. And it did. It proved out what he 00:29:28.130 --> 00:29:31.319 calls the coming agent -to -agent economy. We 00:29:31.319 --> 00:29:33.460 are moving toward a world where your company's 00:29:33.460 --> 00:29:36.220 AI agent negotiates a contract with your vendor's 00:29:36.220 --> 00:29:38.799 AI agent. They settle the terms, execute the 00:29:38.799 --> 00:29:41.839 transaction, and simply log a summary for the 00:29:41.839 --> 00:29:44.710 humans. The agents are actually becoming the 00:29:44.710 --> 00:29:47.750 primary actors in the sequence of commerce. It's 00:29:47.750 --> 00:29:49.369 a fundamental shift in how business is done. 00:29:49.730 --> 00:29:52.549 Okay, that is incredible on a macro scale, but 00:29:52.549 --> 00:29:54.450 let's bring it back down to earth, right to the 00:29:54.450 --> 00:29:57.549 listener's desk today. Let's do it. We know the 00:29:57.549 --> 00:30:00.369 historical

theory. We've seen the massive corporate 00:30:00.369 --> 00:30:03.609 deployment. But how does an individual actually 00:30:03.609 --> 00:30:06.769 implement this? It's a great question. If I'm 00:30:06.769 --> 00:30:08.829 a marketing director or a small business owner, 00:30:08.970 --> 00:30:11.529 I don't know how to code an autonomous reasoning 00:30:11.529 --> 00:30:14.970 loop. Where do these grokbots even come from? 00:30:15.109 --> 00:30:17.529 Am I just locked out of this revolution unless 00:30:17.529 --> 00:30:20.190 I hire a software engineer? That is the exact 00:30:20.190 --> 00:30:22.490 bottleneck the Read Multiplex project addresses, 00:30:22.670 --> 00:30:25.650 and it brings us directly to the toolkit, specifically 00:30:25.650 --> 00:30:29.119 a resource called grokbot .rocks. Let's talk 00:30:29.119 --> 00:30:31.740 about the sheer scale of this directory. As of 00:30:31.740 --> 00:30:34.859 late October 2026, this directory tracks over 00:30:34.859 --> 00:30:40.279 7 ,539 public Grokbot templates across 26 different 00:30:40.279 --> 00:30:43.940 categories, and it is updated weekly. It is essentially 00:30:43.940 --> 00:30:46.819 the new App Store. But the sources note that 00:30:46.819 --> 00:30:49.160 this marks the end of software being sold as 00:30:49.160 --> 00:30:51.740 rent. You aren't paying a monthly subscription 00:30:51.740 --> 00:30:54.539 for a SaaS tool to do these things anymore. You 00:30:54.539 --> 00:30:57.650 are deploying an agent to do it. To explain how 00:30:57.650 --> 00:30:59.650 a non -technical person approaches a directory 00:30:59.650 --> 00:31:03.509 of 7 ,500 bots, Brian uses a brilliant analogy 00:31:03.509 --> 00:31:07.170 from the 1950s. The paint -by -numbers kit. No, 00:31:07.190 --> 00:31:09.170 I like this analogy. Think about what happened 00:31:09.170 --> 00:31:11.589 when those kits were introduced. Millions of 00:31:11.589 --> 00:31:13.630 people who had never held a paintbrush in their 00:31:13.630 --> 00:31:15.970 lives, who had no formal training in composition 00:31:15.970 --> 00:31:19.990 or color theory, suddenly became painters. Why? 00:31:20.470 --> 00:31:22.750 Because they started by filling in numbered outlines. 00:31:23.150 --> 00:31:24.890 They didn't have to face the terror of the blank 00:31:24.890 --> 00:31:27.890 canvas. The grokbot .rocks directory is that 00:31:27.890 --> 00:31:31.009 numbered outline. You use these thousands of 00:31:31.009 --> 00:31:33.730 existing pre -built templates to learn the pattern. 00:31:34.009 --> 00:31:36.029 You deploy a bot someone else built to handle 00:31:36.029 --> 00:31:37.950 your calendar and you watch how it works. You 00:31:37.950 --> 00:31:39.849 color inside the lines. And once you understand 00:31:39.849 --> 00:31:42.410 the mechanism of how an agent operates, then 00:31:42.410 --> 00:31:44.490 you can step up to the blank canvas and begin 00:31:44.490 --> 00:31:46.710 building your own custom bots. But the way you 00:31:46.710 --> 00:31:48.869 interact with these bots is different. You don't 00:31:48.869 --> 00:31:50.549 even have to hunt through the directory manually 00:31:50.549 --> 00:31:53.109 to find what you need. This is where we get into 00:31:53.109 --> 00:31:55.190 a concept Brian calls the superpower prompt. 00:31:55.660 --> 00:31:58.039 And we need to define this because it isn't just 00:31:58.039 --> 00:32:00.980 typing a question into a chat box. Right. A super 00:32:00.980 --> 00:32:03.039 prompt is fundamentally different from a standard 00:32:03.039 --> 00:32:05.380 query. When people use a basic chat bot, they 00:32:05.380 --> 00:32:07.359 give it a task like, write an email to my boss. 00:32:07.500 --> 00:32:10.140 That relies entirely on the AI's generic default 00:32:10.140 --> 00:32:13.359 training. Which is usually pretty mediocre. Exactly. 00:32:13.680 --> 00:32:16.160 A super prompt, as Brian defines it, delivers 00:32:16.160 --> 00:32:20.009 a program and a map. A program and a map. You 00:32:20.009 --> 00:32:22.869 are injecting a multilayered set of logical constraints, 00:32:23.069 --> 00:32:25.670 a cognitive framework that teaches the AI how 00:32:25.670 --> 00:32:28.490 to behave before it even attempts the task. So 00:32:28.490 --> 00:32:30.690 it's teaching it the rules of the game. It's 00:32:30.690 --> 00:32:32.769 the difference between handing someone a wrench 00:32:32.769 --> 00:32:34.990 and handing them a wrench along with a master 00:32:34.990 --> 00:32:37.130 class in plumbing and a strict set of safety 00:32:37.130 --> 00:32:39.710 regulations. So how does that work practically 00:32:39.710 --> 00:32:42.269 with a directory? You deploy a super prompt by 00:32:42.269 --> 00:32:44.589 pointing your Grokbot directly at the grokbot 00:32:44.589 --> 00:32:48.839 .rocks URL. You explicitly instruct it to study 00:32:48.839 --> 00:32:50.819 the skill sets that are already established in 00:32:50.819 --> 00:32:53.339 those 7 ,500 templates. You tell it to match 00:32:53.339 --> 00:32:55.859 those existing proven skills to the specific 00:32:55.859 --> 00:32:58.019 goals you want to achieve for your business and 00:32:58.019 --> 00:33:00.440 then integrate those working recipes into its 00:33:00.440 --> 00:33:02.779 own behavior. You're essentially telling your 00:33:02.779 --> 00:33:05.299 AI to go read the instruction manuals of thousands 00:33:05.299 --> 00:33:08.319 of other AI agents and teach itself how to do 00:33:08.319 --> 00:33:10.819 the job you need done. Yes, but, and this is 00:33:10.819 --> 00:33:14.019 vital, with strict safety rules included directly 00:33:14.019 --> 00:33:16.220 in the prompt. You build the yellow light system 00:33:16.220 --> 00:33:18.539 into the agent's instructions. You explicitly 00:33:18.539 --> 00:33:22.619 tell it, never send, post, buy, or delete anything 00:33:22.619 --> 00:33:25.220 without asking me first. The person who used 00:33:25.220 --> 00:33:27.440 to spend their

morning deciding which manual 00:33:27.440 --> 00:33:30.589 sequence of clicks to perform, opening the email, 00:33:30.769 --> 00:33:33.930 copying the data, pasting it into the CRM, now 00:33:33.930 --> 00:33:35.630 spends their morning deciding which high -level 00:33:35.630 --> 00:33:37.930 results to pursue, while the bot handles the 00:33:37.930 --> 00:33:40.890 clicks safely. Let's make this incredibly concrete. 00:33:41.369 --> 00:33:44.250 The sources map out 10 specific handoffs, 10 00:33:44.250 --> 00:33:47.609 manual sequences that you can hand off to a Grok 00:33:47.609 --> 00:33:50.069 bot today. Let's walk through them. I want to 00:33:50.069 --> 00:33:52.730 look closely at a few of these because it paints 00:33:52.730 --> 00:33:55.190 a vivid picture of how our days are about to 00:33:55.190 --> 00:33:58.569 change. Let's look at email triage. This is perhaps 00:33:58.569 --> 00:34:00.710 the most universal pain point for knowledge workers. 00:34:01.009 --> 00:34:04.190 Think about how you handle email right now. You 00:34:04.190 --> 00:34:07.849 open an inbox with 150 unread messages. You manually 00:34:07.849 --> 00:34:10.070 scan the subject lines, trying to separate the 00:34:10.070 --> 00:34:12.210 newsletters from the internal company memos from 00:34:12.210 --> 00:34:15.050 the urgent client requests. It takes forever. 00:34:15.329 --> 00:34:18.289 You open a client email, read the history, open 00:34:18.289 --> 00:34:20.429 your calendar to check your availability, draft 00:34:20.429 --> 00:34:23.849 a response, and hit send. It takes hours. If 00:34:23.849 --> 00:34:25.610 you hand this off to a Grok bot, the mechanism 00:34:25.610 --> 00:34:28.570 is entirely different. The bot is connected to 00:34:28.570 --> 00:34:31.969 your inbox via API. It reads the 150 emails. 00:34:32.360 --> 00:34:35.199 Instantly. Instantly. It uses its semantic understanding 00:34:35.199 --> 00:34:37.980 to categorize them. It archives the newsletters. 00:34:37.980 --> 00:34:40.320 It highlights the three urgent client emails. 00:34:40.519 --> 00:34:43.659 It accesses your calendar, drafts three perfect 00:34:43.659 --> 00:34:46.380 responses offering available meeting times, and 00:34:46.380 --> 00:34:49.059 leaves them in your drafts folder. You wake up, 00:34:49.079 --> 00:34:51.519 review the drafts in three minutes, hit approve, 00:34:51.699 --> 00:34:54.460 and you just saved two hours of your life. And 00:34:54.460 --> 00:34:56.880 we see a similar transformation with report assembly. 00:34:57.260 --> 00:34:59.469 No, report assembly is the worst. In the old 00:34:59.469 --> 00:35:01.829 sequence, an operations manager might spend every 00:35:01.829 --> 00:35:03.969 Friday afternoon pulling numbers from Salesforce, 00:35:04.309 --> 00:35:07.269 downloading CSV files from a marketing platform, 00:35:07.389 --> 00:35:09.809 and manually copying and pasting those metrics 00:35:09.809 --> 00:35:12.269 into a PowerPoint slide deck for Monday's executive 00:35:12.269 --> 00:35:14.889 meeting. Such a waste of human potential. When 00:35:14.889 --> 00:35:17.309 handed off to a grokbot, the agent logs into 00:35:17.309 --> 00:35:20.070 the platforms using secure credentials, extracts 00:35:20.070 --> 00:35:23.130 the specific KPIs, formats them into the exact 00:35:23.130 --> 00:35:25.150 PowerPoint template you've provided, and even 00:35:25.150 --> 00:35:27.150 writes a brief executive summary highlighting 00:35:27.150 --> 00:35:30.329 anomalies in the data. The human manager doesn't 00:35:30.329 --> 00:35:33.369 build the report. They review the insights generated 00:35:33.369 --> 00:35:37.289 by the report. And the sources list several others 00:35:37.289 --> 00:35:40.369 that follow this exact pattern. Summarizing dense 00:35:40.369 --> 00:35:43.929 research briefs. Managing local file inventories. 00:35:44.389 --> 00:35:46.969 Monitoring specific industry topics on social 00:35:46.969 --> 00:35:49.769 platforms like X. Drafting content in your own 00:35:49.769 --> 00:35:52.329 established voice. Checking vendor invoices for 00:35:52.329 --> 00:35:55.090 mathematical errors before payment. summarizing 00:35:55.090 --> 00:35:58.010 tax and legal paperwork, and tracking project 00:35:58.010 --> 00:36:00.389 status across multiple software tools like Jira 00:36:00.389 --> 00:36:03.489 or Asana. Every single one of these is a job 00:36:03.489 --> 00:36:06.130 someone in an office is doing by hand this week. 00:36:06.329 --> 00:36:08.369 And every one of them can finish before your 00:36:08.369 --> 00:36:10.650 coffee cools if handed off to an agent. If you 00:36:10.650 --> 00:36:12.909 are currently doing these by hand, you have to 00:36:12.909 --> 00:36:16.510 ask yourself why. You are walking the dusty library 00:36:16.510 --> 00:36:19.590 stacks when the search engine is right there 00:36:19.590 --> 00:36:22.409 in front of you. You are attached to the sequence. 00:36:23.480 --> 00:36:25.619 you are mistaking the friction of the copy and 00:36:25.619 --> 00:36:28.440 paste for the value of the final report. Exactly. 00:36:28.739 --> 00:36:31.900 Now, all of these tools, these frameworks, these 00:36:31.900 --> 00:36:34.780 super prompts sound absolutely incredible, but 00:36:34.780 --> 00:36:37.460 they obviously require a highly specific map 00:36:37.460 --> 00:36:40.800 to navigate safely and effectively. You can't 00:36:40.800 --> 00:36:42.800 just stumble into this. You can't just blindly 00:36:42.800 --> 00:36:45.519 stumble into an agent -to -agent economy without 00:36:45.519 --> 00:36:48.079 a guide, or you will end up causing chaos in 00:36:48.079 --> 00:36:50.559 your organization. Right. Here's where it gets 00:36:50.559 --> 00:36:52.969 really interesting. For those trying to figure 00:36:52.969 --> 00:36:55.510 out how to actually deploy this without breaking 00:36:55.510 --> 00:36:58.489 their company's IT protocols, the source of our 00:36:58.489 -->

00:37:01.030 deep dive today becomes incredibly relevant. 00:37:01.349 --> 00:37:04.710 Right. Brian Rommel has published a massive operational 00:37:04.710 --> 00:37:07.630 playbook on his Read Multiplex platform. This 00:37:07.630 --> 00:37:10.429 isn't just theoretical musing. It is an actionable 00:37:10.429 --> 00:37:12.690 blueprint for deployment that sits behind the 00:37:12.690 --> 00:37:15.369 members line. For Read Multiplex members, you 00:37:15.369 --> 00:37:18.230 get access to the exact master prompt Brian uses. 00:37:18.920 --> 00:37:21.300 The super prompt with all the complex safety 00:37:21.300 --> 00:37:23.760 rules and cognitive frameworks already built 00:37:23.760 --> 00:37:26.320 in. You don't have to guess how to code the yellow 00:37:26.320 --> 00:37:28.840 light system. He gives you the text. You get 00:37:28.840 --> 00:37:31.199 10 ready -to -paste prompts specifically designed 00:37:31.199 --> 00:37:33.420 for those exact handoffs we just discussed, like 00:37:33.420 --> 00:37:36.599 email triage and report assembly. The operational 00:37:36.599 --> 00:37:39.780 value is significant. The platform includes a 00:37:39.780 --> 00:37:42.199 comprehensive seven -step playbook to help you 00:37:42.199 --> 00:37:44.760 hand off your very first sequence this week safely. 00:37:45.260 --> 00:37:47.099 And for larger organizations trying to scale 00:37:47.099 --> 00:37:49.699 this, there is a detailed department -by -department 00:37:49.699 --> 00:37:52.159 handoff map covering sales, support, finance, 00:37:52.360 --> 00:37:55.280 operations, marketing, and HR. It maps out exactly 00:37:55.280 --> 00:37:57.800 which tasks should have red, yellow, and green 00:37:57.800 --> 00:38:00.199 light permissions for each specific domain. And 00:38:00.199 --> 00:38:02.179 then there is the insider guide at the bottom 00:38:02.179 --> 00:38:05.079 of grokbot .rocks, which is constantly updated. 00:38:05.440 --> 00:38:09.500 Members get Brian's 12 top bot picks and the 00:38:09.500 --> 00:38:11.780 nine best bot stacks, meaning which specific 00:38:11.780 --> 00:38:14.130 bots to chain together in a sequence. Whether 00:38:14.130 --> 00:38:16.010 you're running a small business, operating as 00:38:16.010 --> 00:38:17.949 a solo creator, or just trying to manage a busy 00:38:17.949 --> 00:38:20.829 household with privacy in mind. There are 62 00:38:20.829 --> 00:38:23.929 practical tips from actual AI builders, a five 00:38:23.929 --> 00:38:25.869 -point safety checklist to run before adding 00:38:25.869 --> 00:38:28.849 any community bot to your system, early access 00:38:28.849 --> 00:38:30.949 to Brian's newest bots before they hit the public 00:38:30.949 --> 00:38:34.329 list, and even 25 % off custom consulting work 00:38:34.329 --> 00:38:36.969 with him. It is important to understand the larger 00:38:36.969 --> 00:38:39.449 philosophy driving this research. Brian Rommel 00:38:39.449 --> 00:38:41.369 isn't just building these tools for the sake 00:38:41.369 --> 00:38:43.250 of corporate efficiency. What's his ultimate 00:38:43.250 --> 00:38:47.329 goal? He views this moment as documenting humanity's 00:38:47.329 --> 00:38:50.050 greatest technological leap. The mission is to 00:38:50.050 --> 00:38:52.389 prepare individuals and businesses wisely for 00:38:52.389 --> 00:38:55.869 what he calls a multi -planetary future. A multi 00:38:55.869 --> 00:38:59.090 -planetary future? He is analyzing and sharing 00:38:59.090 --> 00:39:01.489 these hard -won lessons so that people aren't 00:39:01.489 --> 00:39:03.960 left behind like... the hand typesetters or the 00:39:03.960 --> 00:39:07.380 slide rule engineers. He wants people to survive 00:39:07.380 --> 00:39:09.659 the fourth revolution. And this brings us to 00:39:09.659 --> 00:39:12.750 an important point. This is the Read Multiplex 00:39:12.750 --> 00:39:15.429 podcast. Be sure to support Brian and his work 00:39:15.429 --> 00:39:17.230 and become a member to see the rest of the story. 00:39:17.510 --> 00:39:21.230 It's vital work. This is entirely independent 00:39:21.230 --> 00:39:24.030 research. There's no massive tech conglomerate 00:39:24.030 --> 00:39:26.349 or AI company secretly funding this field work 00:39:26.349 --> 00:39:29.429 behind the scenes. If this conversation has ignited 00:39:29.429 --> 00:39:31.389 your imagination about what is possible for your 00:39:31.389 --> 00:39:34.809 career, your business, or your daily life, it's 00:39:34.809 --> 00:39:36.690 highly recommended to support the person mapping 00:39:36.690 --> 00:39:39.550 the territory. It's about value for value. It 00:39:39.550 --> 00:39:42.429 is the only way independent, on -the -ground 00:39:42.429 --> 00:39:45.610 fieldwork of this depth can continue in an industry 00:39:45.610 --> 00:39:49.050 dominated by massive tech giants. So, to synthesize 00:39:49.050 --> 00:39:50.869 all of this, what is the core takeaway for the 00:39:50.869 --> 00:39:53.510 listener? The core lesson we have to take away 00:39:53.510 --> 00:39:55.849 is that our digital tools are no longer just 00:39:55.849 --> 00:39:58.429 passive calculators waiting for us to press buttons. 00:39:58.849 --> 00:40:01.269 They are no longer just screens we stare at. 00:40:01.369 --> 00:40:04.610 They are tireless, autonomous armies waiting 00:40:04.610 --> 00:40:07.070 for direction. The only limiting factor left 00:40:07.070 --> 00:40:09.690 in the modern economy is our own emotional attachment 00:40:09.690 --> 00:40:12.710 to the old manual sequences of our days. Which 00:40:12.710 --> 00:40:14.849 leaves us with a final lingering question for 00:40:14.849 --> 00:40:17.489 you, the listener. Think about your desk right 00:40:17.489 --> 00:40:20.139 now. Think about the routine you're going to 00:40:20.139 --> 00:40:22.769 log into tomorrow morning. The sequence you are 00:40:22.769 --> 00:40:26.869 so used to. If a grokbot recalculates your entire 00:40:26.869 --> 00:40:29.190 workday tomorrow, stripping away every single 00:40:29.190 --> 00:40:31.750 mechanical step, all the email triage, all the 00:40:31.750 --> 00:40:34.389 data entry, all the report assembly. All the 00:40:34.389 --> 00:40:37.090 friction you've used to define a hard day's work. 00:40:37.429 --> 00:40:40.570 What does your judgment alone actually look like? 00:40:41.010 --> 00:40:43.650 When the comfortable, mindless sequence is entirely 00:40:43.650 --> 00:40:47.679 gone, are you ready to be the artist of your 00:40:47.679 --> 00:40:50.309 own career? That is the true challenge of the 00:40:50.309 --> 00:40:53.610 Grokbot era. It forces us to figure out what 00:40:53.610 --> 00:40:56.449 we are actually worth when the busy work is gone. 00:40:57.039 --> 00:40:59.219 Thank you for joining us on this deep dive into 00:40:59.219 --> 00:41:02.019 the Read Multiplex research. Stay curious, question 00:41:02.019 --> 00:41:16.780 your sequences, and don't get left behind. leads 00:41:16.780 --> 00:41:20.380 the way with research deep and true technology 00:41:20.440 --> 00:41:25.420 and history he connects them all for you Brian 00:41:25.610 --> 00:41:27.849 Romer leads the way with research deep and true 00:41:27.849 --> 00:41:30.309 technology and history he connects them all for 00:41:30.309 --> 00:41:32.809 you ancient sparks to modern code he traces every 00:41:32.809 --> 00:41:36.150 threat go exploration knowledge widely spread 00:41:36.150 --> 00:42:09.420 read multiplex go but human moves on love equation 00:42:09.420 --> 00:42:18.869 for AI zero human companies build Thank you. 00:42:21.530 --> 00:42:23.510 Thank you.