×
Evaluate Your Team's AI Readiness

Take our official assessment and explore training paths to build secure, compliant AI workflows.

Access Portal

AI at Work Podcast

Ep #9 Why Some Early Adopters of AI Are So Stressed (The Busywork Paradox)

AI at Work, AI Fundamentals, Featured Podcast

Audio episode

Listen to this episode

Summary

Elijah and Kevin kick off the new year with a blunt reality check, AI is not just “a tool,” it is reshaping how work feels. They unpack a surprising downside of automation: when AI removes the low stress busywork, people can end up stuck in high stakes decision mode all day. They also break down why “vibe coding” is getting confusing, the difference between developer-grade tools (Cursor, Windsurf, Claude Code) and consumer app builders (Replit, Opal, Lovable). Finally, they dig into the messy middle of apps, connectors, context windows, and why cross-checking outputs across models is becoming a new baseline skill.

Takeaways

Automating “minutiae” can raise stress, you lose the natural recovery moments in your day.

AI transformation is rarely just one department, sales bleeds into ops, finance, and everything else.

Vibe coding now means two things: pro dev tooling vs consumer prompt-to-app tools, and they break differently.

Bigger context windows help, but they also increase drift and confusion unless you add checks and structure.

The new power move is verification: multiple chats, multiple models, and feedback loops, not blind trust.

Chapters

00:00 Intro + New Year banter
01:21 Focus, presence, and productivity intentions
04:07 “The time before”: why this AI moment feels different
05:30 When AI removes busywork, and raises stress
08:30 Why AI adoption can’t live in silos
11:05 Where most organizations really are with AI
13:31 Multitasking with AI: the “stew pot” workflow
16:55 Inconsistent models and building on shifting ground
19:38 Vibe coding: dev tools vs consumer no-code
22:43 Singularity talk and the importance of domain expertise
25:41 GPT apps/connectors: promise vs reality
32:09 A real win: Claude + mobile reminders
39:57 Bigger context windows, same verification risks
48:54 NotebookLM and workflow tools that actually help
57:44 The new pace of work, and what it means going forward

 

 

 

Explore practical AI implementation: https://assessment.ascendlabs.ai/

Book a conversation with Kevin: tidycal.com/kevinwilliams

Show Notes & Links: https://www.spark6.com/podcast

Submit listener questions:

[email protected]

[email protected]

Check out Kevin’s stuff:

AscendAI: https://ascendlabs.ai/

Follow Kevin on LinkedIn: https://www.linkedin.com/in/kevinguywilliams/

Check out Eli’s Stuff:

SPARK6 Agency: https://www.spark6.com/

Sign up for FREE AI Framework Friday Newsletter: https://www.spark6.com/newsletter

Follow Elijah on LinkedIn: https://www.linkedin.com/in/elijahszasz/

Transcript and citation details

Transcript

00:00 Play from here Kevin, I just can't help but dance to that intro music.

00:14 Play from here I don't know about you.

00:15 Play from here It's fantastic.

00:16 Play from here It's so AI-rific.

00:17 Play from here It's lovely.

00:18 Play from here It's so AI-rific.

00:19 Play from here It's so AI-techno club.

00:22 Play from here Happy New Year!

00:23 Play from here Greens, did you stay off of LLMs?

00:25 Play from here What do you think?

00:26 Play from here Of course not.

00:27 Play from here Not even close, man.

00:28 Play from here I bet you, I know you took a little vacation, I bet you brought your laptop and were just

00:33 Play from here like grinding away at every opportunity that your family wasn't beating you down on.

00:36 Play from here You know, having an opportunity to play with stuff is so rare, like the day-to-day is always

00:42 Play from here fierce.

00:43 Play from here You know what?

00:44 Play from here I always laugh when clients will tell me, oh, I just don't have time for you to save

00:48 Play from here me time.

00:50 Play from here And here I am, I kind of have the same problem.

00:52 Play from here And I have the advantage of trying to build systems that are AI-first, right?

00:57 Play from here I'm still...

00:58 Play from here You're not always eating the dog food, are you?

01:00 Play from here Well, I'm eating the dog food, but I'm still awash in communications and things like that.

01:05 Play from here And having everybody just go away sort of blissfully gave me a bunch of time to play

01:10 Play from here with things that I've been wanting to play with.

01:12 Play from here I want to hear about some of these things, but I want to know first if you have any New

01:14 Play from here Year's resolutions, because, I don't know, there's like two groups of people, those who

01:20 Play from here do that stuff and those who don't.

01:21 Play from here I'm trying to decide which one I would, which box I'd put you into, and I'm undecided.

01:25 Play from here So do you have any?

01:27 Play from here I'm not really that guy in general.

01:30 Play from here I would say that I did make some resolutions about productivity and focus and being present

01:37 Play from here in general, that in just a world that's just full of distractions.

01:40 Play from here I'm not a big social media guy or anything like that, but it's just so easy to get wrapped

01:46 Play from here into something as big as what we're dealing with here.

01:50 Play from here It really is.

01:51 Play from here And I've found myself just thinking about this stuff 24-7, which surprised nobody that

01:56 Play from here knows me, but I'm in danger of becoming sort of monodimensional.

01:59 Play from here I sort of laugh that whenever I have an opportunity to brief somebody on AI, I'm saving my wife

02:05 Play from here the pain of having to hear about it from me.

02:09 Play from here Isn't that the truth?

02:10 Play from here I, yeah, sometimes have to catch myself mid-monologue and then just censor the whole conversation

02:16 Play from here because I can see the eyes glassing over, picking up on the subtle social cues like

02:25 Play from here her leaving the room, just walking away.

02:29 Play from here So yeah, my wife's a therapist and AI hasn't taken her job yet, but she came up with this

02:34 Play from here thing that I think she saw from some witch on Instagram, which was on the solstice, you

02:39 Play from here write down 13 things that as if they've already happened, like an intention that you want

02:44 Play from here for this coming year, but it's as if it's already happened.

02:47 Play from here And you fold them up so that you can't see them anymore.

02:50 Play from here And then every night, starting on the solstice, you burn one of these and you don't know which

02:54 Play from here one.

02:55 Play from here And then on New Year's, you open that last one and that's supposed to be your North Star

02:58 Play from here for the year and all the rest, release into the universe of the AI gods, whatever you

03:03 Play from here believe in.

03:05 Play from here And yeah, so that was kind of a fun way to get cajoled into something that I would have

03:10 Play from here ordinarily done as far as intentions or resolutions or anything like that.

03:15 Play from here But I agree on just the presence thing in general.

03:18 Play from here There's just so much happening right now, especially in this world that we operate in,

03:21 Play from here that it is so difficult to just kind of pay attention to what you're doing and not get

03:25 Play from here distracted by the next shiny object.

03:27 Play from here Which to your point, the break is a good time to do that because you actually have the time

03:31 Play from here without all of the regular Zoom calls and everything else and client work where you

03:36 Play from here can start to play with this stuff.

03:37 Play from here So any big things that you played with or big ahas that you had over the break?

03:41 Play from here I wanted to just add one more piece to that.

03:43 Play from here I said this to my wife the other day that I have this feeling that we're in the time

03:47 Play from here before that we've been doing this for a long time.

03:51 Play from here We've been drinking the Kool-Aid.

03:53 Play from here Some people believed it.

03:54 Play from here Some people didn't.

03:56 Play from here I'm absolutely feeling a turning of the corner such that everyone's in there in on some sort

04:03 Play from here of continuum.

04:04 Play from here But they get that even if most of the promises turn out to be false, this has the potential

04:10 Play from here to really change so much, not just about work and the way that we do things, but about society

04:15 Play from here and the world and et cetera.

04:17 Play from here And you can think back to other times in history where these sort of purgatory periods.

04:24 Play from here And I think that we're in one of those.

04:26 Play from here And in 10 years, when we are grayer and older, we're going to look at this particular period

04:32 Play from here as the time before.

04:33 Play from here I'm wrestling with that a little bit as far as what it means.

04:36 Play from here Obviously, it has ramifications for our businesses and everyone's businesses.

04:41 Play from here I had this anecdote yesterday in a workforce special interest group that I run in Utah.

04:46 Play from here And a woman had brought up a friend of hers that's a paralegal at a major enterprise-based

04:54 Play from here company that I won't name.

04:56 Play from here And this company is leaning hard into AI and AI implementation to a degree that's pretty

05:01 Play from here impressive to be honest.

05:03 Play from here And the paralegal made the observation that her job, all of the minutia of her job is

05:08 Play from here now pretty much gone away.

05:10 Play from here All of the filing and the little pain points.

05:12 Play from here And these are the things that we tell everybody, hey, you can get rid of pain.

05:15 Play from here You don't need to do those stupid things.

05:17 Play from here Start with the stuff you hate to do.

05:18 Play from here Yeah.

05:19 Play from here Start with the stuff you hate to do.

05:20 Play from here Yeah.

05:21 Play from here Okay.

05:22 Play from here But the observation was now with all of the minutia gone, every iota of her day is spent

05:28 Play from here on high impact, high meaningful, kind of stressful work.

05:33 Play from here So she's in stress mode all day because usually there'd be recovery.

05:40 Play from here There might be some drama that's going on with a contract.

05:43 Play from here You're moving data in Excel sheet cells, basically.

05:46 Play from here Yeah.

05:47 Play from here But that's almost a reset.

05:48 Play from here But it's low stress.

05:49 Play from here It's low stress.

05:50 Play from here It's low stress.

05:51 Play from here It's just kind of staring at the computer, moving stuff around.

05:52 Play from here But what happens to us if all of our time is then spent in like alpha mode, full on

05:59 Play from here decision mode?

06:01 Play from here As entrepreneurs, that's sort of where you live most of the time.

06:05 Play from here But I don't think that a lot of people are particularly prepared for that.

06:07 Play from here And I don't think-

06:08 Play from here Your mortals don't really deal with that so well.

06:10 Play from here Well-

06:11 Play from here You're not the first person I've heard bring this up.

06:13 Play from here I've heard this a couple of times now.

06:15 Play from here And I think that, I mean, there's really something there.

06:19 Play from here And I think it's part of that evolution of how everybody begins to reconcile with all

06:24 Play from here this, right?

06:25 Play from here On one side, you have, oh my goodness, are there even going to be paralegals?

06:28 Play from here My job is going to go away.

06:30 Play from here And I think what you witnessed, and so many other people are experiencing, is that that's

06:35 Play from here not really the case very often, and you are going to be doing much more high impact work

06:40 Play from here that could also be just a much greater demand on your energy, on your focus, on every aspect

06:49 Play from here of your day.

06:50 Play from here Like you might just really come home feeling completely beat up because you didn't have

06:53 Play from here those down times of just the minutiae that you're used to, that you can kind of do in

06:57 Play from here your sleep, between the really big decision-making or stressful calls, or what some people just

07:03 Play from here might call their zone of genius, right, where you're kind of putting everything into a certain

07:06 Play from here amount of time.

07:07 Play from here So I think it's just one of the adjustments everybody's going to have to learn to deal

07:11 Play from here with as we figure this out.

07:14 Play from here And yeah, to your point about this being one of those before times, and we're going to

07:19 Play from here look back in 10 years and be like, oh, wow, do you remember that when we were trying to

07:22 Play from here figure all this stuff out?

07:23 Play from here And here we are.

07:24 Play from here I think what makes it so different than, and maybe not, like maybe people had these

07:29 Play from here same feelings during the agricultural revolution or the industrial revolution or the birth

07:34 Play from here of the internet.

07:35 Play from here But I feel like there's so much of this factor of we don't know until we get there right

07:40 Play from here now.

07:41 Play from here I mean, and it's always been there, but to a degree that is really extreme right now.

07:45 Play from here Yeah.

07:47 Play from here And before times, you know, both of us lived through 1999, 2000, and there's sort of that

07:52 Play from here feel as well, where it's like this discovery of trying to figure out what the right approach

07:59 Play from here is in an entirely new technology.

08:02 Play from here And it can be really, really challenging to find the right thing.

08:06 Play from here What is the right thing to focus on in an organization?

08:10 Play from here And what I find in conversations is the right thing is different from most organizations,

08:15 Play from here that there's some basics, like you need to have literacy, you need to have leadership

08:19 Play from here that's aligned, et cetera.

08:22 Play from here But it's not like the market has matured enough that you can say, this is the thing that I

08:27 Play from here do, and it's going to be relatively straightforward to solve.

08:31 Play from here The technology just permeates everything in an organization.

08:35 Play from here So if you try and lean in and solve just like a sales problem, okay, we are going to bring

08:40 Play from here a hammer, an AI hammer to sales, and we're going to figure this out, very, very quickly,

08:46 Play from here you realize that sales bleeds over into operations, bleeds over into finance, et cetera.

08:51 Play from here And if you haven't considered how that's all going to work, it just isn't going to work.

08:55 Play from here Right.

08:56 Play from here You almost need this kind of holistic view of everything to make these singular decisions

08:59 Play from here in some cases.

09:01 Play from here It's really wild.

09:02 Play from here I think another thing that really differentiates this between other super cycles that we've

09:06 Play from here seen in human history is just the rate of change as well, right?

09:10 Play from here The pace is exhausting, for lack of better words.

09:13 Play from here It really is.

09:14 Play from here So it's not only deciding, well, where do I apply this?

09:17 Play from here But what tool do I use?

09:19 Play from here And okay, after a bunch of research and talking to experts like you, who are really good at

09:25 Play from here deciding where in an organization a certain tool might be applied, all of a sudden two

09:29 Play from here weeks later, actually, you know what guys, I've got a whole new tool set because all

09:34 Play from here of these things happened, and now this is no longer relevant.

09:39 Play from here And then that might switch again, a month later, because another frontier lab comes

09:43 Play from here out with something mind blowing and just an order of magnitude better than the previous

09:47 Play from here tool.

09:48 Play from here So that's another thing that I think really trips people up.

09:51 Play from here I know it's tripping me up.

09:52 Play from here It's how I spent a lot of my break is evaluating a lot of tools from the different frontier

09:56 Play from here labs and really waffling with a lot of things based on just the nuance of what I could do

10:02 Play from here with specific workflows, given the features of these different applications.

10:07 Play from here It's a lot to navigate.

10:09 Play from here And we're living and breathing this, you know, outside of a few rare organizations that have

10:14 Play from here just gone full in on it and have really identified their use cases and started to build the capacity

10:21 Play from here to solve them.

10:22 Play from here I'm finding that nine out of 10 organizations are still on the very beginning stages.

10:29 Play from here They again are beyond that, should we or shouldn't we, they get that they should, you know, they

10:33 Play from here get that now we have to really lean in, but it's very rare to see people to go beyond

10:40 Play from here just like simple chatbot type applications and start thinking about solving real problems.

10:46 Play from here And then if they are solving real problems, they tend to be doing so in a way that just

10:51 Play from here replicates whatever the humans were doing before, as opposed to thinking a little bit

10:55 Play from here bigger.

10:56 Play from here Now there's nothing wrong with that.

10:57 Play from here That's a great place to start, but that's still where we are.

11:00 Play from here You're not doing like grand transformations like some people think is going on, I would

11:07 Play from here say.

11:08 Play from here Or even leadership using some of these models for real strategic thinking, using some of

11:14 Play from here these models as a companion, using deep research.

11:17 Play from here Some of the things that are available to everybody, like these just off the shelf features that

11:22 Play from here anybody can use, yet I think it's still very rare that anybody does use them anywhere close

11:27 Play from here to their capabilities of how they can be helpful in different organizations.

11:30 Play from here That's another thing.

11:31 Play from here I don't know if you're seeing that as well, but it's something I've experienced.

11:34 Play from here It's very superficial and prompting is still important.

11:38 Play from here Context is still super important and understanding the limitations of both is important.

11:44 Play from here And without understanding, particularly busy people don't get deep enough to get to the

11:49 Play from here unlocks.

11:50 Play from here They haven't committed those dozens of hours of just leaning into it such that it's a constant

11:56 Play from here companion that's solving multiple problems at once.

12:01 Play from here Simultaneous workflows.

12:02 Play from here There's another piece of human behavior that we've been told all through our professional

12:06 Play from here careers is a really bad idea, that you shouldn't be multitasking, you should be focusing in

12:10 Play from here on things.

12:11 Play from here But the latency that's involved in using these tools to the nth degree, at least for me,

12:17 Play from here has me doing 10 things at once because I'm literally running deep research on another

12:23 Play from here monitor right now that I know is going to take 25 minutes to do.

12:28 Play from here And I just have to mentally remember to come back to that later in the day.

12:33 Play from here It's like making a stew or something or a crockpot dinner.

12:37 Play from here I'm going to start this thing and it's going to go off and it's going to do research and

12:43 Play from here then I need to check in on it.

12:45 Play from here I need to stir the carrots a little bit and make sure that it's proceeding.

12:50 Play from here But that is a totally different style of work for anybody who's not an air traffic controller.

12:57 Play from here Maybe that's the corollary.

12:58 Play from here That we're all like air traffic controllers who are constantly having to do attention

13:01 Play from here switching between different flights and different verticals.

13:06 Play from here And if that's not resonating to you who are listening to this, you're probably not doing

13:11 Play from here it right to a degree.

13:14 Play from here It's a problem, but it's how you get the most out of these tools is to realize where you're

13:22 Play from here pushing them into a project and letting them do their thing and where you're real time

13:27 Play from here going back and forth and which part of the project you're basically fully automating

13:32 Play from here behind the scenes.

13:33 Play from here One of those things we might look back on is, do you remember when you used to have

13:36 Play from here to wait for deep research?

13:37 Play from here Wasn't that crazy?

13:38 Play from here And you'd have to remember to come back to it.

13:40 Play from here And I'm laughing with you, not at you, because that is my life and I'm doing it across multiple

13:45 Play from here models and I'm even doing the same research across multiple models because I want them

13:49 Play from here to then verify each other's work and poke holes in each other's outcomes.

13:53 Play from here And one thing that saves me is I set the reminders, the notifications on mobile, so that even

13:59 Play from here if I step away when it's done with this research, it sends me a notification and says, hey,

14:03 Play from here I'm done because yeah, there's always so much going on and I've got so many things running

14:08 Play from here at a given time that I'll just completely forget and come back to it and, oh, wait,

14:12 Play from here that browser tab was open and I had all this research running and I forgot about this task.

14:17 Play from here The notifications have been helpful for that.

14:19 Play from here Like that's, that's, that's been very helpful.

14:21 Play from here Yeah.

14:22 Play from here So over the break, yeah, I did have some protected time as well to go deep on a few things.

14:27 Play from here And like you said, we're in the middle of this and you feel like we're in one of those

14:31 Play from here moments where it's the look back moment, like things are really starting to shift.

14:35 Play from here Everybody knows that they're going to have to start embracing this stuff if they haven't already.

14:39 Play from here the acceptance part and I feel like for people deep in it, some other things

14:44 Play from here are starting to shift as well.

14:46 Play from here You had sent me a message about code debugging.

14:49 Play from here For me, a break away from work is a time where I can like sit down and build

14:52 Play from here things that I want for myself or my business and so I got to do a bit of that

14:55 Play from here as well and you know, one thing I'll say is cloud code has like just gotten

15:00 Play from here completely bananas, like there's weird stuff going on and that's, that's another

15:05 Play from here thing that people who don't spend a lot of time with this usually don't realize

15:08 Play from here if they're just kind of doing just the high level of prompting and trying to

15:11 Play from here get through things is they won't pick up on the nuance of something working

15:15 Play from here completely different, even though there hasn't been an announcement from a

15:18 Play from here Frontier Lab or there hasn't been like release notes or anything else, something

15:21 Play from here just all of a sudden feels different or it's acting different and you and I talk

15:25 Play from here about that all the time, right?

15:26 Play from here Like, wow, this thing just started swearing at me or this, this got really

15:31 Play from here personal or this had great insights and whoa, this just got super sycophantic

15:35 Play from here again and I thought they trained that out of the system prompt.

15:37 Play from here What happened, right?

15:38 Play from here We notice all those things and so that was one of the things like the cloud

15:42 Play from here code stuff has gotten pretty crazy.

15:45 Play from here I don't know if you caught, Elon Musk had tweeted, we have entered the

15:48 Play from here singularity, which is a pretty bold statement, albeit coming from a pretty

15:52 Play from here bold guy, but still, you know, when you talk about big inflection points as to

15:56 Play from here where we are and where this is going with work saying that we've entered the

15:58 Play from here singularity, that's pretty crazy to come out of the holiday break and see.

16:03 Play from here I've got a couple of different reactions there because it feeds off of spending a

16:06 Play from here couple of weeks with people who are, I don't know, muggles in, in AI, as it

16:12 Play from here were, like family members and things like that, that you're not drinking the

16:16 Play from here Kool-Aid and man, I could imagine I'm just exhausting to be around, but trying

16:21 Play from here to like, a minute and you're exhausting to be around, to, to untangle some of

16:26 Play from here these pieces and, and some of the things you just said there are really important

16:29 Play from here for people in our audience to understand.

16:32 Play from here One that strikes me is this idea that it's not that you're building on sand

16:36 Play from here because that's not totally fair, but you're building on a landscape

16:41 Play from here that has opinions and it changes.

16:43 Play from here And I don't think that most organizations have really grappled with this yet.

16:48 Play from here And I've mentioned this on previous shows, but it's this idea of deterministic

16:52 Play from here versus probabilistic outputs and what that means from a strategic perspective.

16:56 Play from here Like it makes people's heads melt when they think that, oh, wow, I built something

17:00 Play from here today, it's working, this is the SOP, this is the way we're going to do it.

17:04 Play from here And then it's a Tuesday.

17:05 Play from here And for reasons that you can't define, it's now going to

17:09 Play from here decide to do different things.

17:10 Play from here And those different things maybe aren't as reliable.

17:14 Play from here Maybe they aren't as accurate and you go up to your little model selector.

17:18 Play from here In our case, it's going to be via API or whatever, but in, in most consumers

17:22 Play from here cases, they're going to be in CLAWD or they're going to be in GPT and you see,

17:25 Play from here oh, I'm in 5.2, why the heck is this so different right now?

17:31 Play from here And that lack of consistency is a very real problem when you're leaning into

17:37 Play from here doing solutions development for clients or internally or whatever it is.

17:41 Play from here And it takes a much different mindset to grapple with that.

17:45 Play from here It's all very organic and hazy.

17:49 Play from here Yeah.

17:49 Play from here I mean, it's a problem for me.

17:51 Play from here I mean, not just the organizations I work with, but it's, I mean, it's a, it's a

17:55 Play from here problem for me when things start acting completely different without notice.

17:59 Play from here Right.

17:59 Play from here And then I go to another tool and I'm like, oh, this is so much better.

18:03 Play from here And then I remembered it.

18:04 Play from here Maybe I should try this in the tool that I was in before and seeing what it's doing

18:07 Play from here now, and it's doing something else completely different as well.

18:10 Play from here So yeah, that's a, that's a real challenge.

18:13 Play from here Like that is, that is a real challenge.

18:14 Play from here Something else that I think people will find useful is the way that, that Eli is

18:20 Play from here going to talk about, you know, using CLAWD code versus say vibe coding.

18:25 Play from here This is a definition that's getting really slippery.

18:28 Play from here And I think that people need to pay attention to.

18:29 Play from here That we, we've, we've talked about vibe coding before, but it means

18:33 Play from here different things to different people.

18:35 Play from here And I think for our audience, splitting vibe coding into two different pools that

18:40 Play from here they can address in separate ways is important.

18:42 Play from here The first, and what he's likely talking about is development vibe coding.

18:47 Play from here This is using tools like Cursor or Windsurf in a coding environment.

18:52 Play from here And if you have a dev team, they're going to be very conversant with this.

18:57 Play from here And it can range, it can range from basically auto-complete to doing huge

19:01 Play from here blocks of code to with CLAWD code now being able to operate semi-autonomously

19:06 Play from here for up to five hours, which is sort of hard for non-developers to conceptualize.

19:11 Play from here But it can actually go off and it can run code on its own for a number of hours.

19:15 Play from here And you can see how that could be a huge unlock for a, for a small dev team

19:19 Play from here to be able to get stuff done quickly.

19:21 Play from here However, there's a whole different type of vibe coding, which is for, it's being

19:26 Play from here presented to mere mortals as something that you can just use human language to

19:31 Play from here produce an output that you can use to create a little app and that's using

19:35 Play from here tools like Repl.it or Google's Opal, Lovable, et cetera.

19:40 Play from here Which many are powered by CLAWD code, just to note.

19:43 Play from here They're powered by CLAWD code, but it's, it's different.

19:45 Play from here It's a different thing and it is a different level of skill.

19:49 Play from here So we totally need to do like a, a 2026 challenge type thing.

19:53 Play from here And I totally would challenge anybody listening to this to go

19:56 Play from here out and try and vibe code things.

19:58 Play from here You will discover that there's a lot of pain that you're think, that you think

20:02 Play from here you're avoiding through doing just the consumer grade vibe coding that would

20:06 Play from here be easier to avoid if you had the skills to use the development type tools.

20:12 Play from here Although I see people bridge the gap, like non-technical people who have

20:16 Play from here leaned into things like Cursor.

20:19 Play from here It's not a particularly easy learning curve if it's not something you're used to.

20:23 Play from here So instead what people try and do is muscle their way through vibe coding

20:27 Play from here solutions that are missing pieces.

20:30 Play from here And they end up in a world of hurt because they can't fix the

20:34 Play from here things that are wrong there.

20:35 Play from here And the vibe coding consumer tools basically have the memory of a flea.

20:40 Play from here So you might fix something in hour one and be delighted.

20:44 Play from here And then in hour eight, it decides to break that and not tell you, and it has

20:49 Play from here no memory of what it did in hour one.

20:52 Play from here So you end up with a, with a compounding effect of errors.

20:55 Play from here They make a lot of these tools really, really hard to use, but that's important

20:59 Play from here for you, for listeners to understand the difference between those things.

21:02 Play from here And I think we're probably going to see new terms for this because it's highly

21:06 Play from here confusing and it means wildly different things to the one person versus another

21:11 Play from here person.

21:12 Play from here Yeah.

21:12 Play from here And it's the former, not the latter that inspired that tweet from Elon, right.

21:16 Play from here As far as the, the things that are happening with cloud code and him saying

21:20 Play from here something like we have entered the singularity means that the, the, the,

21:25 Play from here the raw coding abilities of this model are getting so good that people are now

21:29 Play from here kind of looking ahead and saying, okay, we're now entering this era of recursive

21:33 Play from here code where if something can code that well autonomously, then it's the, the

21:39 Play from here models are very soon going to be self-improving or that's the theory.

21:43 Play from here And that's kind of the entry point for the singularity.

21:46 Play from here Or in other words, this event horizon that we all cross where there's this

21:50 Play from here completely unpredictable outcome on the other side, as far as where does this

21:54 Play from here technology go?

21:55 Play from here Is it going to run away?

21:56 Play from here Can we fully get our arms around the whole thing?

21:59 Play from here A lot of speculation in different directions there, but it's, it was really

22:02 Play from here interesting to first of all, see this stuff and read that tweet and just kind

22:06 Play from here of think about like what's going to be happening in 2026 here.

22:09 Play from here The time before.

22:11 Play from here Before, right?

22:12 Play from here Yeah.

22:12 Play from here Yeah.

22:13 Play from here And a quick note to what you were saying as well with just the difference between

22:16 Play from here like this AI assisted coding and then what most people call vibe coding, which

22:20 Play from here is this conversational prompting where you can make a really pretty semi-functional

22:26 Play from here prototype of something that you would have never been able to do a year ago.

22:31 Play from here Right?

22:31 Play from here Like that's, that's, that's very cool, but it's not really quite at, oh, these

22:35 Play from here are finished projects that any non-technical person can pick up.

22:38 Play from here I think the difference of somebody being able to do that and bring it across the

22:41 Play from here finish line is kind of like the story you were telling about the paralegal that you

22:45 Play from here were working with at that firm being that she's now being put into all of this

22:49 Play from here really intense work with no downtime because she has a domain of expertise.

22:54 Play from here Right.

22:55 Play from here And she couldn't have even used those tools to get rid of all the busy work

22:58 Play from here without that domain of expertise.

23:00 Play from here And it's the same thing with people who are building software, building

23:04 Play from here apps, doing the vibe coding.

23:05 Play from here The people who are getting really far have a domain of expertise.

23:08 Play from here They know a good output from a bad output.

23:10 Play from here They can go back and debug things.

23:12 Play from here And your regular, just mere mortal who's like, Hey, I'm going to build an app.

23:17 Play from here I've never done this before, but I can have a conversation.

23:20 Play from here So I'm going to put this prompt in.

23:21 Play from here We'll often run into problems because they don't come

23:24 Play from here from that domain of expertise.

23:25 Play from here And we're going to see that everywhere across every industry.

23:28 Play from here You're speechless.

23:30 Play from here Speechless.

23:30 Play from here Speechless.

23:32 Play from here Yes.

23:33 Play from here Yes.

23:33 Play from here You are.

23:34 Play from here Yeah.

23:36 Play from here Yeah.

23:36 Play from here Just again, pausing to think.

23:38 Play from here I know, man.

23:39 Play from here It's like going back to how you spent most of your break, which is inside of your

23:42 Play from here head, like all of this stuff going on.

23:45 Play from here There's a lot of just open-ended trippy stuff happening right now.

23:49 Play from here I'll tell you one like open-ended trippy thing I came across.

23:52 Play from here Have you played around much with the GPT apps yet, which are basically

23:57 Play from here they've replaced connectors.

23:58 Play from here Now we don't have connectors.

23:59 Play from here We have apps.

24:00 Play from here And we'd spoken before about how the connectors were not so great.

24:04 Play from here Sometimes they'd work.

24:05 Play from here Sometimes they wouldn't.

24:06 Play from here Most of the time they wouldn't.

24:06 Play from here Or it was really difficult to find out what the utility and the connector was.

24:11 Play from here You could even do a little recap for what is a, what is a connector?

24:13 Play from here What is an app?

24:14 Play from here I don't know how much you've played with this ecosystem.

24:16 Play from here I have to admit that I had a really rough few weeks with GPT.

24:19 Play from here That there's something that has changed in 5.2 that just does not agree with me.

24:24 Play from here So you can tweak your personal settings in there to have different approaches,

24:28 Play from here more formal, less formal, more verbose, less verbose, but there's this reinforcing

24:35 Play from here or reaffirming behavior that it's adopted.

24:39 Play from here That drives me absolutely bonkers.

24:41 Play from here That it's like, you're not crazy.

24:44 Play from here This is built into the system.

24:46 Play from here I didn't, I wasn't asking you if I was crazy.

24:48 Play from here I was looking for an answer.

24:50 Play from here And you're not wrong.

24:52 Play from here And you're thinking this one drives me nuts.

24:55 Play from here And I I've I it's literally caused me to shift a lot of my behavior away from GPT.

25:01 Play from here However, that brings up a little bit of a challenge.

25:03 Play from here I want to get back.

25:04 Play from here I do want to explore the app thing, but it, it, it, it.

25:07 Play from here A lot of people out there have this challenge in that they can't quite find

25:11 Play from here the right model that fits, fits their needs.

25:13 Play from here And the reality is you will, at this point, I have tens of thousands of chats

25:17 Play from here and that is context that is memory.

25:21 Play from here That is established work.

25:23 Play from here And for me to move an established piece of work, be it deep research, be it a

25:27 Play from here project or whatever, and start over, I will have an okay chance at Claude because

25:33 Play from here I use Claude for some very specific things and it, and it knows more about me.

25:38 Play from here But Gemini would be really, really hard for me to start from scratch.

25:43 Play from here So it's been painful to have that sort of mobility as far as moving around.

25:49 Play from here But it is funny from a commercial environment that I, I really hate GPT

25:53 Play from here right now, like I can't believe I'm saying that because it's been like,

25:56 Play from here you know, my BFF, but it just is driving me crazy and I will, I am using the

26:02 Play from here other platforms more as a result, which seems like a little bit of a

26:06 Play from here misstep from a product perspective.

26:09 Play from here What makes me happy to hear that is the, it's not just me kind of worldview

26:14 Play from here because I've been bouncing models like crazy.

26:17 Play from here I've been bouncing across models like crazy because I've been having

26:20 Play from here similar problems, similar issues.

26:22 Play from here Don't like, I don't like some of the recent changes.

26:25 Play from here I've been leaning into Claude a lot lately for a lot more things.

26:29 Play from here And Gemini as well.

26:30 Play from here Yet, you know, I've also been doing the whole cross-examination with different

26:34 Play from here prompts and these different models.

26:36 Play from here And sometimes I will get something that just feels like

26:38 Play from here markedly better from GPT still.

26:40 Play from here So we are in this weird time where you do have to experiment, right?

26:44 Play from here You have to experiment with your different use cases and workflows and the type of

26:48 Play from here responses that are most helpful for you because it's not only changing so fast,

26:52 Play from here but all of these models are feeling really different right now as well, as far as

26:56 Play from here like what they're capable of and their tone and everything else.

26:59 Play from here But the apps, the apps.

27:01 Play from here So real quick, you know, we have one developer on our team who is convinced

27:05 Play from here that this is, this is the next big thing, right?

27:08 Play from here That this is, we need to be putting our attention on this because this is like,

27:13 Play from here this is like iOS in 2008 all over again.

27:15 Play from here They've opened up the app store for developers and everybody can build this

27:19 Play from here stuff and publish it on GPT.

27:22 Play from here And then you're going to be early and you will be the, you'll, you'll be the

27:28 Play from here publisher who made the great app and you'll get the lion's share of users for

27:32 Play from here this.

27:33 Play from here So it was one of those things where the connectors, like my experience was awful.

27:37 Play from here And this is really just rebranded connectors.

27:39 Play from here I would try to hook up Canva and I'm like, cool, it does nothing.

27:41 Play from here Like it's giving me literally nothing useful out of this connector.

27:45 Play from here I can't do anything.

27:46 Play from here Not only that, it's really rough in regards to any information as to what it's

27:51 Play from here even supposed to do.

27:52 Play from here Like it's not very well documented, but the things that it's supposed to do, it

27:55 Play from here wouldn't do either.

27:56 Play from here Right.

27:56 Play from here So I'm like, okay, I'll open up the app store and I'll look around and I'm like,

27:59 Play from here Oh, Spotify.

28:00 Play from here That's a very recognizable icon.

28:02 Play from here I'll click on that.

28:03 Play from here I use Spotify every single day.

28:04 Play from here So I click on the connector and I'm like, what would be a, what would be a real

28:09 Play from here softball to throw right down over home plate here?

28:12 Play from here Make me a great playlist for working out.

28:15 Play from here It's going to be for lifting weights.

28:17 Play from here Let's go heavy into these kind of rage against the machine, nine inch nails type

28:22 Play from here stuff, but then give me some modern stuff on top of that, but in the same genre.

28:25 Play from here Great.

28:26 Play from here No problem.

28:26 Play from here And it starts ripping out all these artists and songs.

28:28 Play from here So I'm like, Oh, this looks pretty good.

28:30 Play from here Cool.

28:30 Play from here And then, so I'm like, I assume I go to Spotify and now I have the playlist.

28:34 Play from here Nope.

28:34 Play from here I'm like, okay, so where's my playlist?

28:38 Play from here My bad.

28:38 Play from here You're not wrong to question that.

28:40 Play from here You're not crazy.

28:41 Play from here It's not there.

28:43 Play from here There must've been a problem.

28:44 Play from here Look now.

28:44 Play from here I'm like, nope, still not there.

28:46 Play from here Yeah.

28:46 Play from here Well, you have a right to be mad here because it didn't work.

28:49 Play from here I'll tell you what, I'll generate a link and that'll go to your playlist.

28:51 Play from here Cool.

28:52 Play from here Where's my link?

28:53 Play from here Well, you're not wrong to question why the link's not there.

28:55 Play from here It did nothing.

28:56 Play from here This is where I ended up.

28:57 Play from here Kevin said, this is actually what it said.

29:00 Play from here Well, you have all the artists and songs.

29:02 Play from here You can just copy these now and then go into Spotify and build a playlist.

29:07 Play from here I'm like, that's kind of what I could do a year ago.

29:10 Play from here Isn't it?

29:10 Play from here What's the deal?

29:12 Play from here I don't know.

29:13 Play from here I don't know if this is really going to be 2008 all over again, because when I

29:17 Play from here downloaded the flashlight app for my.

29:19 Play from here iPhone in 2009, it worked and I really liked it and you know, now flashlights

29:24 Play from here part of the OS, but I don't know exactly where that's going.

29:27 Play from here So have you messed around with this stuff at all?

29:30 Play from here No, not really.

29:32 Play from here Most of our focus internally is workflow development.

29:35 Play from here So we're more on the API tool set implement implementations, which I believe

29:39 Play from here you can call the apps into them as well.

29:42 Play from here But I haven't done it personally.

29:43 Play from here Yeah.

29:44 Play from here Yeah.

29:44 Play from here I'm always trying to split my attention between this kind of goes back to your

29:49 Play from here point about how quickly this is moving.

29:51 Play from here And is this all built on sand?

29:53 Play from here Not really, but it's evolving very quickly.

29:56 Play from here Is that building custom, building custom apps, custom workflows, or

30:00 Play from here even no-code stuff like N8n.

30:02 Play from here That's where we are today.

30:03 Play from here But you would think that what is happening in the app store or what's

30:07 Play from here happening with other connectors and features of these models might

30:11 Play from here replace some of that, right?

30:12 Play from here So I'm always trying to split my attention between here's some really

30:15 Play from here relatively complex stuff, very complex stuff for somebody who doesn't live in

30:19 Play from here this every day versus here's what I could do if I just paid my 20 bucks a month

30:25 Play from here for one of these models and all of its features out of the box, because I would

30:29 Play from here think or hope to think that's the direction that a lot of this would be

30:34 Play from here going, that it is going to be very user-friendly, right, without having

30:38 Play from here really to have all of these development skills or a bunch of third-party tools

30:43 Play from here or both this stuff altogether.

30:44 Play from here So I'm trying to keep an eye on both of those.

30:46 Play from here So I felt compelled to try this and was very disappointed, but

30:50 Play from here here's where I wasn't disappointed.

30:52 Play from here In the mobile version of Claude, I looked at some of the tools and I saw that it

30:57 Play from here had the Reminders app in my iPhone in there as a connection, I thought, oh,

31:02 Play from here interesting, cool, so I enabled it.

31:04 Play from here Then I went to the coach I built in a project, which I've just dumped endless

31:09 Play from here amounts of documentation, including like an entire journal of 10 years.

31:13 Play from here I mean, tons and tons of context and what's amazing is I've barely scratched

31:16 Play from here the surface of the amount of memory that I can use for that project and I've

31:19 Play from here given it everything, everything, a whole kitchen sink of like everything to

31:24 Play from here reference about helping me as an executive coach, so I thought, wow, wouldn't this

31:29 Play from here be cool instead of having to go into the app and hit the plus button and do a

31:33 Play from here reminder instead of due date to just have a conversation, right, to just turn

31:37 Play from here on voice and say, hey, could you remind me to do this task I want to do on this

31:41 Play from here date, here are the notes, here are the subtasks, it did it flawlessly, like it

31:45 Play from here just put it right in the app flawlessly.

31:47 Play from here One caveat is that it does only work on mobile.

31:49 Play from here You can't do that on desktop, but you can access it on desktop later, right?

31:53 Play from here But here's where it got really cool.

31:54 Play from here So as part of that downtime over new years, making a bunch of different

31:58 Play from here changes, both business, personal, everything else, and you know, like you

32:02 Play from here said, we're just constantly running in a million different directions

32:05 Play from here trying to assess all of this.

32:06 Play from here So I just had this big stream of conscious Google doc, just vomiting out

32:11 Play from here like paragraphs and paragraphs using whisper flow, right, using the dictation.

32:15 Play from here So it's probably way more than I needed.

32:17 Play from here I would have never typed that much, but I was talking for hours about all this.

32:20 Play from here And I took that unstructured document and I connected that to Claude and I said, I

32:25 Play from here want you to take this and I want you to talk me through it and then prioritize

32:29 Play from here where all this stuff goes, turn it into tasks, put it into reminders, give it due

32:34 Play from here dates based on its priorities.

32:36 Play from here And after about an hour, um, I had like 70 tasks going into the new year, all

32:42 Play from here with these different due dates.

32:43 Play from here Amazing.

32:44 Play from here Like, like amazing.

32:46 Play from here Anybody, anybody can do this because I didn't build anything.

32:49 Play from here I didn't do anything fancy.

32:51 Play from here I clicked a connector button on my iPhone and that was it.

32:54 Play from here So the ghost of Christmas future here is in about three weeks, you're going to

32:58 Play from here recognize that all the tasks are not in your project management system.

33:02 Play from here And now you're going to be really upset because your reminder system isn't

33:05 Play from here connected to ClickUp and now you're managing multiple systems and you're going

33:09 Play from here to have to build some sort of agent that moves things from your reminders app over

33:13 Play from here into ClickUp, which I happen to know that you use.

33:16 Play from here I know.

33:17 Play from here Don't, don't think I didn't think about this.

33:19 Play from here I will for context, I really do kind of keep my personal to-dos or just general

33:25 Play from here goals different than like stuff that the whole company's looking at that I'm

33:29 Play from here publicly putting on a board and moving around.

33:31 Play from here So yes, that all happens in ClickUp, but I've always really liked the native

33:36 Play from here applications on iOS because they're just super reliable.

33:40 Play from here They're snappy.

33:41 Play from here They sync seamlessly across, you know, desktop, tablet, mobile, everything else.

33:44 Play from here I, you know, I forgot that there was even a reminders app on iOS to be honest.

33:49 Play from here Like I hadn't looked at that thing since it came out because it was crappy.

33:52 Play from here It's really good.

33:53 Play from here Like it's really good.

33:54 Play from here In fairness, this is a real challenge that everyone needs to keep in mind that our

33:59 Play from here information is centered in all kinds of different places and it's a known issue

34:03 Play from here for the LLMs and they're using things like apps and connectors to try and

34:07 Play from here pull those pieces together, but those things are all built on the way things

34:12 Play from here were before before and trying to make all of that work requires a lot of digital

34:17 Play from here duct tape and until we can get to a point where cleanly you can have a conversation

34:24 Play from here with HubSpot through your LLM and reasonably expect things to land in

34:30 Play from here ClickUp on the other side or Notion or Reminders or whatever it is and

34:34 Play from here turn things around vice versa.

34:36 Play from here This is going to be kludgy for a while, but I do have hope that we're going to get

34:40 Play from here there and part of the driver here is enhanced context.

34:44 Play from here So we haven't revisited, we haven't visited this on the show, but this context

34:49 Play from here is of course the amount of information that an LLM can hold at once and yield

34:53 Play from here reasonably accurate results and that context curve keeps increasing with it

34:59 Play from here being millions of words in Google at this point and hundreds of thousands

35:02 Play from here of words in the Kludd and in GPT.

35:05 Play from here So as that context window gets larger, it can hold more information in longer

35:12 Play from here from more sources, which will make it more valuable for most tasks that we have.

35:17 Play from here So, you know, we, here we are, we're, we're kind of complaining about this,

35:21 Play from here but we've come a long, long way.

35:23 Play from here When GPT-3.5 launched, it was only like 3,600 words.

35:27 Play from here So now we're into the hundreds of thousands or millions and we're just

35:30 Play from here impatient for us to get to that next stage because it should be working

35:35 Play from here more seamlessly and it just isn't yet.

35:38 Play from here Yeah.

35:38 Play from here Yeah.

35:38 Play from here But like you said, it's getting so much better, right?

35:40 Play from here Where in the early days, if you wanted a big context window, like if you wanted

35:46 Play from here real memory, right, you would have to jump through these development hoops

35:51 Play from here and hook up some sort of a database, right?

35:54 Play from here You've, you've got to like put in some sort of a vector database.

35:57 Play from here You've got to like wire up a Pinecone or a Superbase or something to give you

36:02 Play from here context outside of what was just in the off the shelf model and it's now starting

36:07 Play from here to approach that point where you can get really far without doing any of that.

36:13 Play from here And like you said, the distance that has come in a year is pretty staggering, right?

36:18 Play from here So to put it in perspective, they call it tokens, right?

36:21 Play from here And tokens, there is a rough equivalent to the number of pages or words, right?

36:26 Play from here So in Claude, which is my favorite girlfriend right now, that is, let's see,

36:32 Play from here I think we're up to 200,000 tokens and that's roughly 150,000 words of context.

36:39 Play from here GPT 5.2, you've got 400,000 tokens, so roughly twice as much.

36:44 Play from here Then you've got Gemini 3 at about a million tokens and a lot of talk about

36:48 Play from here 2 million being right around the corner, right, that's a lot, like that's, that's

36:54 Play from here a lot, that's like books of words.

36:56 Play from here I think this is in terms of something that's going to be really practical for

37:00 Play from here our audience is to understand that yes, you can cram a lot of information into

37:06 Play from here that context and yes, it will be pretty usable within that, that set amount of

37:12 Play from here material, but you still have to be careful about cramming too many things

37:16 Play from here in that look too similar, for example.

37:19 Play from here So if, I don't know, you're a pen manufacturer and you have, you have

37:24 Play from here 5,000 different SKUs of pens.

37:27 Play from here If you have a catalog of all of your pens and they're all black pens that

37:32 Play from here have gel ink and they're like minor differences between them, you're going

37:36 Play from here to see that the LLM will bleed those concepts together and you're going to

37:41 Play from here end up with hallucinations that are coming out the other end, so it's still

37:46 Play from here pretty important to think about more regimented approaches to this and those

37:50 Play from here are things like RAG, which, you know, we're not a technical show, but

37:55 Play from here increasingly people need to understand that retrieval augmented generation is

37:59 Play from here what RAG is and it's the ability to lock in pieces of data and when your

38:04 Play from here application demands it, you need to have a sense of that and I think that's what

38:09 Play from here I'm putting out there is to recognize that even though context is getting

38:13 Play from here really big and even though the individual platforms are using versions

38:17 Play from here of RAG to try and lock in pieces of information such that they come out

38:21 Play from here accurately, you're going to be in a place where it seems like you can just

38:25 Play from here dump in 100,000 documents or 100 documents into Gemini and have it just

38:31 Play from here work, but there's a high probability with the more complexity that you add

38:38 Play from here that the results will drift, which increases your obligation to have some

38:42 Play from here sort of like a feedback loop in the process to check things and I don't see

38:47 Play from here a lot of that in even sophisticated users that they get to a point or

38:51 Play from here semi-sophisticated users, they get to a point where they've built their custom

38:54 Play from here GPT or their project or whatever and they just sort of start blindly relying

38:59 Play from here on it and injecting some gut checks into it, some things that you need to look at

39:05 Play from here is really important and you sort of referred to this earlier with a practice

39:08 Play from here of using multiple models, so either multiple models or multiple chats to

39:12 Play from here check outputs is generally my MO, particularly if it's something that's an

39:16 Play from here important document, so I will output the document in one chat and then start

39:22 Play from here another chat to sort of gut check it and it's remarkable how much it beats

39:27 Play from here itself up sometimes, like, oh, well, that was terrible information, it's like,

39:30 Play from here well, you gave me that information, like what's wrong with you?

39:33 Play from here Or if you're really sensitive, using a different platform

39:36 Play from here model to look at it as well.

39:38 Play from here That's my, that's my favorite.

39:39 Play from here That's my favorite.

39:41 Play from here This was the answer that Gemini gave me.

39:46 Play from here Seems better.

39:47 Play from here I might do this.

39:49 Play from here I, listen, and I do, I actually do this all the time.

39:52 Play from here I'd actually done a little and, and workflow that would kind of like make

39:55 Play from here them all fight with each other because in each other's outputs and try to

39:59 Play from here see what was a better answer, it was amazing to see some of the models

40:01 Play from here concede and say, on second thought, like this is actually a much better

40:04 Play from here approach that you got from Claude or from Gemini, right, which is super

40:08 Play from here interesting, to your point about that context and things starting to get

40:11 Play from here muddled for the person who isn't going to be coding or ragging some vector

40:15 Play from here database, separating things in projects.

40:17 Play from here Even if you are in one single Frontier model is also a good

40:21 Play from here way to kind of keep that clean.

40:23 Play from here A little bit fuzzy as to how this bleeds over as things evolve, right?

40:28 Play from here Like how it's taking things from different projects or different

40:31 Play from here conversations, but for me, it's been a pretty clean approach.

40:34 Play from here I know that Claude does have some pretty definitive lines between if

40:39 Play from here I build a project in Claude, it's not going to go and then pull any reference

40:43 Play from here from a separate project in Claude, like it stays in its lane and at least

40:48 Play from here when asking the model, it says that it stays in its lane, but who knows?

40:51 Play from here Because some of the most unreliable information I've gotten about models

40:55 Play from here is when I asked the models for information about their features.

40:58 Play from here This actually, I've really only seen it happen a lot with actually both

41:01 Play from here Gemini and GPT, where I'd say, Hey, is it possible for you to do this

41:06 Play from here because I want to leverage this feature?

41:09 Play from here And it says, no, sorry, that's a limitation.

41:11 Play from here I can't do that.

41:12 Play from here I'm like, no, actually I know you can do that because I just read a press release.

41:16 Play from here Oh, I stand corrected.

41:17 Play from here You're right.

41:18 Play from here I can do that.

41:18 Play from here So it's sometimes difficult to even figure out what these models can do based on

41:26 Play from here their own intelligence about themselves.

41:28 Play from here Well, you're stumbling right into training windows as well, which is

41:32 Play from here something that we talked a lot about in 2023 and 2024 because these are the

41:37 Play from here cutoff windows before which the model was trained and here we are in a time

41:44 Play from here when all the models have browser capability, but it's really, really

41:49 Play from here important for anybody listening to understand that yes, the model can go

41:54 Play from here out and it can get a snippet of information from today's news, but it

41:58 Play from here isn't trained on that information.

42:01 Play from here So that information hasn't been incorporated into its neural net such

42:06 Play from here that the ramifications and the connections and such make sense.

42:10 Play from here Instead, it's relying on previous information, which I mean, correct me if

42:15 Play from here I'm wrong, but I believe that GPT's, GPT-5's cutoff window is somewhere in

42:20 Play from here the early 2025, so it doesn't know things.

42:25 Play from here And this is also a problem for you if you have web information that you're

42:31 Play from here trying to convey through your brand, that it may have been trained on the

42:34 Play from here different products or different things that are on your website previously, but

42:38 Play from here as those change over time, that training gets looser and looser and it doesn't,

42:44 Play from here by saying it, you know, anthropomorphizing it, whatever, the system can't really

42:47 Play from here make sense of it in a contemporaneous way, even if it feels like it can.

42:52 Play from here So always know that, that the trained information is lagging way behind the

42:59 Play from here contemporaneous information that you're pulling, so you have to be really careful

43:03 Play from here about where you're pulling things from.

43:05 Play from here And this, so we didn't, we won't get a chance to talk about skills, but

43:08 Play from here skills are something that we need to talk about going forward for sure.

43:12 Play from here As far as using skills within CLAUD and now within OpenAI, they're important

43:18 Play from here because they're locking in the, the most current approach to dealing with a

43:24 Play from here platform, not trying to address a platform challenge based on information that it

43:30 Play from here was trained on, you know, six months ago or a year ago.

43:33 Play from here And you'll run into this all the time with like, you know, your accounting

43:36 Play from here platform, you can ask GPT about your accounting platform and how to solve

43:40 Play from here problems, but it really doesn't know how the platform may have changed in the, in

43:46 Play from here the interim, unless you're using something like a skill.

43:48 Play from here Right.

43:49 Play from here Or unless you're thoughtfully prompting as well, right?

43:51 Play from here Because that's, that's another thing is that it will give you incorrect

43:54 Play from here information before it will just automatically search the web for the

43:58 Play from here most current information.

43:59 Play from here But.

43:59 Play from here If in your prompt, you say, search the web

44:02 Play from here for the most current information and then ask the question

44:05 Play from here or then look for whatever data you're trying to get,

44:08 Play from here you'll see it doing that.

44:09 Play from here It will search the web and then fetch that.

44:11 Play from here But it's very quick.

44:12 Play from here And especially with the automatic model routing,

44:14 Play from here in its own interest to save as much energy as possible,

44:18 Play from here it will just go and grab the lowest hanging fruit

44:21 Play from here from whatever it was trained on previously,

44:22 Play from here which might've been a year ago, right?

44:24 Play from here And just give you something completely inaccurate

44:26 Play from here or false sometimes as well, which is what I've seen happen.

44:29 Play from here Yeah, yeah.

44:30 Play from here Well, Kevin, we're running up on the hour here, bud.

44:32 Play from here Is there anything else that you kind of played with

44:34 Play from here over the break?

44:35 Play from here Because I felt like we didn't get to catch up at all.

44:37 Play from here And we both spent a lot of time just pounding away

44:41 Play from here on these things and having a chance to play

44:43 Play from here without as many distractions.

44:45 Play from here For me, like these just very consumer level connections

44:50 Play from here that I found in Cloud that worked amazing

44:53 Play from here with my Google Drive, that worked flawlessly

44:55 Play from here with my native iOS apps, that was a huge unlock.

44:58 Play from here I had some similar experiences with Gemini as well.

45:01 Play from here I didn't have any luck with GPT, as I mentioned.

45:04 Play from here Were there any new tools or big aha's

45:06 Play from here that you kind of came out with from the break

45:08 Play from here after having a little downtime?

45:10 Play from here We certainly have spent enough time talking about Notebook LM

45:13 Play from here in previous episodes, but now it's gone beyond toy

45:18 Play from here to being like functional utility.

45:20 Play from here That's just core to the way that I do things

45:24 Play from here beyond just the output quality and things.

45:26 Play from here And you know, it's just going to continue to get better

45:29 Play from here and more useful.

45:30 Play from here Since that conversation, I've used it a ton.

45:32 Play from here Did you see any advancements like over the last month

45:34 Play from here since we spoke about it last time?

45:37 Play from here I mean, I've actually seen some degradation of quality.

45:39 Play from here I think that they might be throttling it back.

45:41 Play from here That if you're doing big infographics or decks or whatever,

45:45 Play from here they all have a look by the way, but it's a beautiful look.

45:48 Play from here And it's just already everywhere.

45:50 Play from here But it is a tell, right?

45:51 Play from here It's like the LM version of an M-dash in GPT's writing.

45:56 Play from here Like people who know, know, like they see it

45:58 Play from here and they're like, Notebook LM, nice, cool.

46:01 Play from here So my wife uses M-dashes in her real writing,

46:03 Play from here just for the record, as a lawyer.

46:04 Play from here And it's her first name.

46:05 Play from here I know, I've heard the same thing from other lawyers.

46:07 Play from here Like you are chat GPT, like you, like, wow.

46:12 Play from here She sends them to you like in text messages.

46:14 Play from here You're like, wait, are you using GPT

46:16 Play from here to talk to me right now?

46:17 Play from here She definitely is not.

46:18 Play from here I know that she is not.

46:20 Play from here I'm very, very comfortable with that.

46:23 Play from here I would say that one, just interesting output from that

46:26 Play from here is didn't know this about myself,

46:29 Play from here but I'm apparently more visual than I thought.

46:31 Play from here That's being able to take a monster document

46:35 Play from here and being able to turn it into a visual representation,

46:38 Play from here be that a deck, be that an infographic,

46:40 Play from here for me has been a huge unlock

46:42 Play from here as I'm trying to absorb information from every corner.

46:45 Play from here And I just don't have the bandwidth to read,

46:47 Play from here particularly academic literature

46:49 Play from here and the ability to kind of create a visual

46:51 Play from here to kind of create a cheat sheet,

46:53 Play from here like the cartoon version of the MIT iceberg study,

46:56 Play from here like, which is totally another topic

46:58 Play from here that we should talk about at some point,

47:00 Play from here but that's like, that's been an unlock.

47:03 Play from here Other things I would say,

47:06 Play from here more granular work within workflow builders.

47:11 Play from here I recognize that they're a bandaid by the way,

47:13 Play from here that we will get beyond these workflow builders

47:16 Play from here at some point, but approaching testing

47:19 Play from here from a workflow perspective

47:22 Play from here versus an all-in-one perspective

47:24 Play from here has been a bit of an unlock for me

47:27 Play from here that versus trying to vibe code a whole solution,

47:29 Play from here a workflow allows you to create

47:31 Play from here a complicated step of a set of events,

47:35 Play from here and you can isolate the individual thing

47:37 Play from here that you want to fix and get that thing right.

47:41 Play from here And you might end up with a heck of a lot more steps

47:43 Play from here than are necessarily needed,

47:45 Play from here but it's going to work as it goes through the process

47:48 Play from here in a way that is much less frustrating

47:51 Play from here than trying to code things from scratch.

47:53 Play from here So-

47:54 Play from here And you're speaking to N8n for the most part still, right?

47:56 Play from here N8n, yep, is still my go-to for sure.

48:00 Play from here Although I intend to spend more time in Opal,

48:02 Play from here which is Google's solution since we're a Google shop,

48:06 Play from here and the idea of using vibe coding in the right places,

48:09 Play from here like vibe coding for a UX that's nice and pretty

48:12 Play from here and has a bunch of fields,

48:13 Play from here and then tying that into a workflow

48:15 Play from here as far as data connectors and data storage

48:18 Play from here and conversions and things like that

48:19 Play from here is just a much cleaner way to go about it

48:21 Play from here for most non-developer type people.

48:24 Play from here Right, right.

48:25 Play from here Yeah, I've also been spending a lot of time in Notebook LM

48:28 Play from here and just a lot more time,

48:29 Play from here we were talking about Cloud and just where it really shines,

48:33 Play from here and especially for this kind of more personal coaching

48:36 Play from here and just language that is less sycophantic

48:38 Play from here and more directive.

48:39 Play from here And I feel like it is very quick to just cut me off

48:43 Play from here and say, okay, cool, you have your answer,

48:44 Play from here now go do your work,

48:45 Play from here like it doesn't want to just keep talking to me forever.

48:48 Play from here I'm like, wow, very notable difference in the model.

48:49 Play from here One thing that I realized

48:51 Play from here because I am newer to using that

48:53 Play from here as heavy as I was with Gemini and GPT

48:57 Play from here is that when you build one of these projects,

48:59 Play from here you're kind of stuck with whatever model

49:01 Play from here you decided to use for that project,

49:03 Play from here which is a little bit of a bummer.

49:05 Play from here And I don't know how many projects

49:06 Play from here you've been running inside Cloud,

49:08 Play from here but I would often switch models

49:10 Play from here inside a custom GPT, for example,

49:13 Play from here or in a Google gem, for example,

49:16 Play from here where I'd say, okay,

49:17 Play from here these are a bunch of like quick, fast, easy answers,

49:20 Play from here and I'm just going to keep banging away on this,

49:22 Play from here and I want it right away.

49:23 Play from here And then, ooh,

49:24 Play from here this is going to require some more thinking, right?

49:26 Play from here Or this is going to be a deep research project.

49:28 Play from here Can't do that,

49:29 Play from here like can't do that inside of a Cloud project,

49:33 Play from here which is a little bit of a bummer,

49:36 Play from here because like you said,

49:37 Play from here you really do start to feel invested

49:39 Play from here in these conversations,

49:40 Play from here like in these different threads.

49:41 Play from here So to interpret just a little bit

49:43 Play from here that if you're in GPT,

49:45 Play from here which I literally have that GPT window up right now,

49:47 Play from here up in the upper left-hand corner,

49:49 Play from here you can constantly play around

49:51 Play from here with whether you want the auto answer,

49:53 Play from here whether you want a quick answer,

49:54 Play from here whether you want a thinking answer,

49:55 Play from here or the pro answer,

49:57 Play from here which will burn up a lot of your capacity,

49:59 Play from here but will take a lot more time.

50:01 Play from here And internally,

50:02 Play from here it's doing recursive loops,

50:05 Play from here such that it's looking over the material.

50:07 Play from here And that is very,

50:09 Play from here that's a useful skill to develop,

50:11 Play from here is that do I want to be in the slow lane?

50:13 Play from here Do I want to be in the fast lane?

50:14 Play from here What do I need to consider?

50:15 Play from here How long is this going to take?

50:17 Play from here And as I was saying earlier with your stew pot,

50:19 Play from here like a pro deep research run

50:22 Play from here can take 35, 45 minutes

50:24 Play from here for it to go off and do its thing.

50:26 Play from here And you better really want that.

50:28 Play from here But I spend a lot of time in thinking,

50:30 Play from here which takes more time.

50:32 Play from here But if I need to like reorder something or whatever,

50:35 Play from here you can switch into that fast lane

50:37 Play from here and make things happen.

50:38 Play from here So yeah, there's a limitation of Cloud there.

50:40 Play from here I always use Opus 4 or 5,

50:42 Play from here which is the latest and greatest

50:44 Play from here and best thinking model,

50:45 Play from here which takes a while,

50:46 Play from here but it will burn up your capacity way too quickly.

50:50 Play from here Yeah, you mentioned that.

50:51 Play from here Yeah, that you kept hitting walls, right?

50:53 Play from here I kept hitting walls.

50:54 Play from here And it's so frustrating because I probably,

50:57 Play from here hey, I'm looking at you, Dario Amadei,

50:59 Play from here that like from a product adoption perspective,

51:02 Play from here this is killer.

51:03 Play from here And Dario's telling you to use Sonic.

51:05 Play from here What are you doing, man?

51:06 Play from here I can't, I can't.

51:07 Play from here He's doing fine.

51:08 Play from here But I can't rely on the product

51:11 Play from here from a workflow perspective.

51:13 Play from here I can only use it for special circumstances.

51:16 Play from here I can't spend more than 30 minutes with it

51:18 Play from here without blowing out my limits.

51:20 Play from here And then, you know,

51:22 Play from here I don't pay for the $200 a month plan.

51:24 Play from here I think you do.

51:25 Play from here So, but I think I would still get tapped out on it.

51:28 Play from here Whereas GPT and Gemini-

51:29 Play from here You will still get tapped out.

51:31 Play from here Yeah, yeah.

51:32 Play from here It's funny, because I was going to ask you

51:33 Play from here for your projects,

51:34 Play from here which model are you typically running?

51:36 Play from here And you're running the big boy.

51:38 Play from here I basically, there's no point for me to go to Cloud

51:42 Play from here if I'm not going to run Opus 4.5.

51:44 Play from here Like, I can get fine results out of GPT

51:47 Play from here when I'm tolerating it,

51:49 Play from here or Gemini using a lower grade model.

51:53 Play from here You know, hey, here we are,

51:54 Play from here we're using all of these things,

51:56 Play from here and it does ebb and flow.

51:58 Play from here You know, I was asked by a client just yesterday,

52:01 Play from here like, you know, hey, we've got a lot of people.

52:03 Play from here Do we really need to be running GPT versus Gemini?

52:07 Play from here I said, you know, hey,

52:09 Play from here you're trying to get a stable platform in place

52:13 Play from here that's going to work for the largest number of people.

52:16 Play from here Gemini's probably a fine bet at this point,

52:19 Play from here but you, as the CEO,

52:21 Play from here you may want to run multiple models

52:23 Play from here to solve different problems.

52:25 Play from here And if you have power users

52:26 Play from here who need to solve similar problems,

52:29 Play from here you might expect to need that person

52:31 Play from here to have a Cloud account as well as a Gemini account.

52:34 Play from here But for most people, for most applications,

52:37 Play from here Gemini's more than fine.

52:38 Play from here It's great.

52:39 Play from here Yeah, yeah, yeah.

52:40 Play from here And we're talking, you know, leadership roles,

52:42 Play from here real strategic thinking,

52:43 Play from here where you're not just,

52:44 Play from here hey, summarize these meeting notes for me.

52:47 Play from here Make me a deck, but really like,

52:49 Play from here hey, I'm going to dump in all of my financials

52:52 Play from here and a competitor analysis,

52:54 Play from here and I want to then have you ask me,

52:58 Play from here you know, a hundred different questions about my business

53:00 Play from here and what some different options for my plans

53:02 Play from here for Q2 should be.

53:04 Play from here That, when you are using these different models

53:06 Play from here and you're using the real thinking versions of them,

53:10 Play from here you'll get different outputs.

53:12 Play from here And we're living in these strange times

53:14 Play from here where it is well worth it to do that, right?

53:18 Play from here I mean, you can actually get months of work done

53:21 Play from here in an afternoon by trying these different models out,

53:24 Play from here seeing what kind of responses you get,

53:25 Play from here finding best of class, what works for you.

53:28 Play from here Yeah, these are wild times, Kevin.

53:30 Play from here Yeah, it is.

53:31 Play from here It's funny.

53:32 Play from here We've probably closed with this idea

53:33 Play from here that it is startling when you get, quote,

53:36 Play from here months or hours of work in a few minutes.

53:38 Play from here And when it happens overly fast,

53:41 Play from here I find it odd that it's like, that was too fast.

53:45 Play from here Like my brain barely got away from what I'm after.

53:49 Play from here And I frankly need a little bit more time to settle

53:52 Play from here with how I'm thinking about this

53:54 Play from here so I can get to the next step.

53:55 Play from here So that pace, that pace for the paralegal,

53:58 Play from here that pace for the entrepreneur,

54:01 Play from here like that pace is gonna just go crazy going forward

54:04 Play from here and balancing that with the ability of the system

54:08 Play from here to absorb the outputs too.

54:10 Play from here So I'm probably a nightmare in that I'm putting out

54:14 Play from here hundreds of thousands of words a week

54:16 Play from here in like briefs and projects and all of these other things.

54:20 Play from here And I have a AI forward team, of course,

54:23 Play from here so they're using tools to synthesize

54:25 Play from here through all of your outputs.

54:27 Play from here I get to go through my outputs.

54:28 Play from here Yeah, that's Kevin's-

54:29 Play from here Summarize the last hundred pages

54:31 Play from here that Kevin just sent me in Slack.

54:33 Play from here But think about that from, if you're not working

54:36 Play from here for an organization where you know

54:37 Play from here that that's what you signed up for,

54:39 Play from here like that's gonna be really hard for people to deal with

54:42 Play from here in terms of pace.

54:43 Play from here And is it what's the market is really going to demand?

54:47 Play from here Depends on what you do.

54:48 Play from here Depends on like how critical this technology is.

54:51 Play from here That's, yeah, if you're, yeah, so much.

54:55 Play from here I think that really threads back into

54:57 Play from here what you were saying about,

55:00 Play from here are we entering this world where all of this

55:01 Play from here just BS minutia is getting taken off our plates,

55:04 Play from here but are we then just gonna have to be working

55:06 Play from here 10 times harder on the really hard problems

55:09 Play from here all the time without any breaks

55:11 Play from here of dealing with the minutia, right?

55:12 Play from here And that is a great example of it

55:15 Play from here because I see it too.

55:16 Play from here I see like these incredibly verbose outputs and documents

55:20 Play from here and things to review from my team where I'm like,

55:23 Play from here this definitely came from one of the models

55:26 Play from here and now I've got to consume it.

55:28 Play from here I've got to put it into one of my models

55:30 Play from here to just at least get the bird's eye view of this thing

55:32 Play from here because I've got 20 other ones to look at, right?

55:35 Play from here So yeah, our work isn't exactly going away.

55:38 Play from here It's just changing and it's changing really fast.

55:40 Play from here And with that.

55:41 Play from here With that, that is it.

55:43 Play from here That is it.

55:43 Play from here A little optimism, a little pessimism.

55:45 Play from here That's a nice even blend to start the new year.

55:47 Play from here The time before, just think back in history.

55:49 Play from here What did the time before look like

55:51 Play from here and where are the opportunities in the time at?

55:53 Play from here Yeah, I'm looking forward to seeing who was right,

55:57 Play from here who was wrong and who had no idea what was coming.

56:00 Play from here All right, Kevin, until next week, man.

56:02 Play from here Have a good one.

56:03 Play from here We'll see you.

56:04 Play from here That is the pod.

56:05 Play from here See you.

56:06 Play from here Thanks for listening.

56:11 Play from here If you'd like to send us a question to answer on the show,

56:13 Play from here email me at Elijah at spark6.com.

56:16 Play from here That is S-P-A-R-K, the number six.com

56:19 Play from here or Kevin at Kevin at ascendlabs.ai.

56:23 Play from here To learn more about AI strategy

56:25 Play from here and training for your business,

56:27 Play from here visit Kevin's company, Ascend AI Labs at ascendlabs.ai.

56:32 Play from here And to learn more about design and development

56:34 Play from here of custom AI solutions, visit my agency at spark6.com.

56:40 Play from here You can also sign up for my free newsletter,

56:42 Play from here AI Framework Friday, by visiting spark6.com forward slash.

56:47 Play from here Newsletter.

56:47 Play from here Thank you.

56:48 Play from here Thank you.

56:49 Play from here Thank you.

56:50 Play from here Thank you.

56:51 Play from here Thank you.

56:52 Play from here Thank you.

56:52 Play from here Thank you.

56:53 Play from here Thank you.

56:54 Play from here Thank you.

56:55 Play from here Thank you.

56:56 Play from here Thank you.

56:57 Play from here Thank you.

How to cite this episode

Credit the named speaker and AI at Work by AscendAI. Link to the episode or the passage you quoted.

“Ep #9 Why Some Early Adopters of AI Are So Stressed (The Busywork Paradox).” AI at Work by AscendAI, 2026-01-06. https://ascendlabs.ai/podcast/why-some-early-adopters-of-ai-are-so-stressed-the-busywork-paradox/

Recent episodes

View all episodes
×
AI Solutions Graphic
Free Resource
Get Started Now