deadlines + decisions

transcript.

0:00 Sam: What were we talking about?

0:01 Steve: wait last time yeah

0:04 Sam: Nah. No, like two seconds ago.

0:08 Steve: yeah well good to see you sam um we were talking about uh deadlines and uh lists the never-ending list and

0:20 Sam: Yeah. Yeah, deadlines.

0:23 Steve: getting okay with for me at least I'm getting better with being okay with the never-ending list I find that with AI like I keep giving myself bigger lists so I think that's why I feel like I'm more okay with it because I know it's stuff that a lot of it's stuff that I wouldn't even be working on right now and I'm just grateful that it's being done and advancing But the deadline thing, that's like the stuff that we have to do, which I understand why being suppressed about that is normal.

0:57 Sam: Yeah, it's kind of like the goalposts moved a little bit. I think for me, everything was a deadline. I used to scope stuff out all the time. Like, oh, this will take two weeks, you know, or this is going to require this many engineers. Now it's two weeks turns into two hours and it's done. So the idea of deadlines are very loose. For me right now, I made and it's almost like our only deadlines are always based around some type of event or something, right? So we have four sales a year.

1:37 deadline for when that launches maybe a product launch or something like that outside of that right it's kind of really free form and then sometimes you don't want to say stuff where like oh this whole list of things that we have that we would want for this particular sale yeah it's gonna take like two days to get knocked out whereas it used to take months and then we would have to make trade-offs like are we going to get this or that what do we have bandwidth for and now you kind of get everything done uh the nice to haves are the things you get to decide on whether or not you want to do it and sometimes it's like

2:21 like i could just do it and it wouldn't matter but you're like the vibe is a little off you know maybe it will we'll let this cook for a little bit longer and figure out an entirely different path yeah i don't know deadline it's weird the the sense of time has both stretched and compressed where i think um Knowing that I can focus on all these different things, I just, I almost have to block out periods of time where I'm going to think about these really big amorphous concepts, but the execution is so fast when you actually get the clarity there. So yeah,

3:02 I mean, what kind of deadlines have you been working with or that you've maybe not been working with as much?

3:10 Steve: Yeah, so I can relate to everything you're sharing there because things happen a lot faster, right? Things that would take months now take days. The nice to haves are there because they can be for sure in everything we're doing. The deadlines I'm having are around projects like

3:36 yeah I'm working on this working group there's white papers and getting these white papers completed like it was just just a lot like so that's one thing and it's not something that happens all the time so a lot of it's one-off stuff but I tend to say yes to too many one-offs so that's one thing we're an event-based company so There's literal deadlines with every single event. We've standardized, what we end up doing is we standardize topics. So then once it's standardized, well now it's just a production line event. And we've had that down to science for a long time, pre-AI.

4:20 The thing that the biggest deadlines are happening now around new topic development. So we can create new and it used to take us three months sometimes to create a new topic because so much went into it new landing pages, new ads, new new copy for emails and just all the pieces. Now it's just an interview with the person on the topic. put it the AI and I created a whole skill on new topic development where it will produce the

4:57 the outline and the slide deck for the topic based upon the the workshop formula and this belief transfer formula that we codified right so now it because what I keep trying to do is taking we keep removing the bottlenecks and I was a big one in that process and it's not perfect yet it still requires me to get involved and that takes time but those dead so those particular deadlines have been happening more because for example we just gave we just had three new topics launched

5:34 in a six-week period when normally that would we would do one every three months right and not only that but it would usually take three to four iterations of that particular topic testing the ads and everything to get a great result like to get a good result We just launched one. It's a longevity topic and it competed with one of our best known topics like right away off out of the gates with the first one. So that's what excites me too. How do we keep I'm still the bottleneck in those deadlines like how do I keep taking me out of those bottlenecks

6:19 Sam: It's interesting. It's interesting, too, because this idea of a bottleneck, right? There's probably different classifications for it. There's the bottleneck that you could attribute to. admin work drudgery like no one likes making a slide deck it takes forever and for someone to look at it for maybe two and a half minutes for the most part you know and then it's done and how much time went into thinking that through so like that type of bottleneck you can get to a point where skill you have like a format to do this And you get to a first draft in like one minute,

7:03 which is really crazy. And if you think about those types of online, there's all kinds of it, right? Like you have a whole inbox full of emails that you may or may not read, things that might be opportunities that would potentially be missed due to procrastination, due to life happening and you forgetting stuff. You can constantly re-up on that knowledge from that, always extracting. So a lot of the things I guess that we would procrastinate on. or drag out does go to zero which is cool but then there's other bottlenecks that are probably a good thing right so like anything that's going to fall into a creative space or judgment or something like that you actually

7:49 Maybe you even want to extend that bottleneck where because you don't have to focus on the drudgery, the admin, you get to reallocate that time towards coming up with a new topic, iterating, launching more things because you can get a read on the data quicker that you already have. You can pipe that back in, help it in the decision making process. I think like there's all of this compression that's happening and it's very obvious. The thing I'm trying to focus on is where is the human in the loop and where do we need to extend, you know, maybe even play periods or something where we're like really testing out new things, experimenting.

8:32 It's very interesting to see how things are moving towards this decision. base modeling which is actually pretty on the nose for lately if you if you jumped into the whole Jev phenomena thing that's going on right now aha

8:51 Steve: No, I don't know what that is.

8:53 Sam: so there's a new it's not really a new class of model per se but it is dear friend and it's picked up a lot of steam because normally you input some text and it's not a text you input context right so that's either going to be a transcribed you know call that you have maybe it's a screenshot maybe it's a document or a whole you know g drive full of documents maybe do some research on the web you pull that in And the output of that is one of those things, usually an image, right,

9:34 if we're doing image generation or usually text, right, a wall of text. So what Jev is, and I think it's named Jev after the Jevons paradox, which is basically if you make a thing. cheaper more efficient whatever it is it doesn't mean you'll do less of that thing it actually means you'll do way more of it right um so what this this is it's a decision model right so you would feed it a bunch of input and

10:09 the output is not text you give it like up to 255 options i think and it picks one so like easy thing would be like a boolean yes or no so you like feed it a whole bunch of like ad data and it'd be like is this a winner not a winner and it just rapidly classifies this for i think like a hundredth of the cost that you would normally spend and you can get results that are like hundreds of milliseconds latency so it's

10:35 go so fast right so you're not sitting there waiting for it to spit out text that's effectively yes with a whole paragraph afterwards um it just makes a decision so people are using this for all kinds of stuff right like i've seen um people playing chess right you only have so many moves you can make fixed um fixed number of options like anything that falls into that paradigm can be utilized in this way And there's probably some level of a decision tree that you could have where you have options on options. So like if you pick one of these 200 options, that one option has another 200 and you can play that game.

11:17 So I'm seeing people do all kinds of insane things with it, trying to push the limits of it, but it's super cheap. I think I'm probably going to start experimenting with it this week. I don't know how it works for you. Normally I have to hear about something a number of times before I'm like, all right, you dragged me into this.

11:36 Steve: Yeah, that's definitely me. I mean, before I made the jump from open AI to cloud and then from cloud. ai.ai over to cloud code like i had to hear like hey you need to get on cloud code probably three or four times and then now i mean i couldn't imagine the life without it like oh man that would if they took it away right now it's like oh god that would be and and isn't that funny though it's like man they they got us hooked

12:11 Sam: Oh, no, totally code. Like, And the problem with a lot of this stuff, so much problem, but you don't know what the thing is until you start pressure testing it, right? Like in the same way, like I've seen other people using this JEV model, I don't know what I would actually use it for. So I guess that is sort of that black box. I don't have clarity necessarily around what to use it for because it's so new. There's not that many paths forged for it. It's a huge level of uncertainty, right? So those types of open-ended things are probably those bottlenecks that we need to lean into, right? To spend a little time experimenting, failing, figuring stuff out.

12:55 But yeah, to your point with Claude Code, like. You hear about it a number of times people are telling you to use it, you start using it and you don't really understand why it's better. And then at some point you probably have an aha moment where you're like, oh. The computer is not the programs that run on the computer is a computer and a computer can do all these things like connect to the Internet, connect to my Bluetooth, my printer, whatever. And you get to just control these inputs and outputs directly from a terminal. But you wouldn't know that until you start using it.

13:37 So I'm very curious to see what happens from using this new model.

13:41 Steve: Yeah, so I'd like to explore that a little bit more because, you know, I teach people how to transfer their beliefs, like how to shift beliefs and for someone to accept your belief through a process of agreement and challenging the belief. And then part of it is showing how showing a result that's very relatable.

14:08 you know to something that they want right so then so what came up for me was well the best way for for me or so far the way I've adopted cloud code or anything like that was well one strong recommendations like I had to hear it enough times and hear some level of

14:29 relateability like oh man you could be doing this right like okay but i still couldn't what it was very it's very experiential right and it wasn't until i experienced it that i fully was all in but i'm curious like with this jeff thing it sounds like it might be very usable for me with something i'm working on right now so um I'm trying to get action steps and automated runbooks like with every like every meeting that I have. I probably five to six meetings per day at least.

15:12 And there's always an act there's always a next step or there should be. But right now most of that's manual so so I'm creating a process right now where we I get that and I wanted to think like Steve like okay what would Steve do here right so I'm trying to create this second Steve you know AI version of me that would then make the decisions based upon the category of the event like is this a this is an ongoing project or not is it a new thing you And then give me a yes no,

15:50 like, you know, approve those tasks or not, tell it which one was wrong and why or add anything and then learn from that for that category so that hopefully it improves over time. And then the next day, because right now what I find is I'm playing catch up. It's really great what's happening. Like I'll look at the calendar for today and like it's increased my efficiency. I'm able to have so many more projects. Like I could have never done this the way that it's happening right now. But in the morning I'm creating the HTML slide deck for that thing or I'm the Google Doc for that thing or the analysis of the data for this thing and the email.

16:32 But if all that was already ready to go, now I could have more of that time that you're talking about, that play time, to just have more of the big picture. Right now I find myself just in a decision making, that next thing creation thing. and a lot of it I'm the bottleneck because people aren't in my brain I could try to SOP that right for someone else to have a human try to take on some of that but I feel like so much of it could be really automated to a large degree

17:03 Sam: So the thing that I'm thinking about, and again, I know next to nothing about implementation for this model outside of like. It's called Cloud Code to help you set it up. It's with TypeSafe. AI is the name of the company. And you're creating decision-making processes. If you think less about what you do, because you as a human, your decision-making is incredibly complex, right? A few things stood out, though. You can imagine categorization, right? I doubt you would have more than 250 buckets of what you would group any of these different,

17:47 whether it's emails, initiatives, whatever it is. You have a fixed number of buckets within each bucket, maybe subcategories, right? If you could just rapidly classify. everything in your inbox, everything you're getting in Slack, you know, market research, right? That's like coming in in real time, like looking off the web and consolidate all of them into these fixed buckets where it's like, oh, I might not have thought to co-locate these things, but they fall into that same path on the decision tree. And then the simple Boolean, right? Yes, no.

18:26 Does do any of these require Steve to look directly into what's going on here to dig deeper? And I think it's one of those things where. You probably don't know how effective it could be at the beginning because those decisions have to compound over time. So focusing on the categorization feels like the thing that Jeb would potentially be useful for. Categorization of the team, of the topic, and then also whether or not it requires human eyes on it. It's just like a starting place.

19:07 Steve: Yeah, it's helpful. And I just started this today. It'll be interesting to see where it is tomorrow. And then I could ask it about that. That's a great way to get it started. One of the things you talk about decision making and having human eyes on it. On some of these projects I'm working on, like, I find that I actually do best with my eyes on it after,

19:37 you know, where I almost like, because I'm working on, it's because I see it as like a future project that I just want to, I want it to keep because it takes time to build these things, right? You know, some of these projects, if it's a bigger one, it might take a whole week of cooking, right?

19:56 Sam: For sure.

19:57 Steve: and I'm the bottleneck I don't have time to decide and I would rather like so hey and I told it that today I said hey I don't have time to make a decision here like you make the decision and I'll look at it after what are your thoughts on that

20:13 Sam: I do that every night. Every night I'm like, hey, I want you to. optimize seo you know just a whole bunch of extra i'll give an insanely long prompt that i'm just word vomiting at you know voice mode until it transcribes usually paragraphs and then i'm like and by the way i'm going to sleep right now so just do what you're going to do and we'll figure it out tomorrow weirdly the

20:42 approximation that it lands on is like always like i would always grade it at least a c plus you know what i mean like we're usually getting incredibly close uh and that is something that has changed dramatically this year like this idea of one-shotting

21:00 Steve: Yep.

21:01 Sam: They were agents were completely unreliable on this. They needed to have their hands held the whole time. But now, I mean, you have chain of thought, you have, you know, it having the ability to like in a Claude code setting, it has direct access to the computer. It's going to be able to read the logs that are being spit out. It's going to be able to make network requests. Do what it needs to do to get to the end. Yes. Sometimes it makes some very questionable decisions to get there, which is usually why having.

21:34 some engineering guardrails in some type of way like best practices even if it's as simple as running a particular skill or plug-in over whatever it spits out at the end but to that point of looking at it at the end I mean yeah it's just like a different way of work right like you're kind of delegating this is a force multiplication problem more than anything else where it's How do you take you and then average that out against engineer Claude and, you know, product designer Claude and you kind of have this team of sub agents working for you with you to get to some shared goal at the end.

22:19 but i think that's the weirdest part right like a lot of people are maxing on all these agents that they're running but really what they're doing is running a team and you can do that effectively or not uh you will get something out of it which i think is the interesting part because like i said like a year ago you let it go off the rails is going to end on an error page and just stop now it'll just keep cooking until it has something So it is it is different because it'll cook for a long time. Right. That sort of feeds into this ADHD parent, which where it might be a superpower now because context switching, you're like, I'm going to have this project go run off and do this thing.

23:02 It's going to take five minutes or it's going to take. 50. And in that interim, I'm going to go focus on this other project, which is probably leading to the, I have 40 projects that I'm working on at any given point in time, which is both really cool and also kind of a curse at the same time. So, I mean, that's what I think on it.

23:22 Steve: Yeah, so that's why I want the, I mean, I want as close to finished as possible. in one shot so now it's just decisions around styling function features to fine-tune and the ADHD superpower I do think it is very much a very useful thing because I can just jump between us all but the trap that I keep getting into is is not

23:55 setting aside the time for completion like it can less it's a has a strong deadline I'm not and those deadlines are usually exterior there's some consequence right but I have all these projects so close to completion so I think the thing for me I personally need to just set and be strict like okay this is completion hour this is Okay, I'm going to complete one of these projects, one of these things instead of leaving it at 85, 95 percent.

24:29 Sam: I mean, that's always been a problem, though, right? Like the last 10% usually takes 50% of the time. It's, yeah, the finishing part is tough, too, probably even more so because because you can move so quickly. You don't have the sunk cost fallacy of I just burned three weeks on this project. I need to finish it. It's kind of like, well, it was a good idea. And then you kind of let that fade because you don't have the buy-in. You don't have the grind, the sweat and the tears over all of the previous time leading up to that. So there is kind of this weird discounting of work that happens too,

25:12 where it's like you have all these projects, but they're all less valuable at the same time. So then you're like, where's the value metric there?

25:21 Steve: Yeah.

25:21 Sam: Is it the totality of it?

25:22 Steve: As you're talking, I'm thinking, okay, I need a visual. I'm so visual, right? I need a visual to see all the projects and a tier ranking like where I can. you know move them up or down on the scale um also have some some because i'm a visionary adhd in mind like i need i need to be reined in and i i'm i'm self-employed right i i don't have anyone telling me like hey steve like get your ass back into here you know take care you know i'm the ceo of the company i'm but also responsible for sales i don't have anyone telling me like hey how How's sales going this week? Do you have, you know, where are your KPIs and metrics?

26:08 And I don't know, maybe I need to just get someone, I need to hire someone to boss me around.

26:15 Sam: I mean, you can probably make Claude do it. Probably.

26:20 Steve: Right.

26:21 Sam: I mean, knowing that this is being recorded, you're probably going to feed the transcript into something later. I mean, I would 100% visualize this with an artifact. Simple as that. Visualize, like look at all of the repositories I have on my computer. visualize all of them and then when you're looking at all of them probably spin open a new voice tab where you're just kind of tearing down all of the projects and maybe even doing assignment of well you know this thing is a nice to have project this one's actually generated income this one potentially could and just like stack rank everything based on that

27:04 uh that's just like a starting point i've i don't know how much you've been using like the artifacts for me are really nice because they're just throwaway html so you get like the the information sharing that you'd get from a pdf but the flexibility of like a raw html page so you can put anything on there i've done all kinds of them because I think most people are visual learners right so for the most part

27:35 Steve: Yeah, I very much prefer the artifacts, HTML presentations versus the Google Docs and things like that. It's quick, it's easy, it's really... and like even for like proposals like people are very impressed with it too like i take i have a meeting from the previous week oh here's a proposal like oh wow that's that's amazing and i feel like it's like it's they sell themselves because people are wild betting got everything and it captured it and they can now see it too not just read it and it makes sense

28:15 Sam: yeah the speed thing I think is becoming the differentiator there where it's like if you are not if you have to be dragged along to use these tools you're gonna have a really hard time because everyone else who stopped setting up slide decks manually like you're you're going from just you have an idea and then you need to execute and build on this idea and the deliverable of that is a slide deck that's consumed in two seconds it's a lot better to just have idea to visual representation so it takes you two seconds to build it and then two seconds for it to be digested it's just value transfer at the end of the day or belief transfer depending on you know what leverage you're trying to pull there Uh,

28:59 but yeah, like there's so much of the way we work is changing. Um, I, I think we're almost approximating something that is what it means to be a human. Like, what are we building towards? Right? Like, what are we even doing here? Um, And when you take a lot of the drudgery away, when you take a lot of the building pieces that aren't actually fun, they're pretty repetitive, tedious, have to get done to get a product or at least that's typically how it's been. You collapse that to zero. Now they're all just little creative emotional creatures walking around the world just making decisions, having taste.

29:43 And I don't know, like what what is the next iteration of work? One thing that I've been pushing a lot is kind of in a way what we're doing here, but what you can do with other folks, too, is to just have these calls where the end of it is a deliverable of some kind like you've talked about stuff that you want to get done. Vast majority of those action items can just be done right by an agent.

30:11 Steve: Right.

30:11 Sam: So when you think about that mindset, then, well, now we're in a call. I'm not taking notes, right? I am just completely engaged in the conversation. We're exchanging ideas, brainstorming, you know. changing our minds in real time and getting towards something and then as soon as you're done with that you have the entire transcription the entire record of the whole thought process that led up to it and then you get to decide how that gets implemented sometimes it takes two seconds which is an incredible feedback loop

30:47 Steve: I find, yeah, there's a couple of things that came up for me as you're sharing this. And I think like our next session, we talked about like the creation and the quality of the creation and the pain points that I'm seeing and that you're seeing. I don't know if we go into it today because I think it's a deep topic, but everything on social media.

31:16 everybody's ads everybody's flyers everything is looks so the same and and just know it's and and I don't know if I'm more sensitive to it but I have to assume that the rest of society who aren't doing a lot with AI they they know and some of it looks really good but then some of it's so much the same but then I was presenting a proposal on you know that I made with a slide deck and there's one of those things that I'm producing fast I didn't proof it well there was so much extra stuff on it and I couldn't even like

31:53 It just so now I know that I need to give it more guardrails. I need to refine it. I think that's that's probably where society is going to go in this next phase. Right. You know, somewhere around there.

32:05 Sam: Well, yeah, we made building effectively free. So then you have to decide what to build and more importantly, what to subtract from what was built. But yeah, definitely a topic for another time because we could go so deep on that. This was a good dive.

32:22 Steve: Yes, this was. And I'm going to take what we did here. I'm going to build some stuff. I'll try. I'll see if I can get the visualization down before next week and also that thing I was telling you about that I want to build with just my daily. routine right it's a daily routine at the end of day and begin a day and see how much of that I can automate and I'll let you know what I wanted to come up with

32:46 Sam: And I'll try and get my hands dirty with Jev, this new model, and see what I can push it to do if anything interesting comes up.

32:57 Steve: Yeah, that's cool. I like getting in with the deliverable and then you know what we kind of need is like we both share next time and just pull it up like all right, here's the thing.

33:06 Sam: A little show and tell.

33:07 Steve: Yeah, yeah, why not?