36. What AI Is Teaching Us About Ourselves

Summary

Podcast Summary: "What AI Is Teaching Us About Ourselves"

At Work with The Ready – Episode 36
Hosts: Rodney Evans & Sam Spurlin
Air Date: November 3, 2025

---

Main Theme

This episode delves into how the rapid evolution of artificial intelligence is not only transforming organizational work but also reflecting and amplifying the existing strengths and dysfunctions within those organizations. Hosts Rodney Evans and Sam Spurlin explore the realities—hype, hope, pitfalls, and innovations—of AI adoption at work, challenging listeners to move beyond cost-cutting and into thoughtful experimentation and redesign.

---

Key Discussion Points & Insights

1. The Hype vs. Reality of AI

Is AI a Lasting Transformation or Just a Fad?

• Sam notes that while some believe AI is overhyped and transient, it's crucial to separate the froth of the AI investment “economy” from real, enduring tool development ([05:26–06:20]).
• Rodney likens the moment to the dot-com era: there may be a bubble, but the core technology—the internet then, AI now—will endure and become fundamental ([06:37–07:48]).
  • “I think it is undeniably a phase shift. I don’t see any universe in which it’s just a dot dot. It’s not going to be like DAOs. The capability that exists today, even in how nascent we are, is already pretty astonishing.” (Rodney, 06:37)

2. Mirror & Amplifier: AI Exposes Organizational DNA

• AI tends to reflect—and often worsen—whatever issues, processes, or strengths already exist within a company ([03:17–04:39]).
  • “AI is currently, for most organizations, a mirror for what is already true...probably making it more so that way.” (Rodney, 03:34)
• The common quick-win approach is just swapping out human labor with AI, without redesigning workflows or fixing underlying issues—“a paint job of AI” ([12:37–13:39]).

3. Polarities in AI Posture

Wait-and-See vs. Hype and Hope

• There’s a split between organizations ignoring AI due to skepticism or fear, and those jumping in out of FOMO, without a real plan or clarity ([09:47–10:26]).
  • “Neither of those is a great posture to take right now because neither of them provides any kind of clarity or momentum within the organization.” (Rodney, 10:15)
• Instead, the hosts advocate for deliberate, small-scale experimentation and learning.

4. From Prompt Engineering to Workflow Transformation

• Hosts reflect on the rapid obsolescence of best practices—like prompt engineering—due to relentless AI tool evolution ([11:46–12:32]).
• The “copy and paste” mentality—just making old processes faster with AI—doesn’t solve underlying work design problems ([13:39–13:55]).

5. Garbage In, Garbage Out – The Data Hygiene/Workflow Challenge

• Rodney raises the dilemma: Should organizations wait until their data is clean before applying AI, or hope that future AI will circumvent data problems? They lean toward acting now, cleaning data/structures as they go ([14:18–16:16]).
  • “I am skeptical of any course of action that is purely wait at this point. Because I don’t think you learn a lot by waiting.” (Sam, 16:00)

6. The Ongoing Role of Humans

• For now, humans are needed for context, judgment, and “sanity checking” AI outputs ([16:16–17:35]).
• Crucially, humans must clearly define which problems are worth solving with AI before deploying it ([17:35–18:05]).

7. Scaffolding for Successful AI Integration

• The Ready’s approach: build “scaffolding” including skill-building, clear constraints, documented principles, and shared spaces for learning—all developed iteratively by experimenting ([21:00–23:01]).
• Pre-existing cultural elements—like rhythms for sharing and experimentation—set the stage for smoother AI adoption ([23:01–24:11]).

8. Sensible Policy & Experimentation

• Avoid overbearing policies or “wild west” anything goes approaches; strive for “minimum viable constraint” that enables broad but safe experimentation ([24:24–27:26]).
  • “You got to get to minimum viable constraint. …You got to do the work of what minimum viable constraint is within which people can go buck wild.” (Rodney, 26:50)

9. Experimentation Trumps Planning

• There are no playbooks. Over-planning is futile in this fast-evolving space; learning by doing, sharing stories, and iterating is the path forward ([28:02–32:27]).
  • “Anybody trying to sell you a kind of end-to-end answer in the AI realm…and how it’s applied to your organization is selling you a story and … FOMO.” (Sam, 28:40)
  • “Who cares what the roadmap is? …What if you did the opposite of that?” (Rodney, 32:05)

10. Better Use Cases than Cost Takeout

• Many orgs see “first idea: best idea” as reducing headcount by replacing people with AI, but this is a missed opportunity ([33:42–34:52]).
  • “What if we focused so much more on the numerator [value] and less about making that denominator [cost] as small as possible?” (Sam, 34:39)
  • “Anybody can do a cost takeout…If you’re trying to shift to a new way of being…you don’t start with that.” (Rodney, 35:34)

11. AI for Work Redesign, Not More of the Worst Work

• Beware simply accelerating bad workplace practices (e.g., email, PowerPoint) [38:27–39:19].
  • “We don’t need to have agents making tons and tons more and faster of the worst aspects of work.” (Rodney, 39:04)
  • “Let’s not make more of the worst stuff.” (Rodney, 39:15)

12. Hopeful Vision: AI as Enabler of Self-Management

• AI could finally make organizational self-management and distributed creative power workable at scale by acting as a personalized connector, information filter, and problem-solver ([40:04–44:28]).
  • “I think there’s a future…where AI could actually bring self-management…to more organizations…pushing creative power to the edges.” (Sam, 41:11)
  • “A knowledge body that we can be in discussion with…has been a missing piece…like the fuel in self management. And I too feel like quite bullish on that.” (Rodney, 44:03)

13. Vital Practice: Breaking AI to Understand Its Limits

• Don't just experiment to see what works; actively try to “break” AI and discover its limits and failure modes. This builds realistic trust and avoids blind spots ([44:33–45:18]).
  • “We need…to really understand what it's not good at…The only way to do that is to try to break it…and noting when it’s not [giving the right answer].” (Sam, 45:09)

14. Personal Experimentation and Growth

• Don’t just use AI for “productivity.” Reflect on how it makes you better—more creative, original, or insightful ([45:18–47:01]).
  • “At as an individual…what I don’t think most people have really honed a skill around yet is like, where is this making me better and where is it not?” (Rodney, 45:18)
  • “The worst use cases are outsourcing your thinking and just getting dumber. And the best use cases are, like, how to hone and sharpen and use it to challenge you.” (Rodney, 46:35)

---

Notable Quotes & Memorable Moments

• On AI as a Mirror:
  “AI is currently, for most organizations, a mirror for what is already true…probably making it more so that way.”
  — Rodney, [03:34]

• On Hype vs. Real Value:
  “I think it is undeniably a phase shift. I don’t see any universe in which it’s just a dot dot. It’s not going to be like DAOs…even the current use cases are astonishing.”
  — Rodney, [06:37]

• On Experimentation vs. Planning:
  “Anybody trying to sell you a kind of end-to-end answer in the AI realm…and how it’s applied to your organization is selling you a story and…FOMO.”
  — Sam, [28:40]

• On AI for Cost Takeout:
  “First idea, best idea for most organizations with AI is like, oh, cool, we don’t have to pay people anymore…Such a impoverished way to view what is possible here.”
  — Sam, [33:42]

• On AI and Work Redesign:
  “We don’t need to have agents making tons and tons more and faster of the worst aspects of work.”
  — Rodney, [39:04]

• On the Absence of Playbooks:
  “There this time, like for real, for real, there are no playbooks.”
  — Sam, [28:40]

---

Timestamps for Important Segments

• Hype vs. Lasting Value: [05:26–07:48]
• AI as an Amplifier/Mirror: [03:17–04:39]
• The Copy-and-Paste Problem: [12:37–13:39]
• Data Hygiene Dilemma: [14:18–16:16]
• Importance of Defining Problems: [16:16–18:05]
• Building AI Scaffolding: [21:00–24:11]
• Minimum Viable Constraint: [24:24–27:26]
• Why There Are No Playbooks: [28:02–32:27]
• Cost Takeout Trap: [33:42–34:52]
• Self-Management Potential: [40:04–44:28]
• Testing AI Limits: [44:33–45:18]
• Individual Growth: [45:18–47:01]

---

Takeaways for Listeners

• Don’t overfocus on short-term cost savings—rethink what work can be, don’t just automate the status quo.
• There are no AI playbooks. Experiment, share, and stay flexible.
• Establish clear, simple boundaries that let people explore AI safely.
• Human context, critical thinking, and judgment are irreplaceable—especially as AI’s outputs become less transparent.
• AI could unlock creative, decentralized, adaptive ways of working—if used intentionally.
• Test AI’s limitations, don’t blindly trust it; understand where it fails.
• Strive for personal augmentation and growth, not mere productivity gains.

---

This episode is an invitation to embrace disciplined experimentation, radical clarity, and a focus on transformation over mere automation—a mindset that will define both individual and organizational success in the age of AI.

Transcript

[A]

You know, that idea of first idea, best idea, which is like an improv, like, just do the, like, the first thing that comes to mind. I think in a lot of organizations, they're doing that with AI in the sense of cost takeout, like, first idea best idea for most organizations with AI is like, oh, cool, we don't have to pay people anymore.

[B]

Hey, everybody. Welcome back to Outwork with the Ready. I'm Rodney Evans, and the man in black is Sam Sperlin.

[A]

That's what they call me every day.

[B]

I think they usually call Johnny Cash that. But for today, it's you. Yeah, that's right, y'.

[A]

All.

[B]

AI is rewriting work right now, right in this moment. We feel like the future of work has arrived. And so we're not talking about it like it's not here yet. And now the question we're asking on this show is not if you're going to evolve or adapt, but how you are going to design work for what's next and what's frankly, imminent.

[A]

Yeah, that's right. Work design is no longer optional. And the teams that treat it like a side project are going to and are being left behind. And the ones that treat it as essential will keep up with the pace of change.

[B]

So that's what we're going to talk about today. All our clients want to talk about is AI. Every sales call is about AI. Every article I read is about AI. It's all we talk about. And we haven't actually done an app on AI in. In a little minute here. But before we do, let us check in.

[A]

Rodney. I'm gonna keep it real simple today. Check in. Question. What's good?

[B]

Oh, what's good? Well, tomorrow I am taking a long weekend. I'm going to LA with my husband and some good friends. We are going to eat a lot of delicious food, and we are going to hang out with our producer Jack on Saturday for some of our food tour, which I think that's all pretty good.

[A]

Sounds amazing. And Jack, I feel like, is the right person to kind of meet up with for a food tour. He's gonna fit right in the Pulse.

[B]

He's gonna fit right in. It's gonna be great.

[A]

Definitely does.

[B]

He definitely does. Absolutely. What about you, Sam? What's good?

[A]

Well, Rodney, on this day of Thursday, October 9th of our recording, the Detroit Red Wings play their first game of the 2025, 2026 season. So that is good. I'm gonna buy myself a Little Caesar's pizza. The worst pizza. And I'm gonna bring it Home. I gotta drive about 25 minutes to go get it.

[B]

Wow.

[A]

And I'm gonna eat it because the Little Caesars is are owned by the Ilitch family. The Ilitch family owns the Red Wings. The ritual is that I eat Little Caesar's pizza on the first game of the season, regardless of how far away it was. And yes, it was easier when I lived right next door to a Little caesars and not 25 minutes away. But the routine must continue, and that's what's good.

[B]

This is what I love about sports fans. It's obviously not sports, but what I love is how much, like, mythology and pageantry and like, background story there is. Like, it's just a pizza, man. But no, it's not just a pizza. It's a tradition rooted in. Yeah, it's very Game of Thrones. I love it.

[A]

Yep, that's right. Maybe I'll be wearing a robe. We'll see. Wow.

[B]

It's a visual. That is.

[A]

All right, let's do this thing, Rodney. Let's do it. I feel like this has been a long time coming.

[B]

Yeah, it has. And we don't have exactly a repeating pattern yet because most of what we're seeing is just very chaotic. But. But here's what I will say to put a fine point on what we're going to talk about today. A lot of organizations that we work with come to us because they have a difficult time making meaningful progress. And in the case of startups and scale ups, that' often because of the chaotic nature of scaling and how hard it is to prioritize when opportunity and pressure are abundant. And at the other end of the spectrum, with more legacy and bureaucratic firms, you know, they tend to have a lot more sand in the gears and less ability to adapt on a dime. And in both cases, they call us because they're like, how do we get after it? And in today's case, it. It is AI. The problem that we're seeing is, regardless of where you are on that spectrum, AI is currently, for most organizations, a mirror for what is already true. So however your organization is, AI is A, showing you exactly and B, probably making it more so that way. And so we're going to talk a little bit today about how that's happening and what's going on, and then, as always, what to do about it instead.

[A]

Cool. And I wonder if we can actually start with a question that I'm sitting with, because I feel like every time I post or say something publicly about AI, there are always a handful of folks in the comments who are saying something like, AI is all hype. We shouldn't be thinking about it this much. It's destroying everything. So my question for us just to briefly hit up front is what are the chances this all blows over as some sort of, like, mass delusion phenomena? And we look back in like, two years, like, oh, remember when AI was a thing? Good thing we didn't, you know, go all in on that. Like, let's just take 30 seconds here. Do you have a take on that?

[B]

I want to hear your take first. This came up in AI Coffee Club today, so I have a very recent take.

[A]

Okay, cool. So my take on this is that I think it is important to disambiguate the AI economy and all of the companies that are doing increasingly strange and circular investment in each other around GPUs and data centers and other deals like that, from the tools that are actually being created themselves. And I'm not the first one to make this observation, but I think there is something potentially to be pulled from kind of the dot com bust in the early 2000s, where it was absolutely a bubble and it popped and a bunch of companies went out business and still the Internet was a thing and it was still really valuable. And if in 1999 you were like, I don't think the Internet's really going to shake out because of incredible froth that we are seeing, you would not have probably been happy with that decision.

[B]

Yeah, you were wrong.

[A]

I think we'll see a similar thing here. I think we will see a pop. I think we will see a lot of companies go out of business. I think it'll hurt the economy for a while. And I still think there are useful tools that are happening now and will come out of this in the near future that we should be paying attention to.

[B]

Mm. I think it is undeniably a phase shift. I don't see any universe in which it's just a dot dot. I just. It's not going to be like daos. Like, it's just not. And here's why I believe that one is because the capability that exists today, even in how nascent we are into this, is already pretty astonishing. And so the use cases where AI, you know, obviously we all understand that AI does better with, like, data and with, like, strong inputs and analysis than it does in a lot of other places. But even the current use cases in terms of diagnostics, in terms of, like, legal review, in terms of analysis, in terms of like, are astonishing. So I'm just like, I'm not really willing to entertain the, like, this isn't going to be a thing. That's 1, 2. I totally agree with you about like the bubble. And I basically would try to in my mind ignore any news that is not about the technology itself because all of that is just like a sports book right now that it's just like people are betting on crazy shit that is based on nothing. The investment ecosystem is largely irrational. And I think you're right. Like, I think it is the mid-90s right now and they're like pets dot com, am I right? And it's like, probably not, but like, we'll get there, you know. So I try to kind of ignore that and pay more attention to like use cases where people are showing something and you're like, holy shit, it can do what now? Like, I'm much more interested in that. Three is, and this is like my own, you know, very personal take, but I think that AI is a perfect fit for extractive capitalism and the most broken aspects of work. So it kind of feels like a perfect storm to me where 90 plus percent of companies and certainly 90 plus percent of customers are not super happy with like how customer service is going to in like call centers, et cetera, like in high volume places. And most businesses have made that as cheap as humanly possible at the expense of the customer experience. And really great agents will take care of that. Like even the AI already inside of Amazon's native app to deal with most customer service inquiries is better than 75% of the experiences I have talking to a human being. So like it's just this strata of work that's a perfect storm of shit we don't want to pay for, shit that doesn't work now. And a technology that is already proving that it will be able to do this, that I'm kind of like, this doesn't really feel arguable to me. So I try to look at stuff like that where I'm like, where's the convergence of like cost problem and solution? And those are the spots that I think make it kind of an undeniable reality.

[A]

Yeah, I'm right there with you. And I think the other thing, even if you're sitting here and still kind of being a grump about it, when there is a force like this that is radiating throughout all organizations, all of your competitors, like, it behooves you to do your own research and to do your own really investigatory work to figure out if that is actually the stance that you want. So even if you think it's all. I think there's still work to be done to like, determine that for yourself, if that's where you want to land.

[B]

Yes, hard plus one, Sam. And I would say even if you think it's all bullshit and even if you are morally opposed, still go do your own research. Like, I think that the people who are burying their heads in the sand right now for valid reasons, right? Like creative ownership, environmental burden, potential massive unemployment, like, these are all very valid concerns. And the answer to none of them is try to ignore it. Like, it's just, that's not going to work. And so, you know what I'm seeing a lot is the two polarities. It's like the wait and see crowd or the like, hype and hope crowd. I see a lot of CEOs in that camp where they're just like, I'm reading stuff, I'm psyched about it. I don't know how we're going to do it, but man, I hope it works. I don't think either of those is a great posture to take right now because neither of them provides any kind of clarity or momentum within the organization. And that to me, all we're going for right now, like, nobody has this figured out. Nobody is dialed, Nobody is getting even. The people who are publishing case studies, I think half of them are full of shit. So like, don't let FOMO or Feeling behind be the driver here for either wait and see or hype and hope. Get yourself into a place of experimentation and like, what can we try now that we can be learning from meaningfully?

[A]

Yeah. I had my first moment of realizing how quickly things were moving because I was thinking back to like a year ago and a lot of the AI discourse was like, here's how to Prompt Engineer. Prompt Engineer is going to be this like high paid thing and like, here's like a course you can take on it and all the like, tips and tricks to Prompt Engineer. And I just saw pretty respected AI thinker give a presentation a couple of weeks ago and he was like, yeah, basically all of the tools, none of those specific, like prompt engineering things are relevant anymore. Like, they are. They have moved beyond that. And we're talking like less than a year of development. We have completely shifted the conversation on this thing. That a lot of people were pretty stoked was like, the next, like, how do a cowboy AI person.

[B]

Yeah, 100%. What else are you seeing out there, Sam?

[A]

Maybe a good place to start is kind of digging into how you initially teed this up around how AI is kind of a mirror and an amplifier. Because I think the thing that I'm seeing most common or most frequently in a lot of organizations is really kind of just this copy and paste mentality about let's take our organization as it exists and give it a paint job of AI and it'll be like faster and things will work a little bit differently and we'll probably strip out some cost and like we're transforming our organization. And that is a solidifying and a magnifying of a broken OS just with like a broken os, but faster. Which is not what I want to see happen in this phase change that we are talking about. But I think it is a perfectly valid in the sense that not that you should do it, but that you can do it. And I think a lot of organizations are doing it and it's leaving a lot to be desired.

[B]

And really tactically, what does that look like in an organization is that like everybody got a little training and like Everybody got a ChatGPT license or like what is the sort of like broad brush play look like?

[A]

Yeah, the broad brush play that I'm thinking about here is basically how many humans can we replace with an agent of some sort and not really rethink like the workflows or anything, but like how many human behaviors are actually doable by AI and like, let's cut out the human and put out an AI in there instead and leave everything else around it as is.

[B]

Yeah, and this is like, this is actually a really big question that I have about AI because when I wrote like the Ready's AI sort of perspective to get feedback from you guys a few months ago, I was very much in the head of like garbage in, garbage out. And everything that I was reading and hearing from people, particularly about workflow was like, you know, you've got to get your house in order data wise in order to really be using AI. Well, in terms of workflow automation, which made a lot of sense to me, like, you know, if you're feeding it nonsense, you're going to get nonsense back. There is a different camp that says that like the AI will ultimately skip the step of data hygiene and just be able to work around that rather than just automating the garbage that you have. I do have an exit existential question right now, or maybe it's just a design question of like, how long is that going to take? And do you do the data play in order to use AI really effectively today, or do you basically wait for it to be smart enough that you don't have to get your own house in order? I don't know the answer to that. The perspective I'm taking is the former, which is have it solve material problems for us today, even if that requires more manual intervention and more work on our part, because that is part of learning. Even if ultimately that's not a necessary step in a year. But I do think that there are people out there who are like, nah, we're just gonna like let our horrible data debt exist and hope that there is an automation solve for this that does not require us to get our shit together.

[A]

Yeah, no, that's a really good question. I think I fall similar to you in the camp of I am skeptical of any course of action that is purely weight at this point. Because I don't think you learn a lot about waiting.

[B]

That's right. I think that's right.

[A]

You said that. I, and I agree. I think maybe taking action while also holding lightly, like what we're doing right now is not where we're going to be in the future, but that's okay. And like we're, we're going to keep moving there. I think the other thing this makes me think about is why it will be important, I think for a while to have humans in lots of these loops. What humans are going to still be able to do for a while in most organizations is bring the context to the output that is coming from the AI. And I'm sure over time that'll get better and better and better. And maybe we can make the argument that you don't actually need the human in the loop at that point. But I do think there's a lot of context that someone who has been doing the work for a long time or in the organization for a while can sanity check. And even if the AI is, is generating things that are unexpected and don't make sense, at least you have someone now who can like follow that thread. Basically. Is there, is there something to like understand about what is happening here? And even as I'm saying it, which is what like blows my mind about this, I understand that the way AI works is like nobody really truly can understand it. Like you can't really under. You can't really follow the path back because it's not deterministic that way, the way computers have been our entire lives, which is a bit of a mind fuck. But you get what I'm saying?

[B]

I do. And I think to that point, and maybe this is a mental model that we can share with the people listening to this, it's like you're talking about a little bit further into the work. So if you think about as a human in a system, really, whether you're a leader or you're a manager or you're an ic, I don't think it really matters. One of the human jobs right now is getting so, so clear on what problem you are trying to solve.

[A]

Yeah, yeah.

[B]

AI cannot tell you what problem in your organization you should point it to, to solve. I mean, you can ask it and it'll give you answers. But I think that from a change management lens, the human job right now is to really prioritize, like what is solvable, what would show value, what could we learn from, et cetera, et cetera, and getting like crystal clear. And I'll give really tangible examples. First of all, I am a big fan generally of like fix things inside of your own company before you try to tell your customers or clients how to fix shit. That's just like, I don't run a technology company, but that's my general take. Two is like even at the ready right now there are three use cases floating around that are being debated in terms of what order. And like when I say debated, like there will be three sprints and each of these will get made. But still it's a prioritization question. One is research for later stage sales calls. So all of us do research about the people that we're talking to, the organizations, the landscape, the strategic problems, blah blah, blah, blah, blah, as well as looking internally at what conversations we've had before, all of that kind of stuff. So it takes a bunch of human intervention to do that preparation. That's one problem that AI can solve. The second problem is essentially like network management for us. So there is a universe in which AI could do more for us in terms of scraping our linkedins, our own HubSpot, our own Slack, blah blah, blah, blah blah, to do a better job of understanding and managing our network. And three is the state of the business dashboard that we run the company on requires manual work to basically breed three different systems together to give us a data picture. Those are three workflows that absolutely are automatable by people inside of our company. Now first of all, a lot of people that I talk to aren't even that clear. Unlike those are real things that are real problems that would have real value to us. But then from there it's like, okay, in what order is that most important? And we are looking at our strategy for 2026 to make that determination so we don't get Caught in a loop of just like, well, I want this thing and Mia wants that thing and Colin wants that thing and so we're stuck.

[A]

Yeah, well, and the important thing there in everything you just said is that throughout all of this and well prior to what you just described here, we all have been playing with and trying and learning as much as possible. We didn't kind of like sit on our hands and just put our thinking caps on until we decided which three things we're going to go be precious about. It was like, no, we've all been trying very really small scale experiments on all sorts of stuff. So we even have a sense of what problems could it even be directed at.

[B]

That's exactly right. And to that point, Sam, like one of the things I keep saying to clients is like, and it's not really different than any other kind of organizational change, I keep saying, like, you really have to have the scaffolding to do this. And they nod. But sometimes I wonder if they know really what I'm talking about when I talk about the scaffold around something like AI usage. Here's what I'm talking about. And this I think is a lot of the work that we did for the last year was the scaffolding work. So that we're at this point, it is enough skill building. So not just saying to people like, go figure it out. It is enough encouragement, skill building, et cetera. It is clear constraints around what technology is safe to use and in what ways so that we are legally and ethically compliant with our own standards. It is what principles we are designing for. So like where we think we should use AI and where we shouldn't and what our constraints are around that. And it is some clear ways of working. So it's like we talk about AI in these slack channels and in these meetings and this particular team has this particular remit so that that feels clear and it is clarified in our strategy. There is a specific mention of the intelligence age and its intersection with organizational design. So that this isn't like a thing that just exists in the ether. So like that stuff being clear and explicit and fully understood by the whole organization took real work. Like somebody didn't just write that shit down one day and then it was true. It took six months of experimentation and exploration and learning to get us to the point where we have that scaffolding in place and we can start to say, okay, we have a car that like is pretty roadworthy. Where do we want to drive it? And then we get to those problems that I just articulated.

[A]

Yeah, no, I love that. And I think what I would add to that is some even more general scaffolding which existed in our organization before AI was a word on anybody's lips, which are things like really thoughtful rhythms for where we talk about certain things. So we now have a couple of places where various things related to AI are just showing up in rhythms. We already had. Like, we didn't have to create that from scratch. We already had a operating system where most of us have time and space and the expectation that we should make time and space to learn stuff. And, you know, we're not all 100% fully, like, doing billable hours to clients. And then the third thing that I just jotted down is we have and had before all of this a culture of people wanting to share the cool shit that they're doing. So, like, if you went and like tried a thing with an AI tool and it worked, or even if it didn't, like, there's a lot of stuff popping off in our slack all the time where someone's like, check this thing out that I tried or check this out. Like, look what they're doing. And that all pre existed, you know, the AI moment that we are currently in.

[B]

Yeah, I think that's right. And I heard someone say this at a conference. Now I can't remember who it was. Sorry. But I think there's this interesting tension between what it is doing and sort of policy making and the desire for experimentation. And I feel like often it's sort of categorized as like, it just wants to like, lock everything down. But the CEO wants everybody just like trying stuff and finding efficiency gains. And I actually think neither of those things are totally true. I think it a third way.

[A]

You think, right?

[B]

I think there's a third way. I think those functions feel a real responsibility to mitigate risk of technology that is largely untested at an enterprise level and to set enough constraints and create enough safety features that they don't have massive data breaches or they don't produce work slop as an end product and then get sued at hem big consulting firm and I have to pay a bunch of money back. Like, I think it's reasonable that those functions are trying to carve that off. And I don't think that most of them are trying to quell experimentation. And on the other side, I certainly know that this was true of me and other people at the ready and other like, peers of mine that steward other companies. A lot of us needed those kinds of constraints to feel safe Experimenting, Sure. I wanted to try stuff. I had been reading, I was ready and like I didn't want to do something up and I just didn't feel like I knew enough to know what those guardrails were. So I think in a perfect world actually these things should be integrated and generative and instead they're seen as polarities. Intention.

[A]

Yeah. Yeah. I think there's something about polarity, polarities in tension which somehow feels like cleaner, like you're like in one camp or the other and like it's easy to know like who's your ally and who's your enemy and like those are roles that we know how to play generally like in our society and this like in between kind of generative third way. I don't know, I think it comes harder to a lot of folks. It's not as clean, it's messier, but more generative.

[B]

Yeah. So what that looks like is like if your it, whoever has the remit in it is writing like dozens of pages of policy documentation that is essentially useless and illegible to the masses. That's not it.

[A]

Yeah.

[B]

And if your organization writ large is experimenting completely unconstrained on their own tools, with their own personal email addresses, feeding God knows what, that is also not it. Like you gotta get to minimum viable constraint. And I know, look, it's different when you're working with an organization that has 300,000 people. I'm mostly not talking about that. I'm mostly talking to like the mid sized companies that really want to get after this. You got to do the work of what minimum viable constraint is within which people can go buck wild.

[A]

Right, right. So the whole point of experimenting is not so that each individual experiment is conducted and dies on its own, it's so that each experiment and the result from it builds on top of each other into a body of knowledge, a body of practice across the organization. And that's true whether we're talking about AI or any other way of working or org design sort of thing. And in order for that to happen you do need those constraints. You need to some scaffolding for experiment results to build on top of each other over time.

[B]

Yeah, exactly, exactly.

[A]

I think one of the things that I'd like to linger on for a second again and this is from a recent conference talk that I attended and the speaker just straight up said like it's important to remember nobody is an expert in this yet. And specifically the applying AI to organizations. So we've always, I've always found myself saying things like you know, there's no playbooks for like the type of organizational change that we do. And I've always kind of gotten the sense that even some of our best clients would be like, would like nod along, but they would be like, but there's actually like kind of a playbook, right? Like there's, there's sort of a playbook. Just give me the playbook.

[B]

Yeah.

[A]

And like, but there, there this time, like for real, for real, there are no playbooks. And not to make this an episode about like traditional consulting versus like the stuff that we do, we have a whole episode on that. But I do think that is an interesting place for consultancies that have traditionally been paid to sell you answers, an interesting place for them to be. Because anybody trying to sell you a kind of end to end answer in the AI realm and like how it's applied to your organization is selling you a story and they are trying to separate you from your money via the force of fomo. And that is, I don't know, I think about that a lot as we try to like position ourselves like anybody else who helps organizations. It's like, how do you talk about this and bring ideas to an organization when there truly is not, not a playbook? I find it invigorating, but it's also really hard.

[B]

Yeah, I mean the number of places, like whether it's you know, really big consultancies, you know, I think Accenture said they're going to hire 80,000 people in this space, which is, I'm like. To do what? Whether it's like big like that or like, you know, companies that I have a lot of respect for in the AI space and I see what they're talking about in terms of like their quote unquote transformation offering. And it is almost to a company. Make your plan. Like we'll consult with you to create your AI strategy. It feels very early to be overly focused on what the plan is going to be. And I feel like right now what I would like to see more companies doing is getting really clear on their intention, getting really clear on what they're designing for. And when I say what they're designing for, it's like, are we designing for more automation in our tool? Are we designing for more self service operation for our employees? Are we designing for highly paid knowledge workers to not be spending time on admin? Like what are we designing for? I would like for there to be more time on that and then I would like to see more time on what the guardrails are within which Experiments can happen. And then more storytelling, more demo days, more working in public, and less focus on like what the fucking roadmap is. Who cares what the roadmap is? Like the roadmap from six months ago is garbage. And the roadmap engineer will be garbage. Like why? I mean, look, I get, like, I keep saying this on this show. I get that that's a much better business to be in than ours. And it's like nobody really gets held to account for selling plans that don't come true. But I feel like this is a time to really lean into the idea that you're not behind. You just need to have some disciplined experimentation so that you're really learning. And in many ways, unfortunately, learning and planning are in conflict.

[A]

Yeah.

[B]

And you can't do both as priority number one.

[A]

Yeah. And if you separate them out, then they're not, then. Then you really can't do them. I think learning and planning that is like intermingled in a really complexity conscious way. That's cool. But that's not what you're talking about generally.

[B]

No, it's not. And what I'd like to see is like really minimum viable plan and like as much learning as they can stand. And mostly what I see is the opposite is like, let's spend a lot, a lot, a lot of time analyzing what we might do and then a little bit of chaotic energy on trying a bunch of shit to see what sticks. And I'm like, what if you did the opposite of that?

[A]

Yeah.

[B]

I'll give you an example. I talked to an old client the other day that I think is getting this right. They're one of the only companies that I've talked to that I'm like, this you guys are on one is like one of the ways that they're getting after this. Like they've recognized that they're not going to turn their entire workforce into workflow engineers. Like it's just not going to happen. And that they actually do want inspo and use cases and problems to be solved from the edge of their organization. And so they've created a fairly small technical team. And there is a way to submit ideas for automations that can and should be built. Those are public and upvoted. And then the technical team sprints fairly quickly on cranking out MVPs. And I'm like, that's cool. You know, it's done in public, it's retrospected. There's a monthly cadence to talk about wins. Things are sunset that don't get traction. It's a very like agile way of doing this. But I'm just like, that's so smart. That's such a better idea than like a really fiddly, complicated two year roadmap that is just like garbage.

[A]

Totally. And it makes me think about the other kind of like main thing that I've been kind of just stewing on with all of this, which is I think a lot of organizations have kind of, you know, that idea of brainstorming. I don't know if it's from brainstorming or. No, it's from improv. First idea, best idea, which is like an improv. Like just do the, like the first thing that comes to mind. And I think in a lot of organizations they're doing that with AI in the sense of cost takeout. Like first idea, best idea for most organizations with AI is like, oh cool, we don't have to pay people anymore. Great, let's get them out of here and get some AI in here. And I think it is such a impoverished way to view what is possible here. You know, if, if broadly you can break your whole organization down to like the value you create and are able to get from your customers over the cost it takes to make that value. What if we focus so much more on the numerator and less about making that denominator as small as possible? And that's, I hope to see more organizations move away from that really uncreative and I think ultimately kind of doomed way of applying AI into their organization.

[B]

I agree with you. To me this is no different than the CEO who says the strategy is a metric. Yeah, I'm like, no, no, disagree. It feels like the same thing. It's like you are saying the outcome is the work. That's just like cost takeout or headcount reduction or productivity matrix. Those are going to be lagging indicators of the complex body of work. Yeah, Focus on the complex body of work and like leave the lagging indicator to lag and indicate. Don't start with that thing and be like, how do we get it right now real quick? Because what you're going to do is likely not going to be sticky, going to create increased fragility. See also the number of press releases that are like, we fired all these people and then we had to hire them back and it's going to look really good in the short term, but it's actually not going to help like really change the genetic material of your organization to include AI in it. And like that's what you should be going for. Right now, like, anybody can do a cost takeout. Like, and anybody can hire a consultancy to do a cost takeout. Like, it's truly not rocket science. And if you're trying to shift to a new way of being and of operating and of deciding and of executing, like, you don't start with that. Yeah, you start with what you're trying to be or become. And that might be a byproduct. And I think for most companies it will be. Look, I think that like, I do think we're going to see mass unemployment. I do think we're going to see a huge rationalization of the workforce. Like, I do think we're going to see all of that. But I think the companies that will do the best will have that be a byproduct, not the main event.

[A]

Yeah, I think organizations have always, like, if you struggle with the difference between outcomes and output, but you like always kind of have some weird stuff going on in your organization. And I think if you are struggling with it and then you apply AI to it, you're really just ramping it up. Because I think when we're making this argument, which I think is fundamentally like in some ways a pretty like pro human argument that like totally for humans in our organizations in the future, really putting a flag in the, in the, in the sand here. But I, there's something about if we are trying to use AI to keep work into like human sized packets, we're doing something wrong. And what I mean by that is all of the kind of chatter around or like usually kind of like memes or jokes around how like I asked AI to like take this like bullet list and make an email and then the AI sends it to you and then your AI take that took that long email and broke, broke it down into three chunks or like, you know, I'm making like 10 times as many PowerPoints as I normally do. And then your AI is like reading like that is like the skeuomorphic idea of AI that I think we're in and we'll have to work our way through because like using our human brains to basically just accelerate our human workflows feels like such a, a miss.

[B]

Yeah, yeah, it's such a great point.

[A]

But it's, it's hard to like articulate what the new thing is because I'm using my human brain to like try to do that, which is inherently limited to like what I have experienced in many ways. And like it defaults to like human sized conceptions of what work looks like.

[B]

But these are the creative exercises that we should be undertaking. And I feel like to your point, a lot of the memes are in the sort of VEIN of the two things that you just said, email and PowerPoint. Just take those two as example. To me, email and PowerPoint are the two worst trappings of the inefficiency of knowledge work. Like, if you summarize everything that's garbage about working in large organizations, those two things could be the logo and the mascot. You know, so it's like we don't need to have agents making tons and tons more and faster of the worst aspects of work. Yeah, let's just like start there. Like, let's start from that premise.

[A]

Let's not just like more of the worst stuff.

[B]

Let's not make more of the worst stuff.

[A]

I like, I feel like that's a.

[B]

Conversation that I would love to be having more with clients. Yeah, yeah, yeah, totally. So one of the things that I do want to ask you is just what about this are you hopeful about? I think we've talked a lot about what we're seeing out there and why we think it's not the way. What is the way? What are, what are you bullish on?

[A]

I am not confident enough to sit here and say, here is hot take zone. Hot take zone. We kind of retired the hot take zone. We should bring it back.

[B]

I, I, I think it was brought back a few weeks ago and I was very surprised and delighted by it.

[A]

Okay, great.

[B]

You not listen to our podcast. You should. It's good.

[A]

Which is, and I, I feel like I've, I feel like I've mentioned this somewhere, but may maybe not. But I think there is something interesting about, you know, so already we're self managing. Right. And you know, you don't have to be deep into the theory of like holacracy and sociocracy to like basically understand that, you know, we don't have a traditional hierarchy, traditional org chart, nested teams and circles and authorities that work together to do stuff. And there are other, some quite large organizations that have done interesting experiments around various flavors of self management. And I think there is a future, a potential where AI could actually bring self management or things that look and feel like self management, at least some of the principles of self management to more organizations. And partially this is just a feeling and partially it is a, a sense of like, what happens when you push creative power to the edges of the organization. Self management is all about that. And for many, many years, like, there are ways to do that in your operating system and your ways of working. But nothing like AI obviously has been on the scene to really accelerate that. There's often a challenge around. I mean it's true in all organizations, but especially self managing around kind of like information overload and like what information is relevant to you in your roles. And I'm picturing an AI where it really deeply understands your role that you're playing in the organization and what information you need for and who you need to be connected to. And it's a complex network because it's a huge organization and it's not something you can just look up on an org chart and boom. Like your AI is like built your personal like individual dashboard for like what you need to do your role. And suddenly I feel like we're like unlocking new ways of showing up in an organization. That actually makes me really excited. And I notice that point here. Did I talk about stripping out cost or replacing people with AI in any way? So there's something in that future that I hope we continue to push toward as a species.

[B]

I think that's really interesting. I was listening to a podcast the other day about how what AI is really good at is like taking out the translation layer in organizations. And I think in most organizations the translation layer is way too thick. Like there's way too much time spent massaging and stakeholdering and rewriting and blah blah blah blah. So like on the one hand yes, AI is great at that. Conversely I think in self management there is no translation layer and that is also very challenging. So whereas in a traditional organization it's too thick and it's wasteful and ineffective in self management even at a small scale what you tend to have is people feeling either information overload or not clear on what's important or not sure how to Basically it's hard to echo locate in a self managing system. I think you're right that the kinds of problems that exist in a non trad company that have not really been solvable before, like really visualizing how the network is working, what people are working on in an atomized way, how skills are married to needs, what people's sort of reputation and capabilities are like pull based strategy in order to make resourcing or other decisions. Like these are things that were really hard to crack using Google Drive and notion. But I think you're right. Like I think a knowledge body that we can be in discussion with as individual users with individual needs in order to echolocate has been a missing piece that frankly is like the fuel in self management. And I too Feel like quite bullish on that. And I think unlike in trad organizations where it's going to strip something out that's like ultimately kind of sludgy and not that helpful in self management, I think it potentially adds something that has been missing.

[A]

Yeah, that's hopeful. I enjoy thinking about that.

[B]

Do you have any ideas you want to tell people about, Sam that we didn't already tell people about?

[A]

I think just we've talked a lot about experimentation.

[B]

We have.

[A]

So the only little bit that I want to add to that is, yes, experiment to see what works, but I think particularly in AI's case right now, experiment to break it it. We need, you need to really understand what it's not good at so that you're not in the default position of just trusting everything it says because boy howdy should you not do that. So you should really get a good texture, get a good feel for when the AI is full of shit. And the only way to do that is to try to break it and break it in, you know, in domains where you know what the answer is and what it should be saying and noting when it's not. And like making sense of that.

[B]

Yeah, I think that's really smart. I'm going to take the individual lens on this because, you know, we're starting to see studies come out about like how much retention we have of our interactions, particularly with chat. And it's not good. It's, you know, it's really making us dumber. I feel like as an individual, so, you know, we've talked about experimentation and constraints and system design and blah, blah, blah, blah. But at as an individual, what I don't think most people have really honed a skill around yet is like, where is this making me better and where is it not? And what is the partnership that feels like augmentation for me and that pushes my thinking and that makes me sharper or more creative and really like not just more productive because again, more productive is fine, but like I wouldn't as an individual with any authority in any company to do anything. I wouldn't be orienting right now to just like, how can I be more productive? I'd be like, how can I be more of whatever I'm trying to be? How can I be more original? How can I be more creative? How can I be more iconoclastic? How can I be better at sensing trends? How can, like I would tune into what I'm trying to do and I would be experimenting with how my partnership with this technology makes me better at that thing because in many ways, the worst use cases are outsourcing your thinking and just getting dumber. And the best use cases are, like, how to hone and sharpen and use it to challenge you.

[A]

That's motivating. We're to end this episode so I can go, like, think about that some more.

[B]

Go get it. Go get it.

[A]

Yeah.

[B]

Cool.

[A]

All right. Should we wrap here?

[B]

Yeah.

[A]

Thanks to the first swing at this AI AI episode. So we can come back to it in a year and be like, holy.

[B]

We were wrong about everything.

[A]

They're totally different. All right. We're always looking for new topics for the show, so if you have an org pattern that you're having trouble changing, shoot us a [e-mail] this show is.

[B]

Engineered by Taylor Marvin and produced by my friend, Jack Van Amberg. He's our friend, but I'm my friend, too.

[A]

I felt this incredible urge to hop in and say, he's also my friend. So thank you.

[B]

He is our friend. I got. I was sharp. I was thinking about Saturday. I didn't mean to leave you out. Sam at Work with the Ready is produced by the ready, where we help organizations around the world get ready for this moment and change the way that they work with AI. Thanks so much for listening.