Multitasking was always a lie, AI made it more believable

00:00:00,600 --> 00:00:43,060 [Eric] [upbeat music] Welcome back to the Token Intelligence Show. AI is changing the way that we work, and on this show, we will walk you through the state of the art with AI, uh, cut through all the noise, and help you use wisdom to be a great leader and a great worker in this new era that we work in. And today, we're gonna talk about multitasking. This has been a, a subject that is, um, that is valuable and interesting in the realm of work for, I mean [laughs] really, as long- 00:00:43,060 --> 00:00:43,780 [John] For a long time. 00:00:43,780 --> 00:00:47,340 [Eric] [laughs] For a long time. I think that it became more relevant 00:00:48,380 --> 00:01:00,830 [Eric] in modern, uh, knowledge work, especially with digital interfaces, because you can run multiple applications at the same time, right? Even if you think about early computing, 00:01:02,080 --> 00:01:06,260 [Eric] you put a, a single program, physical program- 00:01:06,260 --> 00:01:07,060 [John] Right 00:01:07,060 --> 00:01:08,120 [Eric] ... into a computer. 00:01:09,260 --> 00:01:09,500 [Eric] And so- 00:01:09,500 --> 00:01:11,040 [John] Yeah, like on a, on a disk. 00:01:11,040 --> 00:01:12,920 [Eric] Exactly, on a disk. And so- 00:01:14,140 --> 00:01:14,850 [John] [laughs] 00:01:14,850 --> 00:01:27,890 [Eric] ... uh, you know, I'm sure that, that people remember, older listeners remember doing that, you know, where... Actually, my dad talks about going to college, and there was a computer for the entire campus, right? 00:01:27,890 --> 00:01:27,920 [John] Right. 00:01:27,920 --> 00:01:34,060 [Eric] And you would schedule time to go use the computer that was on the campus, and you had punch cards with the program on it. 00:01:34,060 --> 00:01:34,150 [John] Right. 00:01:34,150 --> 00:01:39,310 [Eric] Right. Um, and then some listeners will remember, uh, floppy disks, 00:01:40,660 --> 00:01:55,940 [Eric] uh, and then, of course, uh, compact disks or CDs. And so these are all, you know, physical programs. But what's interesting is, as computers became more capable and as interfaces evolved, you can run multiple applications at the same time. 00:01:55,940 --> 00:01:56,620 [John] Right. 00:01:56,620 --> 00:01:57,700 [Eric] And so this is- 00:01:57,700 --> 00:01:58,780 [John] Win- Windows. 00:01:58,780 --> 00:01:59,300 [Eric] Yes. 00:01:59,300 --> 00:02:00,220 [John] Right. [laughs] Like- 00:02:00,220 --> 00:02:01,400 [Eric] Multiple windows. 00:02:01,400 --> 00:02:01,460 [John] Yeah. 00:02:01,460 --> 00:02:01,920 [Eric] Exactly. 00:02:01,920 --> 00:02:07,480 [John] I mean, the operating system, right, like, that is the, one of the core innovations is multiple windows. 00:02:07,480 --> 00:02:07,920 [Eric] Yes. 00:02:07,920 --> 00:02:07,980 [John] Um... 00:02:07,980 --> 00:02:09,740 [Eric] Exactly. So 00:02:11,220 --> 00:02:18,740 [Eric] it became very relevant, or it became even more relevant in that environment just because of the interfaces that we use to do our work. 00:02:18,740 --> 00:02:18,829 [John] Right. 00:02:18,829 --> 00:02:19,480 [Eric] Right. And 00:02:20,820 --> 00:02:24,240 [Eric] I wanna dig into how AI is changing that. 00:02:24,240 --> 00:02:24,310 [John] Right. 00:02:24,310 --> 00:02:31,890 [Eric] But first, let's go back to, let's go back to the core of multitasking and what that means for us as humans- 00:02:31,890 --> 00:02:31,890 [John] Yeah 00:02:31,890 --> 00:02:33,460 [Eric] ... performing work in general. 00:02:33,460 --> 00:02:36,960 [John] Yeah, 'cause you just gave a nice overview of what it looks like with technology. 00:02:36,960 --> 00:02:37,040 [Eric] Yep. 00:02:37,040 --> 00:02:50,980 [John] But there's a human aspect of it that has nothing to do with technology, and I remember back to, like, when I f- really, when I first started working, um, ability to multitask was on a lot of job descriptions, and then it- 00:02:50,980 --> 00:02:51,340 [Eric] Yeah 00:02:51,340 --> 00:02:53,350 [John] ... and then it kind of went away. 00:02:53,350 --> 00:02:53,860 [Eric] Yep. 00:02:53,860 --> 00:03:00,520 [John] So I think we should talk about... And, and I think some of that is some, some science that came out around how our brains actually work. 00:03:00,520 --> 00:03:01,000 [Eric] Yeah. 00:03:01,000 --> 00:03:08,320 [John] Um, and then the popul- and then that coming into the popular, um, understanding, and therefore, it kind of evaporated off of job descriptions. 00:03:08,320 --> 00:03:08,480 [Eric] Yeah. 00:03:08,480 --> 00:03:15,780 [John] But what's, what's your memory around, like, the, the, the science around it, and maybe even just personal experience with multitasking? 00:03:15,780 --> 00:03:19,140 [Eric] I s- I actually studied this, not formally- 00:03:19,140 --> 00:03:19,340 [John] Mm-hmm 00:03:19,340 --> 00:03:33,080 [Eric] ... in an academic sense, but I became really interested in this when I, like, early in my career. And my first job out of school, actually, I had a desktop computer. 00:03:33,080 --> 00:03:33,090 [John] Okay. 00:03:33,090 --> 00:03:35,220 [Eric] It was a Power Mac, which was very cool. 00:03:35,220 --> 00:03:35,910 [John] Oh, wow. Yeah. 00:03:35,910 --> 00:03:41,290 [Eric] Yeah, it was a Power Mac tower with an Apple display plugged in. It was super cool. 00:03:41,290 --> 00:03:41,680 [John] [laughs] So- 00:03:41,680 --> 00:03:41,760 [Eric] Um- 00:03:41,760 --> 00:03:43,020 [John] You s- you still have that setup. [laughs] 00:03:43,020 --> 00:03:45,120 [Eric] I still have that set- except for the tower. I do run a- 00:03:45,120 --> 00:03:45,850 [John] Okay. That is true 00:03:45,850 --> 00:03:46,620 [Eric] ... I do run a laptop. 00:03:46,620 --> 00:03:46,710 [John] That is true. 00:03:46,710 --> 00:03:48,020 [Eric] But I do have the Apple display set up. 00:03:48,020 --> 00:03:49,070 [John] Right, right, right. 00:03:49,070 --> 00:03:49,070 [Eric] Yeah. 00:03:49,070 --> 00:03:49,200 [John] Right. 00:03:49,200 --> 00:04:01,970 [Eric] Um, yeah, I've been a Apple display loyalist for a long time. Uh, but then later on, I got a laptop, you know, and so I could sort of take work with me, and then, you know, started exploring a multi-screen setup. 00:04:01,970 --> 00:04:02,000 [John] Mm-hmm. 00:04:02,000 --> 00:04:05,460 [Eric] So having two monitors. Um, and 00:04:06,860 --> 00:04:08,120 [Eric] there... I remember... 00:04:09,410 --> 00:04:19,340 [Eric] Yeah, and also, this is, um, this was pre-Slack days, and so we pri- still primarily used email- 00:04:19,340 --> 00:04:19,560 [John] Mm-hmm 00:04:19,560 --> 00:04:28,060 [Eric] ... for internal communication. And then, uh, live chat started to be used in the business place more. 00:04:28,060 --> 00:04:29,200 [John] Yeah, like Skype. 00:04:29,200 --> 00:04:33,990 [Eric] Skype, yes, so we used Skype. There was also Yammer. Do you remember this? 00:04:33,990 --> 00:04:34,020 [John] Yeah. 00:04:34,020 --> 00:04:36,960 [Eric] It was Twitter for, like, inside of a company. 00:04:36,960 --> 00:04:37,920 [John] Yeah. Yeah. 00:04:37,920 --> 00:04:38,420 [Eric] Um- 00:04:38,420 --> 00:04:39,780 [John] I think Microsoft bought them. 00:04:39,780 --> 00:04:45,700 [Eric] Yep. And they did. Yeah, I think they did. Uh, and then Google Chat was actually used a lot of places- 00:04:45,700 --> 00:04:46,360 [John] Mm-hmm 00:04:46,360 --> 00:04:51,020 [Eric] ... um, that ran on, you know, sort of the Google Workspace suite. And I just remember 00:04:52,420 --> 00:04:57,840 [Eric] feeling a little bit overwhelmed at, like, okay, now there's email, and now there's Skype, which we use- 00:04:57,840 --> 00:04:57,850 [John] Mm-hmm 00:04:57,850 --> 00:05:14,040 [Eric] ... you know, internally. And, and then the iPhone had just come out, and so this is 2008. So the iPhone's, like, a year old, and so people where I worked l- you know, the, the following year, 2008, 2009, started to buy them 'cause the price came down. 00:05:14,040 --> 00:05:14,520 [John] Mm-hmm. 00:05:14,520 --> 00:05:18,640 [Eric] And then now, all of a sudden, you have, you know, email. Everyone has email on their phone. 00:05:18,640 --> 00:05:19,360 [John] Right. 00:05:19,360 --> 00:05:26,940 [Eric] And it created this environment where the form factor kind of made it feel like you could multitask. 00:05:26,940 --> 00:05:27,760 [John] Right. 00:05:27,760 --> 00:05:32,280 [Eric] But I started to notice that it was more overwhelming than productive- 00:05:32,280 --> 00:05:32,290 [John] Right 00:05:32,290 --> 00:05:33,620 [Eric] ... you know, for me personally. 00:05:33,620 --> 00:05:34,040 [John] Right. 00:05:34,040 --> 00:05:49,580 [Eric] And so that's when I started to look into it, and really, the, almost all of the... I haven't, um, I haven't refreshed my research on this in, in some time. Actually, I, I did a little bit, which we'll talk about in the AI section. 00:05:50,700 --> 00:05:52,880 [Eric] But the, um, 00:05:54,040 --> 00:05:58,539 [Eric] all of the science points to the fact that multitasking is a net negative. 00:05:58,540 --> 00:05:59,120 [John] Right. 00:05:59,120 --> 00:06:12,480 [Eric] It creates a sensation of getting more done But divided focus is generally a net negative. Whether that is taking... Uh, and there are two sort of main vectors at least that I explored. 00:06:13,620 --> 00:06:21,380 [Eric] One is that you may be w- trying to work on multiple things at the same time, but all of those things take longer, right? 00:06:21,380 --> 00:06:21,410 [John] Yeah. 00:06:21,410 --> 00:06:25,050 [Eric] So in aggregate, you're not being... You're not getting more done faster. 00:06:25,050 --> 00:06:26,000 [John] Right. 00:06:26,000 --> 00:06:33,810 [Eric] And then there's a quality impact, too, because you're not focused. You, you can't do deep work on a single task- 00:06:33,810 --> 00:06:33,810 [John] Right 00:06:33,810 --> 00:06:34,780 [Eric] ... and that's generally how you- 00:06:34,780 --> 00:06:35,040 [John] Right 00:06:35,040 --> 00:06:35,980 [Eric] ... draw out quality. 00:06:35,980 --> 00:06:37,520 [John] Right. Do you remember Scrum? 00:06:37,520 --> 00:06:38,000 [Eric] Yeah. 00:06:38,000 --> 00:06:40,680 [John] Yeah. So that's one of the agile methodology- 00:06:40,680 --> 00:06:41,400 [Eric] Yep 00:06:41,400 --> 00:06:46,660 [John] ... pieces. Um, so I wr- I read the kind of original book by one of the, the guys that came up with it. 00:06:46,660 --> 00:06:46,720 [Eric] Mm-hmm. 00:06:46,720 --> 00:06:56,880 [John] And one of the big components actually behind it was this multitasking thing, where he was essentially saying, like based on the science, number one, you can only actually do one thing at once. 00:06:56,880 --> 00:06:57,240 [Eric] Yes. 00:06:57,240 --> 00:06:59,520 [John] You can rapidly switch back and forth between- 00:06:59,520 --> 00:06:59,670 [Eric] Yes 00:06:59,670 --> 00:07:04,000 [John] ... do one thing, but you can't actually multitask. That's not a th- for a human, that's not a thing. 00:07:04,000 --> 00:07:04,400 [Eric] Y- yes. Yes. 00:07:04,400 --> 00:07:05,200 [John] Your mind doesn't work that way. 00:07:05,200 --> 00:07:08,900 [Eric] I think that [laughs] is actually a great fundamental point that I missed, is that- 00:07:08,900 --> 00:07:09,230 [John] Right 00:07:09,230 --> 00:07:11,350 [Eric] ... it's not multitasking, it's context switching. 00:07:11,350 --> 00:07:11,360 [John] Right. 00:07:11,360 --> 00:07:12,150 [Eric] You literally cannot work on- 00:07:12,150 --> 00:07:13,500 [John] Rapid context switching- 00:07:13,500 --> 00:07:13,510 [Eric] Yes 00:07:13,510 --> 00:07:14,510 [John] ... is multitasking, essentially. 00:07:14,510 --> 00:07:16,260 [Eric] Yes. You can't do two things at the same time. 00:07:16,260 --> 00:07:17,080 [John] Right, right. 00:07:17,080 --> 00:07:17,120 [Eric] Uh- 00:07:17,120 --> 00:07:21,780 [John] But I still remember a graph from that book where, um, where he shows, 00:07:22,960 --> 00:07:26,539 [John] uh, let's just call it, I don't know, several different tasks, like task A, B, C, D- 00:07:26,540 --> 00:07:26,550 [Eric] Mm-hmm 00:07:26,550 --> 00:07:30,760 [John] ... and then shows on a timeline, like A to B to C to A to B to C, 00:07:31,820 --> 00:07:32,590 [John] six months. 00:07:32,590 --> 00:07:32,599 [Eric] Mm-hmm. 00:07:32,600 --> 00:07:38,020 [John] And then he shows like just A, done, just B, done, just C... And, and it shortens the time. 00:07:38,020 --> 00:07:38,280 [Eric] Yep. 00:07:38,280 --> 00:07:48,450 [John] And essentially that was part of the, the research, um, that, that he, um, had as part of some of the fundamentals in, in the, you know, philo- whatever you wanna call it, the, uh- 00:07:48,450 --> 00:07:48,450 [Eric] Yep 00:07:48,450 --> 00:07:49,480 [John] ... methodology. 00:07:49,480 --> 00:07:50,080 [Eric] Yep. 00:07:50,080 --> 00:07:57,680 [John] Um, and, and that really stuck with me, um, beyond just project management, um, methodologies or methods of working. 00:07:57,680 --> 00:07:57,900 [Eric] Yep. 00:07:57,900 --> 00:08:05,599 [John] It's like, oh, if we have these blocks and we get to, to done, it's way better than like half, half, half- 00:08:05,600 --> 00:08:06,310 [Eric] Yes, exactly 00:08:06,310 --> 00:08:06,900 [John] ... come back and the next half, half, half. 00:08:06,900 --> 00:08:07,320 [Eric] Yeah. 00:08:07,320 --> 00:08:07,480 [John] Yeah. 00:08:07,480 --> 00:08:14,620 [Eric] There's another really good book that touches on a lot of these subjects. The book is called Personal Kanban. 00:08:14,620 --> 00:08:16,160 [John] Okay. [laughs] I don't know that one. 00:08:16,160 --> 00:08:21,580 [Eric] And it's a great, it's a great book. I, I highly recommend it, um, 00:08:22,880 --> 00:08:34,980 [Eric] because it talks about throughput. And so, you know, a lot of this kinda goes, i- it, it is very much in the realm, although, you know, it's, it's personal, right? 00:08:34,980 --> 00:08:35,090 [John] Mm-hmm. 00:08:35,090 --> 00:08:37,440 [Eric] And so it talks about, it's more about personal productivity- 00:08:37,440 --> 00:08:37,450 [John] Right 00:08:37,450 --> 00:08:39,720 [Eric] ... but it extends to team context. 00:08:39,720 --> 00:08:40,040 [John] Right. 00:08:40,040 --> 00:08:43,880 [Eric] But it's very much in the realm of, um, The Phoenix Project. 00:08:43,880 --> 00:08:44,340 [John] Oh, sure. 00:08:44,340 --> 00:08:45,260 [Eric] You know? And then- 00:08:45,260 --> 00:08:45,270 [John] Yeah 00:08:45,270 --> 00:08:52,060 [Eric] ... sort of going back to the goal and, you know, a bunch of the books that The Phoenix Project references that sort of talk about manufacturing and- 00:08:52,060 --> 00:08:52,240 [John] Right 00:08:52,240 --> 00:08:53,200 [Eric] ... and throughput, right? 00:08:53,200 --> 00:08:53,980 [John] Right. 00:08:53,980 --> 00:08:57,840 [Eric] And, uh, Personal Kanban touches on this as well, right? 00:08:57,840 --> 00:08:58,150 [John] Right. 00:08:58,150 --> 00:09:00,500 [Eric] Where, where you limit work in progress, you know? 00:09:00,500 --> 00:09:00,820 [John] Yeah. 00:09:00,820 --> 00:09:01,120 [Eric] And so- 00:09:01,120 --> 00:09:01,890 [John] Exactly. Yeah 00:09:01,890 --> 00:09:05,749 [Eric] ... what is being worked on, you can only have one thing being worked on at a time. 00:09:05,749 --> 00:09:05,779 [John] Right. 00:09:05,780 --> 00:09:11,290 [Eric] And they generally recommend, you know, making it literally physical, so they- 00:09:11,290 --> 00:09:11,370 [John] Mm-hmm 00:09:11,370 --> 00:09:12,590 [Eric] ... you know, they're big on sticky notes. 00:09:12,590 --> 00:09:13,680 [John] Sticky notes, yeah. 00:09:13,680 --> 00:09:21,720 [Eric] Uh, you know, which is tough in a remote work environment. But, um, you know, that way you... I- if the rule is you can only have one sticky note in the column- 00:09:21,720 --> 00:09:21,980 [John] Mm-hmm 00:09:21,980 --> 00:09:24,060 [Eric] ... you know, you actually have to physically- 00:09:24,060 --> 00:09:24,280 [John] Yeah 00:09:24,280 --> 00:09:34,480 [Eric] ... make sure that's the case. So it's... Multitasking is bad generally. I, I would say that this is a big pro- this is a big problem before AI. 00:09:34,480 --> 00:09:35,320 [John] Yeah. 00:09:35,320 --> 00:09:45,820 [Eric] And I think it's probably the silent killer of productivity for a majority of people who I know who work in a, in a digital environment. 00:09:45,820 --> 00:09:46,020 [John] Right. 00:09:46,020 --> 00:09:48,800 [Eric] You know? Not tech, but just generally I'm using- 00:09:48,800 --> 00:09:48,870 [John] Right 00:09:48,870 --> 00:09:54,400 [Eric] ... computers, technology, whatever, right? It, it's... It erodes it at an extreme level, actually- 00:09:54,400 --> 00:09:54,410 [John] Right 00:09:54,410 --> 00:09:55,500 [Eric] ... scientifically. So- 00:09:55,500 --> 00:09:55,580 [John] Right 00:09:55,580 --> 00:10:02,680 [Eric] ... it can lengthen the amount... Trying to multitask can lengthen the amount of time it takes to do something dramatically. 00:10:02,680 --> 00:10:02,840 [John] Yeah. 00:10:02,840 --> 00:10:07,280 [Eric] So it can take a 20-minute task and make it an hour or even multi-hour task, depending on the- 00:10:07,280 --> 00:10:07,290 [John] Yeah 00:10:07,290 --> 00:10:08,300 [Eric] ... complexity. 00:10:08,360 --> 00:10:10,360 [John] So, so you talked about 2008. 00:10:10,360 --> 00:10:10,380 [Eric] Yep. 00:10:10,380 --> 00:10:13,180 [John] So let me fast-forward us to 2018. 00:10:13,180 --> 00:10:13,880 [Eric] Yep. 00:10:13,880 --> 00:10:16,220 [John] And [laughs] it got worse, right? 00:10:16,220 --> 00:10:16,330 [Eric] Mm-hmm. 00:10:16,330 --> 00:10:22,140 [John] Where essentially, like yeah, you have mobile, yeah, you have your, your, your desktop, and then you have Slack or Teams. 00:10:22,140 --> 00:10:22,360 [Eric] Yep. 00:10:22,360 --> 00:10:29,600 [John] And essentially, um, if you have... And, and you're... So have email. Um, if you have notifications on all the time, then you would- 00:10:29,600 --> 00:10:29,609 [Eric] Yes 00:10:29,609 --> 00:10:46,600 [John] ... be interrupted every five minutes or, or more often. And this is pre-AI, but, but the work still was, was like, was focused work. Uh, you thinking, what- whatever your job is, like trying to think and like organize, um, a project or trying to think in code or whatever- 00:10:46,600 --> 00:10:46,680 [Eric] Yep 00:10:46,680 --> 00:10:47,710 [John] ... that was still the work. 00:10:47,710 --> 00:10:47,880 [Eric] Yep. 00:10:47,880 --> 00:10:58,780 [John] I think steering into the AI part, the weird fundamental change here is there's a lot of arguments out there right now for now that the fundamental work should be multitasking- 00:10:58,780 --> 00:10:59,100 [Eric] Yes 00:10:59,100 --> 00:11:03,380 [John] ... when it comes to especially coding, and I think probably aspects of knowledge work. 00:11:03,380 --> 00:11:03,660 [Eric] Yep. 00:11:03,660 --> 00:11:04,260 [John] So I think that's 00:11:05,580 --> 00:11:07,600 [John] fascinating. Like have we not learned anything? 00:11:07,660 --> 00:11:08,660 [Eric] Well, I think- 00:11:08,660 --> 00:11:11,620 [John] Or why is it different? Why, why are people thinking about it this way? 00:11:11,620 --> 00:11:15,700 [Eric] It, it is because when you 00:11:17,040 --> 00:11:22,880 [Eric] start to leverage AI in its full 00:11:24,400 --> 00:11:25,220 [Eric] capacity, 00:11:26,640 --> 00:11:41,880 [Eric] you... There, there's this, there's a sort of a break point, right? And I know that our listeners are across a very wide spectrum, right? I think most people start by using ChatGPT or Claude, and you open a chat, and you have a long interaction about something, right? 00:11:41,880 --> 00:11:41,940 [John] Mm-hmm. 00:11:43,080 --> 00:11:51,820 [Eric] And then I would say the next step is you realize that you can have multiple chats going at the same time. 00:11:51,820 --> 00:11:52,120 [John] Yep. 00:11:52,120 --> 00:12:03,140 [Eric] Right? So like, oh, I can, you know, I can start a chat and talk about this thing. I can start a chat and talk about this thing. Especially, you know, y- it's like, okay, well, I can actually sort of get multiple things going at once. 00:12:04,180 --> 00:12:16,436 [Eric] And then I think the next step is realizing that if th- that is It seems advantageous, especially if something's going to take a long time. 00:12:16,436 --> 00:12:16,836 [John] Right. 00:12:16,836 --> 00:12:26,956 [Eric] So I have a research task, and it's gonna take, you know, Claude or GPT just to say 20 minutes to sort of run this long research task or something, right? 00:12:26,956 --> 00:12:27,216 [John] Right. 00:12:27,216 --> 00:12:33,006 [Eric] And so at that point, you're thinking, "Oh, well, this is interesting. I can kind of game the system," right? 00:12:33,006 --> 00:12:33,056 [John] Right. 00:12:33,056 --> 00:12:39,146 [Eric] So I could start five research tasks, you know, co- irrespective of cost, but, you know- 00:12:39,146 --> 00:12:39,146 [John] Sure 00:12:39,146 --> 00:12:43,456 [Eric] ... I could start five research tasks, and all this stuff could be going on while I'm working on something else. 00:12:43,456 --> 00:12:43,546 [John] Right. 00:12:43,546 --> 00:13:02,506 [Eric] And then I think I'll... You know, for those of us who have migrated from, like, a desktop or a web app interacting with GPT to more of a terminal-based workflow, and so I don't really write code for my job. I'm a writer. 00:13:02,506 --> 00:13:03,116 [John] Right. 00:13:03,116 --> 00:13:07,316 [Eric] But I use AI exclusively through the command line in my terminal- 00:13:07,316 --> 00:13:07,326 [John] Right 00:13:07,326 --> 00:13:12,336 [Eric] ... because there are a lot of, um, advantages to that which we can discuss- 00:13:12,396 --> 00:13:12,406 [John] Right 00:13:12,406 --> 00:13:25,776 [Eric] ... on another episode. But at that point, it-- you really supercharge this, the ability to do a lot of things at once because you can have as many terminal windows open as you want- 00:13:25,776 --> 00:13:26,176 [John] Right 00:13:26,176 --> 00:13:33,266 [Eric] ... and you can have them all running at the same time, and you sort of break free from this interface challenge of having, like, multiple chats in the sidebar and, like, clicking- 00:13:33,266 --> 00:13:33,636 [John] Yeah 00:13:33,636 --> 00:13:33,856 [Eric] ... around, right? 00:13:33,856 --> 00:13:33,886 [John] Yeah. 00:13:33,886 --> 00:13:36,296 [Eric] You can have... You know, I mean, people have, uh, like, 00:13:37,576 --> 00:13:42,626 [Eric] easily see, like, people [chuckles] I work around will have 10 terminal windows open on their computer- 00:13:42,626 --> 00:13:42,626 [John] Right 00:13:42,626 --> 00:13:45,686 [Eric] ... all running different AI jobs at the same time. And 00:13:46,816 --> 00:13:49,216 [Eric] there are a couple things here that are interesting. 00:13:50,376 --> 00:14:08,676 [Eric] One is that it n- that just feels very natural based on the way that the tools work, right? And one of the things that you don't necessarily think about explicitly, but that's just part of the experience, is that it takes time for an agent to respond. 00:14:08,676 --> 00:14:09,276 [John] Right. 00:14:09,276 --> 00:14:23,136 [Eric] And so before, you know, if you were working on something, let's say you were working on coding something, or I was working on a blog post, or you were doing research, right? So let's just talk about researching for a blog post, you know, or trying to figure out- 00:14:23,136 --> 00:14:23,216 [John] Mm-hmm 00:14:23,216 --> 00:14:41,745 [Eric] ... um, you know, how to write a Python script to perform some sort of statistical analysis. You're gonna go search on Google. You know, you're gonna, you know, sort of read articles, scan here, scan there, go back and look at the code, run a test. You're sort of proactively driving this whole time. 00:14:41,745 --> 00:14:41,776 [John] Right. 00:14:41,776 --> 00:14:46,936 [Eric] You know, or you, like, take your hands off your keyboard and sort of stare up and think, like, "Okay, how do I do this," right? 00:14:46,936 --> 00:14:47,676 [John] Right. 00:14:47,676 --> 00:14:56,696 [Eric] And so if we walk through that process even in the first steps, now you're gonna have your, like, AI go do the research for you, right? 00:14:56,696 --> 00:14:56,876 [John] Right. 00:14:56,876 --> 00:15:08,196 [Eric] Here's... So I'm gonna describe the problem, and then it's gonna go out and research. It's gonna find all these blog posts, all that sort of stuff, right? And so instead of you being an active part of doing that, you're waiting. 00:15:08,196 --> 00:15:08,596 [John] Yeah. 00:15:08,596 --> 00:15:08,946 [Eric] Right? 00:15:08,946 --> 00:15:08,976 [John] Right. 00:15:08,976 --> 00:15:14,276 [Eric] And so, uh, a-and even if it's three or four minutes, like, you kick off a task- 00:15:14,276 --> 00:15:14,596 [John] Right 00:15:14,596 --> 00:15:17,396 [Eric] ... and it's like, "Okay, I'm just sitting here for three or four minutes." 00:15:17,396 --> 00:15:17,546 [John] Right. 00:15:17,546 --> 00:15:21,956 [Eric] "I kind of need this information in order to move forward." And so 00:15:23,236 --> 00:15:31,956 [Eric] you get to a place where this multitasking modality feels natural just because of the way that the tools work and the latency of- 00:15:31,956 --> 00:15:31,966 [John] Yeah 00:15:31,966 --> 00:15:32,636 [Eric] ... issuing- 00:15:32,636 --> 00:15:32,646 [John] Yeah 00:15:32,646 --> 00:15:34,615 [Eric] ... you know, a prompt and then waiting on a response. 00:15:34,616 --> 00:15:35,376 [John] Yeah. 00:15:35,376 --> 00:15:37,296 [Eric] And so you end up, I think, 00:15:38,316 --> 00:15:45,596 [Eric] multitasking in a way cra- so you add this layer of multitasking to the work itself, right? So to your point- 00:15:45,596 --> 00:15:45,816 [John] Yeah. Right 00:15:45,816 --> 00:15:54,836 [Eric] ... you have Slack. You have, you know, email. You have all these notifications going on. Um, you know, project management tools, which, by the way, turn all your notifications off. 00:15:54,836 --> 00:15:54,846 [John] [chuckles] 00:15:54,846 --> 00:15:56,706 [Eric] It's the best thing you can do for productivity. 00:15:56,706 --> 00:15:56,756 [John] Yeah. 00:15:56,756 --> 00:16:04,076 [Eric] So you have that layer that's sort of eroding focus, and then now when you add AI into the actual work itself- 00:16:04,076 --> 00:16:04,085 [John] Right 00:16:04,085 --> 00:16:07,025 [Eric] ... it's a layer on top of that. And so I just- 00:16:07,025 --> 00:16:08,436 [John] Which is... It's way worse. 00:16:08,436 --> 00:16:09,276 [Eric] It's way worse. 00:16:09,276 --> 00:16:12,276 [John] Way worse. The, the other one feels insignificant now. 00:16:12,276 --> 00:16:12,416 [Eric] Yes. 00:16:12,416 --> 00:16:14,236 [John] Like, the notifications feel like, "Eh." 00:16:14,236 --> 00:16:15,256 [Eric] Yeah, exactly. 00:16:15,256 --> 00:16:19,536 [John] But the... But that... And, and here's, here's the interesting thing. I think it's the waiting right now which will improve. 00:16:19,536 --> 00:16:19,786 [Eric] Mm-hmm. 00:16:19,786 --> 00:16:21,536 [John] Like, uh, AI's gonna get faster. 00:16:21,536 --> 00:16:21,696 [Eric] Yep. 00:16:21,696 --> 00:16:25,766 [John] I don't think anybody's concerned about that long term. But the other problem is 00:16:26,876 --> 00:16:33,175 [John] I, I understand the desire to have 10 things working at once. 00:16:33,176 --> 00:16:33,476 [Eric] Mm-hmm. 00:16:33,476 --> 00:16:38,855 [John] Um, th-there is a, the, there... If, if you can have them working on the right things and doing good things- 00:16:38,856 --> 00:16:39,036 [Eric] Mm-hmm 00:16:39,036 --> 00:16:40,756 [John] ... like, that is a net positive. 00:16:40,756 --> 00:16:41,396 [Eric] Yep. 00:16:41,396 --> 00:16:52,956 [John] I think the problem is... And there's an interesting thing that happened this week I wanna get to here. The problem is if you're having to constantly switch between the 10 things every five minutes- 00:16:52,956 --> 00:16:53,596 [Eric] Yeah 00:16:53,596 --> 00:16:55,156 [John] ... it's super costly. 00:16:55,156 --> 00:16:55,516 [Eric] Yep. 00:16:55,516 --> 00:17:01,496 [John] And the, like, last, in the last month, there's been a lot of conversation around loops- 00:17:01,496 --> 00:17:01,586 [Eric] Yep 00:17:01,586 --> 00:17:09,616 [John] ... which we could get into more, but the idea is essentially you having to intervene less and you getting it in a positive loop- 00:17:09,616 --> 00:17:09,626 [Eric] Mm-hmm 00:17:09,626 --> 00:17:10,866 [John] ... um, trying to accomplish a goal- 00:17:10,866 --> 00:17:10,906 [Eric] Mm-hmm 00:17:10,906 --> 00:17:31,896 [John] ... for example. But w- an interesting thing that I think it happened yesterday. Yesterday or earlier this week, um, Anthropic came out with, and, and, and there's, and there's other companies already doing this, um, something called Claude Tag. And the concept is using Claude inside of Slack 00:17:33,236 --> 00:17:37,656 [John] asynchronously, giving it a task. It works, responds back. 00:17:37,656 --> 00:17:37,706 [Eric] Mm-hmm. 00:17:37,706 --> 00:17:41,796 [John] And then it can also, like, read from the channel or the place that you talked to it in- 00:17:41,796 --> 00:17:42,016 [Eric] Yep 00:17:42,016 --> 00:17:50,176 [John] ... and, and continue a loop down a path, which is different than having 10 tabs open from a workflow standpoint- 00:17:50,176 --> 00:17:50,186 [Eric] Yep 00:17:50,186 --> 00:17:53,235 [John] ... and then just having to hop back and forth and feed it and be like, "Okay, that's good. That's good." 00:17:53,236 --> 00:17:53,346 [Eric] Mm-hmm. 00:17:53,346 --> 00:18:03,295 [John] "I looked at that. I put yes, yes, yes." Like, by very nature, if you're using it inside a chat application, it actually... I, I think it would reduce that behavior. So that's an- 00:18:03,296 --> 00:18:03,326 [Eric] Yep 00:18:03,326 --> 00:18:08,676 [John] ... interesting thing. Like, A, we don't know what, what people are gonna prefer, um, necessarily. 00:18:08,676 --> 00:18:08,766 [Eric] Yep. 00:18:08,766 --> 00:18:23,660 [John] And then B, um, if we go more down this road of, of using our same interfaces where we talk to humans to talk to the AI, I, I wonder Like, like the open question is like how does that impact multitasking? 00:18:23,660 --> 00:18:24,020 [Eric] Yeah. 00:18:24,020 --> 00:18:24,680 [John] Right? 00:18:24,680 --> 00:18:24,940 [Eric] So, 00:18:26,620 --> 00:18:32,910 [Eric] I will be very interested to see how this plays out. Uh, I agree that there is... 00:18:34,740 --> 00:18:36,400 [Eric] That the form factor 00:18:37,680 --> 00:18:38,030 [Eric] of... 00:18:39,220 --> 00:18:42,720 [Eric] So we move from this I have 10 terminal windows open, right? 00:18:42,720 --> 00:18:42,790 [John] Mm-hmm. 00:18:42,790 --> 00:18:48,360 [Eric] And I think this is, we're talking about generally people who are 10 tabs, 10 windows, right? 00:18:48,360 --> 00:18:48,960 [John] Right, yeah. 00:18:48,960 --> 00:18:50,870 [Eric] People who have crossed this threshold of like I'm- 00:18:50,870 --> 00:18:53,570 [John] People, people that have 50 Chrome tabs open. 00:18:53,570 --> 00:18:53,580 [Eric] Yes, exactly. 00:18:53,580 --> 00:18:54,440 [John] I mean, it's the same thing. 00:18:54,440 --> 00:18:55,840 [Eric] Yeah, yeah. Um, 00:18:57,560 --> 00:19:03,080 [Eric] but if you continue to invest in learning how to use AI, you will get to that place because you realize- 00:19:03,080 --> 00:19:03,160 [John] Right 00:19:03,160 --> 00:19:03,460 [Eric] ... that there's- 00:19:03,460 --> 00:19:03,470 [John] Right 00:19:03,470 --> 00:19:06,340 [Eric] ... potential leverage in having it work in the background. 00:19:06,340 --> 00:19:06,560 [John] Yeah. 00:19:06,560 --> 00:19:06,800 [Eric] Right? 00:19:06,800 --> 00:19:06,930 [John] Exactly. 00:19:06,930 --> 00:19:08,150 [Eric] So whatever that looks like. 00:19:08,150 --> 00:19:08,380 [John] Exactly. Yeah. 00:19:08,380 --> 00:19:11,860 [Eric] I realize the 10 terminal window thing is not [laughs] you know, it's not everyone. 00:19:11,860 --> 00:19:12,640 [John] Yeah, yeah. 00:19:12,640 --> 00:19:17,400 [Eric] But if you haven't tried running multiple AI jobs at the same time, you should. Like- 00:19:17,400 --> 00:19:17,770 [John] Right. Yeah 00:19:17,770 --> 00:19:23,000 [Eric] ... it, you know, um, we'll make w- what we think is dangerous about that with multitasking clear- 00:19:23,000 --> 00:19:23,890 [John] Right 00:19:23,890 --> 00:19:25,060 [Eric] ... uh, before the end of the show. But, 00:19:26,200 --> 00:19:26,400 [Eric] um, 00:19:27,660 --> 00:19:28,020 [Eric] the, 00:19:29,920 --> 00:19:38,510 [Eric] the, the form factor of 10 windows in front of me, the problem there is that they finish at different, they finish jobs- 00:19:38,510 --> 00:19:38,540 [John] Yes 00:19:38,540 --> 00:19:39,210 [Eric] ... at different times. 00:19:39,210 --> 00:19:39,220 [John] Yeah. 00:19:39,220 --> 00:19:43,580 [Eric] And so I h- I'm like constantly... It- it's context switching in the extreme. 00:19:43,580 --> 00:19:43,770 [John] Right. 00:19:43,770 --> 00:19:44,470 [Eric] Right? Which we know- 00:19:44,470 --> 00:19:44,810 [John] Right 00:19:44,810 --> 00:19:46,420 [Eric] ... is counterproductive, like in aggregate. 00:19:46,420 --> 00:19:47,150 [John] Right. 00:19:47,150 --> 00:19:53,780 [Eric] Right? If you put that in Slack, it is much closer to managing a team of people. 00:19:53,780 --> 00:19:54,380 [John] Right. 00:19:54,380 --> 00:19:54,620 [Eric] Right? 00:19:55,660 --> 00:19:56,080 [Eric] Um, 00:19:58,080 --> 00:20:01,740 [Eric] but I'm not sold on that being the solution 00:20:04,000 --> 00:20:08,680 [Eric] because it ultimately comes down to personal discipline in both cases. 00:20:08,680 --> 00:20:09,040 [John] Yeah. 00:20:09,040 --> 00:20:09,390 [Eric] Right? 00:20:09,390 --> 00:20:17,759 [John] 'Cause well, then you're just gonna... But let's, let's play this out. Say you do move all your work there, then y- you're probably just gonna have more Slack notifications. 00:20:17,760 --> 00:20:19,820 [Eric] You, you will have more Slack. Yes, yes, exactly. 00:20:19,820 --> 00:20:20,170 [John] [laughs] Right? 00:20:20,170 --> 00:20:20,770 [Eric] Right? And so- 00:20:20,770 --> 00:20:22,100 [John] Which was a problem before. We already- 00:20:22,100 --> 00:20:22,240 [Eric] Yes 00:20:22,240 --> 00:20:22,740 [John] ... we already know. 00:20:22,740 --> 00:20:24,180 [Eric] Yes, yes, exactly. 00:20:24,180 --> 00:20:25,210 [John] Um. 00:20:25,210 --> 00:20:34,840 [Eric] Um, I mean, to some extent, the physical constraints of how many terminal windows you can fit on your screen on a practical level, especially if you're not using a large Apple display- 00:20:34,840 --> 00:20:35,110 [John] Right. Sure 00:20:35,110 --> 00:20:36,380 [Eric] ... like you're just doing it on a laptop. 00:20:36,380 --> 00:20:36,670 [John] Sure. 00:20:36,670 --> 00:20:39,140 [Eric] It's like again, there's a point at which this doesn't work, and so there's sort of- 00:20:39,140 --> 00:20:39,150 [John] Yeah 00:20:39,150 --> 00:20:39,720 [Eric] ... a physical limit. 00:20:39,720 --> 00:20:39,890 [John] Well- 00:20:39,890 --> 00:20:40,569 [Eric] In Slack, there is- 00:20:40,569 --> 00:20:41,170 [John] I've noticed that- 00:20:41,170 --> 00:20:41,590 [Eric] ... there's no limit, you know 00:20:41,590 --> 00:20:43,960 [John] ... that vertical tabs is, is becoming a thing- 00:20:43,960 --> 00:20:44,230 [Eric] Oh, yeah 00:20:44,230 --> 00:20:44,720 [John] ... because of that. 00:20:44,720 --> 00:20:45,290 [Eric] For sure. [laughs] 00:20:45,290 --> 00:20:46,940 [John] To just, just if you have that problem- 00:20:46,940 --> 00:20:47,100 [Eric] Yes 00:20:47,100 --> 00:20:47,990 [John] ... you can just flip 'em- 00:20:47,990 --> 00:20:48,510 [Eric] Yes, exactly 00:20:48,510 --> 00:20:50,610 [John] ... flip over. [laughs] 00:20:50,610 --> 00:20:50,620 [Eric] Uh- 00:20:50,620 --> 00:20:51,740 [John] But, yeah 00:20:51,740 --> 00:20:52,830 [Eric] ... but the other thing, 00:20:54,200 --> 00:20:57,400 [Eric] going back to first principles, is that if you think about 00:20:58,660 --> 00:21:02,290 [Eric] people d- so good... I- if we take 00:21:03,400 --> 00:21:05,120 [Eric] the 10 windows, 10 tabs- 00:21:05,120 --> 00:21:05,380 [John] Right 00:21:05,380 --> 00:21:09,740 [Eric] ... and we translate that into Slack, it's like, okay, this is much more like managing a team. 00:21:09,740 --> 00:21:10,460 [John] Right. 00:21:10,460 --> 00:21:14,280 [Eric] I think the challenge is that great managers are very rare. 00:21:14,280 --> 00:21:14,400 [John] Sure. 00:21:14,400 --> 00:21:14,629 [Eric] Right? 00:21:14,629 --> 00:21:14,659 [John] Yeah. 00:21:14,660 --> 00:21:18,439 [Eric] And great, great managers, um, 00:21:19,520 --> 00:21:22,680 [Eric] are extremely good at prioritization. They're- 00:21:22,680 --> 00:21:22,710 [John] Right 00:21:22,710 --> 00:21:32,620 [Eric] ... extremely good at reviewing work and giving feedback. They're extremely good at, like, limiting, um, you know, at, at, at creating limits, right? 00:21:32,620 --> 00:21:32,820 [John] Right. 00:21:32,820 --> 00:21:33,580 [Eric] And, and so 00:21:34,640 --> 00:21:39,560 [Eric] I'm not convinced that moving it into Slack is going to play to the, to- 00:21:39,560 --> 00:21:40,040 [John] Right 00:21:40,040 --> 00:21:41,380 [Eric] ... the natural advantage of people. 00:21:41,380 --> 00:21:43,580 [John] The, the thing that it plays to the advantage of 00:21:44,720 --> 00:21:49,020 [John] is a theoretical future where long-horizon tasks work a lot better. 00:21:49,080 --> 00:21:49,390 [Eric] Yes. 00:21:49,390 --> 00:21:50,640 [John] 'Cause they don't work great right now. 00:21:50,640 --> 00:21:50,920 [Eric] Yep. 00:21:50,920 --> 00:22:00,160 [John] 'Cause if you, if every... So you've got a couple employees. If all of your employees could only handle work in chunks of 30 minutes or an hour- 00:22:00,160 --> 00:22:00,270 [Eric] Yeah 00:22:00,270 --> 00:22:03,680 [John] ... reliably, you as a manager, like, that's a rough job. 00:22:03,680 --> 00:22:04,960 [Eric] Yes. Totally. [laughs] 00:22:04,960 --> 00:22:05,980 [John] Like you, you know- 00:22:05,980 --> 00:22:07,360 [Eric] I, I think that's the other thing actually is that- 00:22:07,360 --> 00:22:07,700 [John] Yeah 00:22:07,700 --> 00:22:11,020 [Eric] ... AI is really... It, it is fast. It's getting faster. 00:22:11,020 --> 00:22:11,260 [John] Right, right. 00:22:11,260 --> 00:22:15,060 [Eric] And so if you manage 10 agents, like via Slack- 00:22:15,060 --> 00:22:15,409 [John] Right 00:22:15,409 --> 00:22:19,720 [Eric] ... it, uh, it's, as speed increases, it becomes more problematic, right? Um- 00:22:19,720 --> 00:22:28,320 [John] Yeah, yeah. But, but, but consider a future where you could give a goal in Slack to one of your AI agents that was a week-long goal. 00:22:28,400 --> 00:22:28,460 [Eric] Yes. 00:22:28,460 --> 00:22:31,220 [John] And you check in once a week, then that's very fine. 00:22:31,220 --> 00:22:31,320 [Eric] Yeah, yeah, yeah. 00:22:31,320 --> 00:22:37,229 [John] Or even, like, a whole day. And, um, and I know there's some theoretical benchmarks where we're like, "Sure, it can work on a task for eight hours"- 00:22:37,229 --> 00:22:37,229 [Eric] Yeah, yeah 00:22:37,229 --> 00:22:45,180 [John] ... or whatever, but, but I don't, I don't know many people, without any intervention, get really great results. It... You're still measuring that in hours. Like, I don't think anybody measures that in days- 00:22:45,180 --> 00:22:45,400 [Eric] Yeah 00:22:45,400 --> 00:22:45,670 [John] ... at this point. 00:22:45,670 --> 00:22:58,720 [Eric] I agree with that. I agree with that. A- and I think that those long-horizon tasks are actually a small slice of the pie as far as the day-to-day work that you, that most people are leveraging AI to do. 00:22:58,720 --> 00:22:59,500 [John] Oh, for sure. 00:22:59,500 --> 00:22:59,680 [Eric] Right? 00:22:59,680 --> 00:23:00,500 [John] Yeah. 00:23:00,500 --> 00:23:19,100 [Eric] Okay. So let's, let's get practical here at the end. So there's a guy who works for me. Um, he's a writer. We were, we were having this conversation about multitasking, and, you know, writing is very much a, l- like many things, a process of thinking. 00:23:19,100 --> 00:23:19,780 [John] Right. 00:23:19,780 --> 00:23:26,420 [Eric] And so we were going back and forth just, just discussing this challenge of, wow, like, 00:23:27,480 --> 00:23:32,220 [Eric] I can work on five blog posts at once, but is that good? Like- 00:23:32,220 --> 00:23:32,300 [John] Right 00:23:32,300 --> 00:23:34,760 [Eric] ... is that actually helping me do a good job- 00:23:34,760 --> 00:23:34,770 [John] Yeah 00:23:34,770 --> 00:23:36,679 [Eric] ... on all five of those at the same time? 00:23:36,680 --> 00:23:36,830 [John] Right. 00:23:36,830 --> 00:23:47,990 [Eric] Because, like, really thinking through a single conceptual piece on a deep level, um, tends to produce a better, you know, blog post or whatever- 00:23:47,990 --> 00:23:47,990 [John] Right 00:23:47,990 --> 00:23:56,920 [Eric] ... we're, we're working on. And so I think that's a really practical challenge. My... You know, we, we didn't necessarily land on, you know, how do we fix this. 00:23:56,980 --> 00:23:56,990 [John] Right. 00:23:56,990 --> 00:24:06,140 [Eric] But I think it goes back t- I, I actually think the answer is not different for, like, in, in the AI era than it was previously. 00:24:06,140 --> 00:24:06,300 [John] Right. 00:24:06,300 --> 00:24:10,549 [Eric] Right? And the answer is, like, 00:24:11,740 --> 00:24:13,640 [Eric] get good at prioritization- 00:24:13,640 --> 00:24:14,520 [John] Right 00:24:14,520 --> 00:24:17,660 [Eric] ... and limit the amount of work in progress, 00:24:19,440 --> 00:24:23,060 [Eric] and optimize for maintaining deep focus on- 00:24:23,060 --> 00:24:23,760 [John] Right 00:24:23,760 --> 00:24:26,790 [Eric] ... individual things and getting those done, right? 00:24:26,790 --> 00:24:26,820 [John] Right. 00:24:26,820 --> 00:24:30,510 [Eric] So t- to the point of Scrum- 00:24:30,510 --> 00:24:36,249 [Eric] Two things being done really well is better than four things being done at 50 or 75%. 00:24:36,250 --> 00:24:36,490 [John] Right. 00:24:36,490 --> 00:24:38,050 [Eric] Right? Like, absolutely it is. 00:24:38,050 --> 00:24:38,890 [John] Right. 00:24:38,890 --> 00:24:44,690 [Eric] And multitasking, um, you know, is the enemy of that happening. 00:24:44,690 --> 00:24:45,270 [John] Right. 00:24:45,270 --> 00:24:51,180 [Eric] I do think that we will see, and, uh, we're already starting to experiment on my team with, 00:24:52,610 --> 00:24:55,240 [Eric] uh, ways to do this wisely. So 00:24:56,410 --> 00:25:09,690 [Eric] what... You know, if you think about multitasking and you have the 10 windows open or 10 tabs open, part of the problem is that those are different, uh, maybe all different projects, but they're in all different stages, right? 00:25:09,690 --> 00:25:10,570 [John] Yep. 00:25:10,570 --> 00:25:16,630 [Eric] And so how do you sync with- sync- sequence these things in a way that is wise, right? 00:25:16,630 --> 00:25:16,650 [John] Yeah. 00:25:16,650 --> 00:25:25,150 [Eric] So like, okay, background research is great. Do I need five of those running at the same time, you know, and then pinging me when they're done? Or, you know, how does that work? 00:25:25,150 --> 00:25:25,270 [John] Right. 00:25:25,270 --> 00:25:27,010 [Eric] How do you sequence that work? And so, 00:25:28,310 --> 00:25:37,109 [Eric] um, it's... I don't... I mean, I will run multiple research jobs in the background, but I don't have that all in front of me, and then I will- 00:25:37,110 --> 00:25:37,290 [John] Right 00:25:37,290 --> 00:25:39,430 [Eric] ... intentionally switch context- 00:25:39,430 --> 00:25:39,730 [John] Right 00:25:39,730 --> 00:25:43,130 [Eric] ... to a single result and say, "Okay, I'm going to review this," right? 00:25:43,130 --> 00:25:43,910 [John] Right. 00:25:43,910 --> 00:25:58,210 [Eric] And so I think it just really comes back to prioritization and personal discipline and maintaining focus. If you do that, I think you can leverage AI, AI's ability to run multiple jobs at the same time- 00:25:58,210 --> 00:25:58,220 [John] Right 00:25:58,220 --> 00:26:00,020 [Eric] ... in a really powerful way. 00:26:00,020 --> 00:26:10,610 [John] Right. Well, and, and you have this pattern emerging too where people are doing this fan out thing. So say y- you run 20 research jobs, and then AI fans out to 20, fans back into one. 00:26:10,610 --> 00:26:10,800 [Eric] Yep. 00:26:10,800 --> 00:26:14,890 [John] And then you look at one output. Like, there's, there's creative ways to do it to still get- 00:26:14,890 --> 00:26:15,000 [Eric] Yep 00:26:15,000 --> 00:26:15,930 [John] ... the, like, multi, 00:26:16,990 --> 00:26:23,930 [John] um, multiple context, multi, um, agent workflow, but you not have to interact with all 20 things. 00:26:23,930 --> 00:26:24,610 [Eric] 100%. 00:26:24,610 --> 00:26:25,050 [John] Yeah. 00:26:25,050 --> 00:26:32,370 [Eric] I think the other thing that I would really encourage people to think about, and I'm saying this to myself as well, 00:26:33,410 --> 00:26:36,270 [Eric] because you can do so much 00:26:37,350 --> 00:26:38,750 [Eric] simultaneously with AI, 00:26:40,770 --> 00:26:44,080 [Eric] it sort of sets this baseline expectation of how quickly things can happen. 00:26:44,080 --> 00:26:44,130 [John] Sure. 00:26:44,130 --> 00:26:55,010 [Eric] Right? So it's like, well, I can do hours of research. I mean, I, I just ran some numbers on a project that I did, and it's... I think I saved two weeks of total time across, you know, however long this project was. 00:26:55,010 --> 00:26:55,790 [John] Right. 00:26:55,790 --> 00:26:58,910 [Eric] Um, you know, six or eight weeks of, of just research- 00:26:58,910 --> 00:26:58,990 [John] Yeah 00:26:58,990 --> 00:27:10,790 [Eric] ... right, that I would've had to do manually. And so you start to get this expectation of like, well, wow, things can happen really quickly, right? But reviewing research or, like, producing a good piece of work is still slow [laughs]. 00:27:10,790 --> 00:27:11,479 [John] Yeah. Right. 00:27:11,479 --> 00:27:12,470 [Eric] Like, it, it still takes time- 00:27:12,470 --> 00:27:12,490 [John] Yeah 00:27:12,490 --> 00:27:20,650 [Eric] ... to do those things. And so I think that I'm also having to reset my own expectations, 'cause you can- 00:27:20,650 --> 00:27:20,660 [John] Right 00:27:20,660 --> 00:27:32,070 [Eric] ... naturally get to a place where, like, literally everything is faster. And it's like, no, only parts of the process are faster, and it is dramatic. But being a craftsperson is still tedious- 00:27:32,070 --> 00:27:32,180 [John] Right 00:27:32,180 --> 00:27:33,430 [Eric] ... and time-consuming. 00:27:33,430 --> 00:27:35,820 [John] And, and that part might be a little bit, bit slower. 00:27:35,820 --> 00:27:35,820 [Eric] Yes. 00:27:35,820 --> 00:27:38,910 [John] 'Cause if you'd mainly done the research, you're building internal context- 00:27:38,910 --> 00:27:39,020 [Eric] Yes 00:27:39,020 --> 00:27:40,470 [John] ... then you're actually gonna be a little faster- 00:27:40,470 --> 00:27:40,570 [Eric] Mm 00:27:40,570 --> 00:27:45,350 [John] ... when you sit down to write. So you, you actually probably hurt your speed a little bit- 00:27:45,350 --> 00:27:45,669 [Eric] Yep 00:27:45,669 --> 00:27:46,130 [John] ... on that stuff. 00:27:46,130 --> 00:27:48,710 [Eric] Agreed. All right, so turn notifications off. 00:27:49,830 --> 00:27:53,509 [Eric] Uh, be wise about how you deploy multiple AI jobs at the same time. 00:27:53,510 --> 00:27:54,210 [John] Yep. 00:27:54,210 --> 00:27:56,450 [Eric] Focus on deep work and your craft. 00:27:56,450 --> 00:27:58,010 [John] Yep. Couldn't have said it better. 00:27:58,010 --> 00:28:08,400 [Eric] All right. We'll catch you on the next show. [outro music]