We all work within intractably large and complex systems. How much agency do we really have? This was the question Harry and I explored on episode 40 of the Traction Heroes podcast.

Harry read a passage from Eric Ries’s new book, Incorruptible (which I haven’t read.) It starts by citing John Steinbeck’s story of depression-era farmers dealing with bank officers who’ve been rendered impersonal as agents of a large organization. Organizations are a kind of being:

Living things maintain boundaries between themselves and their environment, resources into energy. They grow, adapt, and reproduce. They exhibit behaviors that emerge from their parts but can’t be predicted by studying those parts in isolation. Most importantly, they display a will to survive that shapes every action they take. Organizations exhibit every one of these properties.

The organization is a superorganism that demonstrates emergent properties. Among these, is a kind of intelligence. This is the entity the farmers were interacting with through its agents.

We’ve all had experiences dealing with such complex emergent systems. One might feel disempowered when considering the organization forms part of a broader socioeconomic system that incentivizes it to act in particular ways.

Incorruptible sounds like an antidote to that kind of thinking. Our socio economic systems provide alternate structures and ways of being — and we can choose them: As Harry put it,

the default world that we’re in is just a construct. You really have an enormous amount of flexibility. You have many more degrees of freedom than you are probably aware of. It doesn’t serve the system to inform you of what those are because those degrees of freedom give you more control and give you more agency.

This conversation was empowering and energizing. As I said in the podcast, this book has moved up my queue — check out the show for more.

(On a separate note, this is the first episode we’ve released as a proper YouTube video. We’d love to know what you think — please comment in YouTube itself.)

Traction Heroes episode 40: Agency

The Kinetic Council invited me to join a panel with Abby Covert, Jessica Talisman, and Larry Swanson about the evolving role of information architecture in a world where AI exists. Things got salty! 🧂

Moore’s law was in full force in 1986. Hardware was getting more powerful, faster, and cheaper. Surely, some technology would come along to do the same for software. Brooks argued this wasn’t in the cards, since software is fundamentally different from hardware. For one thing, it’s of a different order:

The essence of a software entity is a construct of interlocking concepts: data sets, relationships among data items, algorithms, and invocations of functions. This essence is abstract, in that the conceptual construct is the same under many different representations. It is nonetheless highly precise and richly detailed.

Specifying, designing, and testing this construct is difficult. The challenge isn’t implementation but design: “We still make syntax errors, to be sure; but they are fuzz compared to the conceptual errors in most systems.”

Technical advances usually make development easier. No Silver Bullet traces the history of time sharing, unified programming environments, and high-level languages. Object-oriented programming was a promising new technology at the time and there were even rudimentary AIs in the form of expert systems. Brooks examines them and concludes they’re not enough.

Why? Because coding isn’t the hardest part of making software. Instead, the hard part is knowing what to build:

The hardest single part of building a software system is deciding precisely what to build. No other part of the conceptual work is so difficult as establishing the detailed requirements, including all the interfaces to people, to machines, and to other software systems. No other part of the work so cripples the resulting system if done wrong. No other part is more difficult to rectify later.

What will the system do? How will it serve strategic objectives? How will it enable better judgment and allow people to derive meaning from data? These aren’t implementation questions, they’re design questions. Somebody must define the “construct of interlocking concepts” that define the system, aiming for good fit between the system and the context it serves. LLMs can help, but they can’t replace human understanding and judgment, at least not yet.

Brooks calls out four inherent properties of modern software systems:

• Complexity: Software systems are among the most complex human constructs. They’ve only gotten more so as computers and operating systems have grown more powerful and capable.

• Conformity: Software solutions must conform to the goals, needs, constraints, and interfaces of the organizations that bring them forth. This is true whether it’s bought off-the-shelf or developed bespoke.

• Changeability: Anything that lasts does so because it’s able to adapt to changing conditions. Software is inherently more malleable than other complex designed systems, such as buildings.

• Invisibility: Whereas complex physical systems (again, think of buildings) can be represented with mechanical drawings, software specs are inherently abstract. This makes them hard to design.

There’s been progress in the last four decades, but these properties remain fixed. LLMs haven’t changed that. Non-deterministic components add immense complexity and unpredictability to software systems. The ease, speed, and volume of code generation make software more malleable and opaque than ever. And LLMs promise to ease bespoke development, tempting orgs away from one-size-fits-all solutions.

Which is to say, LLMs haven’t changed the nature of software. Instead, they’ve made it more so. So far, the technology’s killer application is developing software: teams can now produce more software, faster. (It’s unclear yet whether it’ll ultimately be cheaper, especially when you consider maintenance costs.)

What LLMs haven’t done yet is replace software wholesale, at least not for tasks that require predictable behavior. And as their true costs and constraints become evident, it’s increasingly doubtful they will. Instead, LLMs will likely become part of systems that include traditional deterministic components — both inside the systems and as part of the development process.

The resulting systems will be more complex, malleable, and abstract than prior ones. They’ll also be better fit to purpose than off-the-shelf solutions. But that requires design, which remains primarily a human challenge. And it’s hard:

it is really impossible for clients, even those working with software engineers, to specify completely, precisely, and correctly the exact requirements of a modern software product before having built and tried some versions of the product they are specifying.

Replace “software engineers” with LLMs, and this sentence still stands. But it also hints at where LLMs come closest to being a silver bullet: in their ability to spin up rapid prototypes. Good software is grown, not built. That is, it evolves from an initial core to a more complex system through an organic approach that respects Gall’s law:

The building metaphor has outlived its usefulness. It is time to change again. If, as I believe, the conceptual structures we construct today are too complicated to be accurately specified in advance, and too complex to be built faultlessly, then we must take a radically different approach.

Let us turn to nature and study complexity in living things, instead of just the dead works of man. Here we find constructs whose complexities thrill us with awe. The brain alone is intricate beyond mapping, powerful beyond imitation, rich in diversity, self-protecting, and self-renewing. The secret is that it is grown, not built.

So it must be with our software systems.

What was true then is true now: technology moves the bottleneck from production to orientation. LLMs make coding easier, much like high-level languages, IDEs, and compilers did in the past. But without shared models, structured context, feedback loops, governance, and clear interfaces, they won’t provide the results leaders expect.

As always, how to build gets easier — knowing what to build doesn’t. AI can help with that too — but it needs steering. The question isn’t “Which systems can we replace with AI?” Rather, it’s “How can AI help us grow systems that better fit our needs?” The answer will consider AI as a system component and a production tool. But forty years on, there’s still no silver bullet — just better ways to find good fit, faster.

AI lets us prototype faster than ever before. That’s great, as long as you remember there’s underlying work to be done. Specifically, production systems require structural underpinnings “screen-level” prototypes often miss.

Ironically, this means prematurely-rich prototypes can impede traction. So I wanted to discuss these issues with Harry. I kicked things off with a reading from Stewart Brand’s The Clock of the Long Now:

In recent years a few scientists (such as R. V. O’Neill and C. S. Holling) have been probing a similar issue in ecological systems: How do they manage change, and how do they absorb and incorporate shocks? The answer appears to lie in the relationship between components in a system that have different change rates and different scales of size. Instead of breaking under stress like something brittle these systems yield as if they were malleable. Some parts respond quickly to the shock, allowing slower parts to ignore the shock and maintain their steady duties of system continuity. The combination of fast and slow components makes the system resilient, along with the way the differently paced parts affect each other. Fast learns, slow remembers. Fast proposes, slow disposes. Fast is discontinuous, slow is continuous. Fast and small instructs slow and big by accrued innovation and occasional revolution. Slow and big controls small and fast by constraint and constancy. Fast gets all our attention, slow has all the power. All durable dynamic systems have this sort of structure; it is what makes them adaptable and robust.

This is a succinct articulation of pace layers, an important idea popularized by Brand. Complex systems are made up of diverse components and subsystems that change at different rates and scales. Resilient systems use the faster layers to experiment with new approaches and the slower layers to “remember” those best fit to purpose. Thus, the system evolves while maintaining coherence.

Brand showcased an early version of this model in his book How Buildings Learn: What Happens After They’re Built. As the title suggests, that version focused on explaining how buildings evolve. The model in The Clock of the Long Now casts a larger net, explaining how civilizations change over time.

Which is to say, this model can be generalized to other complex systems. That’s why it’s at the core of my book, Living in Information, where it explains how information environments remain resilient. That version offers five layers. Here they are, in order of fastest- to slowest-changing:

• Form: the user interface
• Structure: the underlying information architecture
• Governance: how the organization manages change
• Strategy: what the organization does differently to win
• Purpose: why the organization exists

As with Brand’s version, the slower-changing layers are more powerful. But the fast layers also play an important role. A healthy system needs both. Therein lies the challenge: AI’s ability to produce rich prototypes can lead us to hang out too long on the fastest layer. Eventually, alignment and coherence require heading down the stack.

What’s exciting is that this isn’t a return to the slow old days. The same tools that can help us with form can help us with architecture and strategy. But they do require attention. You can start by asking yourself two questions:

• What layer am I acting on?
• What layer should I be acting on?

If there’s a misalignment there, you want to fix that first.

The form layer is easy now. And that’s great: it means we can better describe what we want. But you can’t leave it there. Somebody must look after alignment with the underlying layers — especially strategy, governance, and structure. That’s the work I’m increasingly doing, so it was great to discuss it with Harry.

Traction Heroes episode 39: Pace Layers

In systems thinking, this principle was best articulated by John Gall:

A complex system that works is invariably found to have evolved from a simple system that worked. A complex system designed from scratch never works and cannot be patched up to make it work. You have to start over with a working simple system.

I’ve long promoted Gall’s law to students and clients. It’s been hard going. We want to see products and services in their full glory ASAP. But you can only throw so much cash and person-hours at a problem. Ergo, we got the time-tested idea of a minimum viable product. (It’s no coincidence that orgs with more resources violate Gall’s law more often than scrappy startups.)

But the value of an MVP isn’t just that it allows you to get something that works quickly and cheaply. Instead, the value is that that first try isn’t overspecified to theoretical conditions. It’s only a draft meant to kick off an evolutionary process that leads to a system that meets real-world customer needs.

AI removes these constraints. An afternoon with Claude can yield a comprehensive spec for a very complex system. Further sessions can architect the system and build an initial release that includes bells, whistles, timpani, harps, violas, and all the other instruments in the orchestra. All this, at a fraction of the cost and time it would’ve taken in the past.

That’s amazing. It means we can now design and build much larger systems, faster. This opens new possibilities. Not just one new feature, a new product. Not just a new product, a suite. Not just a suite, a platform. The possibilities seem endless.

But in removing architecture and development constraints, we’re also removing the need to focus on what matters most and the discipline to run it by the market. All that upfront complexity doesn’t necessarily address real-world needs. Instead, it reflects a singular top-down vision that may or may not provide value to others.

Whereas MVPs in the “before times” called for minimal investment before validation, LLMs promise a much more realized vision from the first go. But there’s a significant difference between a sexy concept car and a vehicle customers will take grocery shopping. AI-augmented workflows will give you the former, but not the latter.

Here’s an example. Two years ago, I started building a product called SiteRanger, an AI-powered agent to help small teams manage large websites. This was before Claude Code or any of the current coding agents. Still, I got surprisingly far by using Claude to augment my basic PHP skills.

Together, we built a functional MVP that implemented what I considered to be the core functionality. The problem: my “core” was in fact an open-ended platform. Rather than solve a particular customer problem, it was designed to solve classes of problems.

By the time I got alpha users on board, the system was wildly over-architected. Worse, I learned new agentic systems could provide ~80% of its value. When I looked to pivot, I realized I’d have to move in a completely different direction, one with entrenched incumbents. It wasn’t worth it.

It wasn’t all a loss. This experiment taught me a lot about developing AI-powered software products using AI. But the most important lesson I learned is that AI makes it very easy for individuals and small teams to land in the same trap as resource-rich orgs: no constraints.

Which isn’t to say you shouldn’t use AI. To the contrary, I’m all for accelerating MVP design and development. But the word “viable” is fungible, especially when you have robot engineers. You want to expose the product to the discipline of the market. That means releasing something embarrassingly simple at first. And that requires discipline and constraint — the two things most scarce when working with LLMs.

My friend Karl Fast pointed me to a wonderful scene in the movie A RIVER RUNS THROUGH IT. The main character, a child, brings an essay to his dad, a strict preacher, for evaluation. The dad’s only reply: “Half as long.” The child does, and returns with the edit. After scribbling with a red pencil, the preacher looks at him and says: “Again, half as long.”

Step away from the console and ask yourself: What would I cut if there were no AI building it? Cut, cut, cut. Then imagine Tom Skerritt staring at you over his schoolmaster glasses and saying drily, “Again, half as long.”

The key distinction is between management through principles and top-down management by rules. The latter is impracticable because managers aren’t omniscient. Arbitrary constraints proliferate in such a system, leading to stasis and entropy.

How do you keep things moving in the right direction without proliferating rules? By infusing the organization with principles. The book showcases 41 that made the Kochs successful, including Integrity, Openness, Property Rights, Mutual Benefit, Personal Knowledge, Stewardship and Compliance, and more.

Some of these are more obvious than others. Sidebars throughout the book highlight each principle as the authors narrate stories from their careers that exemplify how their organizations infused a principled culture, stayed ahead of change, out-competed other organizations, influenced policy, contributed to the broader well-being, and more.

The book starts by suggesting it’ll unfold from the perspective of both Charles and Chase, who also works at Koch Industries. But apart from a couple of autobiographical chapters at the beginning, most of the book told from a unified POV. (It’s clear Chase has internalized his father’s worldview.)

The two autobiographical chapters comprise the first part of the book, which is about applying principles to transform your life. The second part is about transforming your business. The final part is about transforming society. It applies the Koch’s principles to thorny social and political issues.

Business books often gloss over mistakes to elevate the authors. This one doesn’t. The Kohns own up to numerous mistakes, including one that cost the lives of two teenagers. Rather, they often own up to their companies’ managers’ mistakes, something I found a bit off-putting. Still, the message is clear: principled decision-making helped them get ahead — and they got in trouble when eschewing such principles.

The world would be better if principled people were in charge. I happen to agree with most of the principles in this book, but IMO the broader point is driving change through clear and well-grounded principles rather than top-down rules that constrain human agency. In promoting this mindset, this book provides a blueprint for building a more harmonious, adaptive, productive, and fair world.

Avoiding catastrophic failure is essential for gaining traction — and many so-called accidents can be prevented with a bit of forethought.

This idea was at the core of episode 38 of Traction Heroes. Harry kicked it off with a short reading from Alan Levar’s There Are No Accidents, Only Collisions. Here’s a subset:

It’s stunning how often simple safety checks were not carried out on tractor-trailers that subsequently crashed. Things that could have been fixed easily. Human error or judgment is not usually the cause of the collision, but rather it is the driver’s failure to make simple routine checks he should make every time he gets behind the wheel to make sure his truck is safe as possible and he won’t needlessly endanger the public. These checks usually take only five or ten minutes before the driver gets behind the wheel of the truck.

Of course, I wanted to probe further. How do these checklists come about? Are they provided by experts? Compiled from firsthand experience? Both?

In any case, this is actionable knowledge — and it can make an important difference. In the absence of checklists and standards, individuals and teams can “drift into failure,” inching toward catastrophe. Success calls for mindfulness and continuous vigilance.

What can you do besides compiling checklists? You can conduct post-mortems after something happens. Perhaps more usefully, you can also conduct premortems before projects begin. Both post- and premortems can inform checklists.

Good leadership calls for clarity, intentionality, and accountability. Establishing routines and standards — and staying on top of them — can help avoid failure.

Traction Heroes episode 38: Checklists

In the third of our Open-Ended Sessions, Greg and I discussed how AI is changing how product and design work are done. In particular, we’re tracking the shift from feature- and screen-level work to more strategic and human-centered system design.

Lots of orgs are choosing to deploy AI as a means to do more of the same, only faster and cheaper. But AI can also be used to unlock new possibilities by augmenting (rather than replacing) humans.

As Greg put it,

The efficiency play can be in service of unlocking human potential, right? So they don’t have to be either-or paths, but they do have to be both at a minimum.

Ultimately, the key question isn’t what the technology is capable of, but what it’s for.

Links

Books and posts mentioned in the conversation:

• Magnifica Humanitas by Pope Leo XIV
• I think Anthropic and OpenAI have found product-market fit by Simon Willison
• The Cost of Being Busy by Craig Hepburn
• The Feature is Dead. The Job Remains. by Greg Petroff
• This Moment We’re In, Ep. 3 (Greg’s conversation with Cindy Chastain)

Transcript

(AI generated — likely contains errors.)

Jorge: All right, hey Greg.

Greg: Hi Jorge, how are you?

Jorge: I am doing all right.

Greg: Yeah.

Jorge: It is interesting times.

Greg: Oh my gosh, yeah. You know, I think when we were talking about perhaps hosting one of our sessions here, it was really about the zeitgeist, like what’s happening this moment. And there are these peaks where all of a sudden there’s some kind of thread that shows up that just begs to be probed. So thanks for encouraging me and us to have these conversations. And for those who are joining us, welcome to number three of our Unfinishe sessions, where Jorge and I just talk about stuff that we think is interesting that’s going on. And we hope you find it valuable as well.

Jorge: Yeah. And it might be worth recapping beyond the open-ended conversations. This is not just something that we’re doing because we want to talk about things that interest us. In some ways, it feels like the transformations that we’re seeing are, sounds kind of heavy-handed to say it, but they’re existential. They kind of are. And these conversations I see as an opportunity to name the things we’re seeing. You have your note taker just joined when he’s trying to join. You see, it is existential.

Greg: That’s right. There we go. Goodbye.

Jorge: It might be worth unpacking what that means, but we have not pre-scheduled these, to your point. We’re kind of calling these conversations when we have sensed a shift in the zeitgeist. And this is the third one we’ve had so far. And I’m going to try to sketch out the conditions that precipitated this one. And then we’re going to circle back to a post that you shared, which I see as kind of like a way to deal more skillfully, for organizations to deal more skillfully with the situation that they find themselves in now. But there are several factors at play here. One is the big shift toward the end of last year. I think we saw a big shift in how people in general are thinking about AI. Agentic software development tools like Claude Code started proving their mettle in the organization, right?

Greg: Yeah.

Jorge: And Simon Wilson recently had a blog post where he said that Anthropic and OpenAI have found product-market fit. And he was referring specifically to these tools, right? So there’s been this shift from a modality that is more focused toward software development. And then we have things like Claude Code that build on that agentic way of working with these systems. So that was a big shift toward the tail end of last year. And then I would say that around the end of the first quarter of the year, we started paying the piper, literally. Organizations have started realizing that that kind of usage can get very expensive, right? And we’ve started to see some organizations start to pull back, capping their people’s budget for using these tools.

Greg: I would just build. I’ve been working as a fractional for a couple of different companies, fractional design leader, for those who haven’t heard that term before, over the last year. And to watch how product organizations have started to use these tools, and then the uptake in them, and then the capabilities that they unlock, and then the speed that teams can work at, and more importantly that you can do more, it’s been really remarkable to watch. And at the same time, it’s causing all kinds of churn because teams are struggling with who does what, how, when, and they’re hitting some of these issues too around utilization and use of the tools and the cost of using the tools, and not just the financial cost of using the tools, which can be significant when teams burn through their tokens, but the cognitive costs of using these tools, because you can build incredibly dense, rich, powerful documents now, but then your peers have to have time to consume them. And one of the things I’m noticing is that many of us are collaborating with our AIs more than we are collaborating with each other. And so there are all these kinds of things that we’re learning in the process of using these tools. And I also think there was some hyperbole around what was going to happen and how this might change things and the number of people, etc., that I think is starting to not show up. I think there are some things that we can think about that are different. So in this moment, yeah, we saw toward—

Jorge: Beginning of the year, organizations were starting to boast of how many tokens they were using, and they had leaderboards for this stuff.

Greg: Yeah.

Jorge: Which is another factor here that I think the emphasis was clearly on adoption as opposed to value creation. And one of the things that has shifted is, well, first they’ve stopped doing that because I think they realized, like, hey, this is really expensive. But also, that’s no way to measure progress, right? What you want to do is you want to be actually creating value. You don’t want to be boasting of how much you’re using the tools, right? So it feels like we’re kind of speed-running the process of maturing into how—

Greg: These tools can serve human needs. Yeah, there’s a whole set of new problems that are showing up inside product organizations around utilization. Are you using the right model? Are you leaving your context window open too long and therefore every time you ask for something you’re burning through credits like mad? Did you budget appropriately for the amount of credits that your team is using? What does finance think about all of that, right? I think there’s a whole bunch of things that are popping up right now. And some of that showed up in some recent announcements with Microsoft deciding to turn Claude off inside of their environment. There’s been conversation about bringing their own code development tools and their own models to bear, so maybe that has part of it. But I also, from what I understand, they were spending a lot of money. And my own experience in watching the team that I’ve been helping, the startup that I’m working with, is we run out of credits on a regular basis, or tokens on a regular basis. And then we have to go and ask for more. And there’s no real governance or process in place for a small company where they’re kind of making it up as they go and trying to understand what it means. They want to go fast, and so they’re willing to spend the money, but they have finite resources. So they can’t spend as much as they maybe think they should, or as the team wants to spend potentially. And there are all sorts of stories around teams that have burned giant holes in the budget of their organization by doing things which may not have been valuable, right? And it goes back to your point around this leaderboard thing. And I think one of the things that there’s a really great post by Hepburn about the cost of being busy that I think is really interesting to me. It was really about the teams that give the tools so that AI can be adopted throughout the organization and people sort of adopt them at their own speed are not really getting all that much productivity gain. They’re getting richer content perhaps, but they’re not institutionalizing the work in a way that takes the work that’s repetitive or less of priority and maybe building agents around that so that you can unleash your people towards new things. And I think there’s another aspect of this, which you and I started, I think this is one of the reasons why I wanted to talk this week, was Pope Leo’s encyclical came out, and it kind of fits a notion that we’ve been talking about for Unfinishe, our practice, which is we want to help organizations discover how they can, you know, the possibilities of unleashing their talent towards new efforts versus a productivity gain where you’re laying people off because you don’t need them anymore, right? I think it’s a very different mental model. And I think that there’s something about this conversation that needs to come up around the humans in the system and the value that these tools create for us, but also the value of work, etc., that I think is a notional, the conversation’s happening now and that’s good. And I think that document was a really important document to come out. And there’s been a lot of conversation around it. I know you have a take on it, but some people will leave it at that and hear what you think.

Jorge: Yeah, and I’m glad you brought that up. And even though this is turning into a kind of long preamble, I also want to bring to the table the conversation that you had with Cindy Chastain on her podcast.

Greg: Yeah.

Jorge: Because that conversation also circled around some of these issues. And I’m going to try to name it, just hearing you talk about it. I’m going to try to name the connecting thread for all of these things that we’re observing. And the thread is the question: but what is it for? It’s like, what are we doing with this, right? What’s the point? The conversation with Cindy, one way to summarize it might be something like: for design and product teams, there are two possible things that you can do with this. You can approach it as a way to do the things you were doing before more efficiently, faster. I don’t know if more cheaply, as is becoming evident, but do, let’s say, more with fewer people, right? Or another possible approach you could take is you could take this as an opportunity to do more with the people that you have, or maybe even with more people in your team, right? Just empower them to explore possibilities that were previously unattainable because of the natural constraints of a team of humans and their limited cognitive abilities and attention. And I wanted to bring that up because, A, I think that we, well, I’d love to hear your take, but I think that we’re both in the second camp. It’s like, hey, let’s see how we can augment people to do more. Because one of the things that motivates us in our shared practice is possibilities, like opening a broader space of possibilities, right?

Greg: Yes, I think with possibilities, intention, like what future do we want to have and how do we help teams and organizations fulfill that in an intentional way? And I love that you brought that up because I think the conversation that Cindy and I had was really about two stories. I’ve been doing this fractional work, and over the last year I’ve worked with two different companies, and sort of not at the same time, too. So one was sort of last year to the end of last year, and the second one is I’m currently engaged with and I started with in January when all these tools started to mature. And there are sort of two conversations. The first one was sort of dabbling with AI, learning how to use the tools, some experiments with vibe coding, etc. The second one was the design team and the product team not only using those tools, but exploring whole new territories that we just didn’t have time for in the past to do. So we were imagining a new product capability, a new product category for a company, not a pivot, but an expansion of this particular organization’s remit that they want to serve. They had a really good idea of a set of problems they wanted to solve. And we just did things that were just hard to argue for in the normal timeframe that software development gets built. And we were able to do them. So we did a bunch of work on information architecture. And one of the things that we did is we looked at competitive tools that were out there, other people trying to solve the problem in similar ways, and we did really deep analysis of how they structured their experiences so we could try to understand why they were doing it. And that helped inform a better architecture for us. We did a bunch of work looking at agentic systems and emerging conversations around how to build MCP apps. And that was exploratory. Frankly, I think it’s exploratory for everyone who’s working in that area. And so we were able to help use the process of making to define the product requirements. And in some ways, we were leading product development in that conversation because we were really trying to understand meaningful outcomes. And we came up with this thing we called the job brief, and it worked for us very well. So I think the opportunity right now is to build much better products. The concept of an MVP to me is sort of like something that needs to be reexamined because most MVPs were always insufficiently great because they were constrained by time and engineering resources. And now, where engineering resources are less constrained, why do we put product into the market that isn’t foundationally awesome? And so we can do that. And I think there’s an opportunity to really ask the organization to rethink how it takes the time of their customers and fulfills their needs and their goals. And we can be much more clear and more polished and build things which on launch make people feel seen, the users of them feel seen in a way that we couldn’t do before. And so I think that to me is an inkling of what these tools should be doing. And that’s just a product development example. But I think you could do this for organizational change. And I think any organization could take a look at it and say, I don’t just want productivity. I want to unleash the creativity of my people towards new outcomes. And that’s the kind of work I want to work on. I think it’s the kind of work that you want to work on. And it’s the kind of work that I think that the Pope’s encyclical talks about, which is meaningful work and placing humans in the system and not just this sort of technocratic cost-out perspective.

Jorge: Yeah. There’s been a, well, you wrote a post about this and I want to get into the post.

Greg: Yeah, sure.

Jorge: The very title of the post has the phrase, “The feature is dead.”

Greg: Yeah.

Jorge: Which is kind of provocative, right?

Greg: Yeah, sure.

Jorge: Why is the product feature dead, Greg?

Greg: Well, I think we’re designing something different now, right? Feature was a component, a widget, a tool that helped humans solve a problem. And in the traditional way of product development and user experience, the actor was the user, right? The user’s on a journey. You build a capability for them to complete a task that’s important for them. They use that tool. They get it done. And so product development would think in increments of capability, features. That’s what we call them in product. And we would build those features in support of customers accomplishing their goals that were important to them on their adventures at work. And what’s changed now is we have this intelligence that we’re working with. And it can be the actor, right? It can make the choices and decisions on the process, especially when you think about more agentic workflows. And so the notion of product discovery and creating and making is different. We’re not making a widget or a tool. I mean, those are still important, and by the way, yes, thank you for calling out it’s a provocative title. Features don’t disappear; they still are important. But I think what’s more important is encoding the intelligence in a system so that it can support humans and their goals, but also allow some autonomy for agents to do things. And that’s a new skill, a new pattern. And the way I was thinking about it is Clayton Christensen coined this term jobs to be done. And jobs to be done is sort of evergreen. It doesn’t go away. You have to balance your checking account or you have to pay, make payroll, or you have to answer a customer call. But the way that you can do that can change over time with the way technology shows up or how an organization chooses to solve that problem. And ironically, a lot of organizations don’t actually really understand what they do. They’ve built sort of Rube Goldberg machines of technology that allow them to accomplish tasks, and individuals in the organization are actors in that to solve a business problem. But very few people in the organization actually understand the outcomes and business goals that an organization is trying to solve for. So I’ll give you an example. The reason why this came up is there’s also something very different in the way that we can build software right now, where it’s more structural and more of a system and less bespoke custom feature development. So this particular startup that I’m working with, we are building a system that can add new capabilities, and we wanted those new capabilities to show up almost like a feed of content, right? The application would be much more like Spotify, where you’d have a playlist of things that you can, instead of a playlist of things you listen to, a playlist of things that you can do. But to do that, then you have to kind of understand, like, what do those things do and why are they valuable and how do they achieve outcomes that are important for the end user or the company that’s buying a product or a service. And so we built this thing. I’m calling it a job brief. It’s a little bit of a hybrid. It’s influenced a little bit from the Josh Seiden sort of outcomes over output construct, being an outcome-centric notion. But what I wanted to do is try to create a recipe. And I built this sort of recipe card. You can’t really see it here, but I’ll share this with folks if they’re interested. And it was really very straightforward: find the job. What is the problem? Sit with a practitioner or a person at work and find out what is the thing that they need to accomplish. Write a job statement, one sentence: who’s the user, what are they trying to do, whether they want to know or decide. Don’t make it about technology at all. What’s the outcome that they’re trying to get to? And then the next layer, which is new, I mean, this is something we’ve done forever, but the next layer is what is the context that we understand in the system? What data do we have access to? Where is the user or the person who’s acting in a moment in their day, in their workflow, in the ebb and flow of an organization? You have to define what done means. When the job is completed, what does success look like? And then the last part of it is really understanding what expertise is around that. And working with people who have deep knowledge of what that outcome should look like, what’s meaningful for an organization. And I would argue that for every organization, that expertise might be different based on the values of that organization, the people they have, the things that they find important. And this is an opportunity to add that human layer into what differentiates company A from company B. Don’t just use a novel solution that everyone uses. Build something that’s respectful of what’s important to you as an organization. And then finally, the last part is find some way to measure it by how often it’s used, by invocation, like how often it gets done. And so that construct is really about understanding how humans in the system work. Now, I’m not naive. Some of this work turns into agents that do the work for you. And ideally, in an organization, you could use this as a way to look at the things in your organization that you can codify, the domain expertise, to gain productivity and to gain efficiency where you can. But my next point of view on that is in service of being able to do more interesting, more valuable work that pushes your objectives farther forward. And so that’s the notion behind it.

Jorge: This is all in the context of design and the projects that you have been working on. You’ve been working, your role has been like a fractional chief design officer, right? So you’ve been leading design teams. Another one of the signals that we are picking up from the environment is that it feels like many organizations are questioning the value of investing in design.

Greg: Yeah.

Jorge: And I was thinking, there was a time when a lot of organizations did not have big design teams internally, right? The idea that design is a function of the organization, I would say at least the current wave, because there have been prior ones, but the current wave, we can probably date back to around 20 years ago after it had become clear that Steve Jobs had saved Apple through design, right? And people were upholding the iPod as an example of a well-designed product. And there are still business leaders out there who are thinking in terms of, like, I want something that is as useful, beautiful, whatever, as the iPod, right? So it became like a touchstone that articulated the value of design for organizations. And that seems to have shifted significantly as a result of AI. And I suspect that it might not be coincidental that Jony Ive and company’s latest product to hit the market has not been well received by a lot of people. I’m talking about the new Ferrari.

Greg: Yeah, right.

Jorge: And I have no, you know, I’m not an expert in that space. I’m not a Ferrari fan or anything. To me, it looked like—

Greg: I lost your audio. Oh, no. I can’t hear you. Now I can.

Jorge: Okay, great. What I was saying: I am not a Ferrari fan, so I don’t have as strong feelings about that product as other people. But it did strike me that at least a lot of the commentary I saw online, I got the sense that it felt like this, like it was over-designed or like it’s precious. Or it’s like it’s a product that is trying to make design, like it’s a design-forward thing. And in some ways, that almost hurt it because it feels like it’s a rethinking of what a Ferrari is supposed to be. And I almost found it to be like a metaphor for what’s happening elsewhere in design. It’s like, I guess the question is: is design necessary now that there’s a Claude design and Figma will do design work? And I think that I’ll stop after this. I think that a lot of organizations have built design functions that are primarily production functions.

Greg: Yeah.

Jorge: And by production, I mean they are set up basically to crank out screen-based experiences.

Greg: Yeah.

Jorge: And they have staffed up with people who are tasked with doing that kind of work. And at least to me, it’s been pretty clear that that’s going away, and it has been clear for a while, right? The tools have been getting better. The kind of design work that you’re talking about when you talk about job briefs. You said that the jobs-to-be-done construct is evergreen. And I think that one of the reasons it’s evergreen is because it moves the level of abstraction. It’s not about how do you execute on the hole in the wall. It’s about how do you determine what it is that the customer needs. And then you can figure out how that’s delivered to them, right? And I’m wondering, I’m just, again, this is open-ended and unstructured, right? So I’m thinking out loud here. I’m wondering if what we’re seeing is a shift from an understanding of design as this kind of screen-level production function to, and then there’s a question mark. And I think that your article tries to answer that question by saying to a role that is more strategic, which is something that designers have been trying to do for a while with not a lot of traction, I would say.

Greg: Yeah, many haven’t. Yeah, there have been some examples of people who have, but yes, you’re right. Let me respond to it. I think when we build products, there are different roles that are really important. And I think when design is seen as only craft and only as screen building, it misses the value that design can bring to the table. Those things are important, but they’re not the only thing. And the challenge is that at some level for many people in many organizations, the AI tooling is capable of doing, I wouldn’t say excellent, but reasonable results on the sort of production-level quality of building a user interface. However, I think one of the things that designers bring to the table is a deep understanding of human behavior and building products that recognize humans in the system and the mental models that people have. And it’s connected to empathy. It’s connected to having the ability to do ethnography, basically sit with customers and understand how they solve problems and what they’re trying to accomplish. It requires some new skills now because when we look at the capabilities of these tools, you can incrementally make things better or you could radically change things. And it doesn’t mean one or the other is the right path. It’s actually what is the right path for the group of humans that you’re responsible for delivering a solution for. Too much change may require a group of people to do things they’re just not unwilling to do or unable to do cognitively. Not enough change is a missed opportunity toward functionally changing the game in a way that produces some really interesting results for an organization or for the people who work in the organization. And we’re in this really messy period where the old rules don’t really apply and people are trying to apply the old rules. So you still see PRD, product requirement documents. You still see engineering burndowns. You still see JIRA tickets created. You still see, this is a software development process, because people anchor to that. That’s what they know. They kind of organize around it. But it’s unclear to me if that process of how we built software is still usable and useful in this new moment. And there’s an opportunity to change that. So where am I going with all this? I think designers bring a couple of skills to a team that are really valuable. And if not there, it’s a missed opportunity for organizations. And they can be very strategic. We are adept at the art of juxtaposition. We can take two different ideas and put them together and discern a net-new outcome. Because we make the think, we make things, we make artifacts, and those artifacts inform us. And that conversation we have with the things that we make allows us to find truth or an answer or a solution or a novel way of solving a problem in a way that can be faster than other methods. It’s not the only way to solve those problems, but it’s a super valuable way to do it. So that’s kind of just one path that designers take. The second part of it is I think it is my personal perspective that it’s inexcusable that you don’t make things great now because the tools let us do that. And that requires discernment. That requires someone who can look at a small radius or a small curve on a user experience or looking at accessibility and making sure that a product is useful for all of us, regardless of our abilities. That we can choose words carefully so that it aids people in the direction of the path that they’re trying to take. We don’t just build a piece of software and expect them to learn how to use it. We can now frame a product in a way that fits the mental model of the people that we’re serving, and we can actually get closer and closer and closer to that because these tools allow us to actually iterate and understand that more successfully. So I guess a long way of saying is I think designers are very important. There are some new things that we have to look out for, like I think interfaces are starting to collapse. We’re doing more of our work in language models, right? We’re speaking with AI. We’re having a conversation with AI. AI systems can not only answer words, they can answer in interfaces, so they can create custom experiences for us or custom tools for us in the conversation. You could just do that. But I think the thing that design brings to the table is an intentional way of doing it, shaping the grammar of an organization, inserting a value set of values into an organization, shaping how answers are delivered. Are they long? Are they short? What’s the structure? What’s the organization of them? Left alone, a large language model may choose to answer a problem differently every time you ask it, and that may not serve the audience that you’re trying to serve, right? So I think there’s a role for design to be very much involved in intentional curation of these experiences and bringing human values into them explicitly. I remember, I’m forgetting who told me this. I’ll think of it in a moment. But all software has opinions in it, whether they are explicitly or implicitly embedded in the software. Some teams are very explicit about it: this is what we do, certain things. Others, it’s just the end result of the people who built it and what they valued. But it’s there, right? We have an opportunity to be very explicit about how we serve people. And again, I think design is the discipline that not just solves business problems but figures out ways to make it deeply human. And I think our opportunity is to help organizations at a different layer, at a different level. And it may not be screens we’re working on. It could be orchestrating the flow of how work happens in a way that makes sense for people. And that’s a new skill. Not all designers are going to pivot or understand how to move into that space. But I think the things that we bring to the table are very useful there. Anyway, that—

Jorge: Was a long answer to your question. Well, and it kind of prompts some follow-up questions. You said earlier in the conversation, and I’m going to be—

Greg: I lost you there for a second there, Jorge. I think your internet took a—

Jorge: Yeah, something’s glitchy. Can you hear me?

Greg: I can hear you now, yes. I wanted to repeat your question or insight.

Jorge: Yeah, so you said, I’m going to paraphrase and probably get it wrong, but you said something earlier in the conversation along the lines of one of the things that we’re grappling with, and this was talking about as designers, is the fact that many of the things that we were doing as humans working with tools are now being delegated to agentic systems, right? And I’m going to put on my CEO hat, right? Like I have to make, I have to determine how I’m going to invest, how I’m going to allocate the organization’s capital, right? And there’s a lot of incentive right now for organizations to invest in capability, like technical capability, right? Compute is what they’re calling it, right? And you said that one of the things that designers bring to the table is a deep understanding of human behavior. Yeah. But if the systems that we’re building are not going to be used by humans, why does that matter? If I’m a CEO looking to invest, what’s the argument in favor of investing in a design team that is going to be crafting experiences to be used by humans? When, hey, isn’t AI going to do all of it? Why am I investing in human experiences?

Greg: Well, I mean, that’s a great question. I think if you’re a cost-out CEO, that may be what you think about, right? You’re like, I don’t need a design team. We’re going to use all these tools, and we’re going to deliver this service with as few people as possible. And we’re going to look at generating as much revenue as possible. And this is a path for us to get there. And my guess is there’ll be many parts of our economy that are going to be efficiency plays like that, that are going to be things that people are worried about from a job perspective. And it’s sort of inevitable. However, I also think that there will be a group of CEOs that have a growth mindset who look at what their organization does in their community or wherever they serve and look for opportunities to provide better services or better outcomes or better products. And I think one of the things that’s missed is there’s this bottom-up and top-down conversation that’s missed. So as an example, I think small organizations benefit massively from these tools because they allow them to punch way above their weight so they can compete at levels that they couldn’t compete with before. And small, intimate teams with these tools that communicate well with each other now have the ability to expand their horizons around the things that they can offer, the services they can develop, the things that they can provide for whoever their customers are, or users are, or whatever their goals are, whether they’re for-profit, not-for-profit, etc. In large organizations, I think the transition is going to be identifying also where in the aspect of the services that they provide that humans are actually important for their customers, right? So I think as humans, people want to hang out with people, right? And I mean, there are people apparently who have chat, cheap, and girlfriends, but I think most of us are, especially even post-pandemic, and I still think we’re in the post-pandemic phase, we want to have a sense of community and connection to each other. And so I think there’s huge opportunity for organizations that recognize how to place their people front and center with their customers or the people that they care about or the things that are important. So I guess what I’m trying to discern is I think there’s an opportunity space for the people who are good at unlocking possibilities and discerning new things to service, developing better outcomes for all of us. And it’s a little bit of a, you know, it needs to be seen, but I think that’s the path that I want to see. I would love to see our politics in this world focus on the opportunity to service and support everyone. I’d love to see our businesses look for opportunities to grow their businesses in new and creative ways. And I would love to see human potential be the big story about being unlocked by these tools versus being laid off. And I think we need to think about that very hard. I think we need to find this is going to be a transition. There’s not going to be without, there already is massive transformation going on in organizations. And I think we are, as a culture, need to be managing this carefully. And again, I’ll come back to why I think design is important, because I think designers can find opportunities to potentially mitigate some of the challenges with AI, from an employment and cultural perspective. But they’re also just going to be great at finding new things to do that we didn’t know we needed before and that are exciting and fun and fulfilling and helpful. And so, if I were a large CEO, one of the things I would be investing in is a creative design team that is just exploring and looking for net-new things that could benefit, that are adjacencies to a company’s core mission that might be new growth opportunities for them. And I think that the tooling allows creatives the opportunity to really shine there. I don’t think we’re there yet. I don’t think people have talked about it yet enough. But I do think that there’s really an opportunity for us to look for ways to be more future-focused and actually try to tackle some of the real problems that we have in the world.

Jorge: I kind of want to double down on what you’re saying there, because in playing devil’s advocate earlier, it may not have been clear just how much I agree with what you’re saying there. But my read is that organizations that are kind of restructuring themselves to maybe, like a phrase that you could, maybe a phrase we could use is something like a post-design world, or like a world in which they’re not placing their emphasis on human experience, right? The organizations that are investing in AI as a way to gain efficiencies by automating the sort of touchpoints that humans have relied on so far, they’re doing so from what I see as like a, like in my mind, I have this Venn diagram that has fear, bad incentives, and a kind of like spectacular lack of imagination. Yeah. The phrase that often comes to mind is this Warren Buffett thing where he says that he advises being bold when others are fearful and fearful when others are being bold. And I just want to kind of double down on what you said about this being a time of unique opportunity, particularly for the organizations that are willing to zag where everyone else is zigging in the direction of efficiencies. Just because there are, like to your point, there are now possibilities that were just previously unavailable. And to only think about the path of, like, how can we make, how can we deliver the minimum possible experience as cheaply as possible is one possible direction, but by no means the only one, right? And I suspect that we are still in the throes of, like, hey, this is a new technology and let’s double down on efficiencies where we have not yet truly explored the possibilities of other ways of using the technology, right?

Greg: Yeah, I mean, I love that framing. I’ve been a big, one of the things I’ve talked about for the last 10 years is this notion of incrementalism versus really understanding the problem, right? And for a lot of organizations, incrementalism was the path towards success because you took small pieces and you got better and you got better over time and you got better over time. It kind of fits the Agile Manifesto. It’s how organizations work. They’re risk-averse, so they don’t want to try to take on too much. We’ll make one small change, we’ll make a change of, you know, and. But it was connected to the fact that things were very difficult to do well, and therefore you had to be careful about doing it. And so instead of going for it and doing something big or trying an experiment and failing, you just made your product slightly better over time or your outcome slightly better over time, your system slightly better over time. And there’s nothing wrong with that as a strategy, by the way. It makes sense and allows you to do things in a sustained way. What is interesting about this moment is that these tools allow us to build things that used to be very, the very expensive part of building software is far less expensive than it used to be. And so an incrementalist mindset may just be a faster way to get to the wrong place, you know? And I think one of the things that we don’t do enough of is discovery work, understanding what is the problem and what would really be meaningful for people. And now we can spend a lot of time in that, right? And in the design space there’s this diagram called the double diamond that came out of the British Design Council that’s a very popular way of describing the process of design. And the first half, the first diamond, is discovery. And then you sort of evolve that into a product, and then you go into an execution phase, and then you deliver and you learn from your customers. And my argument now is that these tools allow us to build the first part of that diamond much bigger and the rest of it much smaller. And what do I mean by that? It means that we can spend, in a short period of time with these tools, we can learn way more about who we’re serving and have a much better understanding of what their problem is and iterate on not just one, but hundreds of solutions that could possibly satisfy those needs and outcomes, and then discern which one is the best, refine it, make it great, and then deliver it at fairly low cost from an execution standpoint. And so that upends the process. It used to be the other way around, that the execution part was so hard that that’s where all the energy was, right? And so it was like, make a small bet, get it out to market, make another small bet, make it out to market. So it goes back to your point of lack of imagination. And I think there’s an opportunity right now for companies to at least have some part of their organization that is thinking boldly and broadly and unhampered by an incrementalist mindset. And I think if I were a CEO of a larger organization, I would be setting up a lab to think about not how to use these tools to do what I do better now. I would be setting up a lab on how do I use these tools to do something I’m not doing now that would help me grow my outcomes that are important to my shareholders or to my customers or to the business community that I’m trying to connect to.

Jorge: I love that framing.

Greg: And I think designers are great at that. They’re not the only one. There need to be other parties in that conversation. It’s a multidisciplinary effort, but it doesn’t work without designers in the room.

Jorge: You know, to circle back to the Pope’s encyclical, the central metaphor that he uses to talk about the possible ways forward for the development of AI-based systems, he uses this architectural analogy from using two stories from the Bible. One story is the construction of the Tower of Babel, which is a kind of technocratic, top-down effort to, now I’m kind of reading into it, right, to control. And it’s an effort that kind of flattens differences between people. And he contrasts that with another story from the Bible, which is the rebuilding of Jerusalem after it had been torn down. And he talks about Nehemiah. I don’t know if that’s how you pronounce it, but this person who rather than dictate top-down this technocratic solution, gathered the people who inhabited the city. And through this kind of consultative approach, led to the rebuilding of this kind of organic, more organic city that responded to their needs. And part of what I’m hearing you say is that, and I think that this is also implicit, if not explicit, in the encyclical, is that AI allows for both of those approaches. We can do the top-down thing really fast now and really comprehensively, and it can turn into a real dystopia. Or we could employ it in this more kind of bottom-up, consultative, human-centered way, and it could also do it much faster with greater scope, with the possibility to explore many, many more alternatives just because the tool allows us to do so much more. So it becomes a matter of how do you choose which of the two approaches you’re going to take? And I think what we’re saying here is we would like to foster a world that follows the second path, the more kind of designerly path, the path that puts human beings, their needs, including their dignity, which is an important word in the encyclical, front and center. And the tools are amazing. They can empower us to do much more than before, as long as it’s in service of that, right?

Greg: Yeah, creating a better world. And the irony is you can have both of those things happen. The efficiency play can be in service of unlocking human potential, right? So they don’t have to be either-or paths, but they do have to be both at a minimum. If we just go the Tower of Babel path, I don’t want to live in that future. I’m very much interested in being intentional about the choices that we make. And again, that’s why I go back to why designers are important. We’re good at that. We are good at intentionally helping craft a narrative and artifacts and things which connect to the way that we want to live. And then we’re good at telling stories around that with the things that we create. Those things that we create inspire people. And it’s a really part of being human that there are things that delight us and bring us joy and make us cry and help us understand and live fulfilling lives, all the things that I think are important, which by the way, that’s not the longest list of important things, but nonetheless, I think we’re coming to a close.

Jorge: I was going to say that feels like a good place to wrap it up. If you want to follow our work, we do have a Substack. It’s called Unfinishe Thoughts, and it’s at thoughts.unfinishe.com. Remember, it’s Unfinishe without the D. And Greg and I post there periodically. It feels like almost as infrequently as we do these live streams, but we should write these things up more.

Greg: Yeah, our plan is to be a little bit more prolific in the rest of this year. But thanks for hanging out with us today. And Jorge, as always, I love hanging out with you and talking about our stuff. And we’ll see you at the next Unfinish.

Jorge: Same here. Thanks, Greg. Bye.

Many teams are being measured for the wrong things: tokens used, agents deployed, etc. Their orgs have focused on tech adoption rather than value creation. It’s a mistake.

I wanted to discuss this with Harry in our most recent podcast, so I read a passage from one of my favorite books, James C. Scott’s Seeing Like a State. To my surprise, he’d read it too.

I won’t cite the whole passage, but it kicks off with a familiar distinction:

Isaiah Berlin, in his study of Tolstoy, compared the hedgehog, who knew “one big thing,” to the fox, who knew many things. The scientific forester and the cadastral official are like the hedgehog. The sharply focused interest of the scientific foresters in commercial lumber and that of the cadastral officials in land revenue constrain them to finding clear-cut answers to one question. The naturalist and the farmer, on the other hand, are like the fox. They know a great many things about forests and cultivable land. Although the forester’s and cadastral official’s range of knowledge is far narrower, we should not forget that their knowledge is systematic and synoptic, allowing them to see and understand things a fox would not grasp.

Scott then unpacks how flattening an ecosystem to a few legible variables leads to a kind of myopia. This is assuming the variables are meaningful, as with land productivity for cadastral purposes. Token maxxing, on the other hand, is folly.

Legibility — instrumenting processes so we can track progress — is essential for traction. But we shouldn’t focus on things we can measure (e.g., tokens used, numbers of agents created) rather than those that matter to the business.

It’s harder to focus on the right measures when we’re acting urgently and/or from fear, as is the case for many teams now. How can we measure what really matters? That’s what Harry and I explore in this episode.

Traction Heroes episode 37: Legibility

The Pope’s namesake, Leo XIII, led the Church during the second half of the 19th century. It was a time of social unrest triggered by technological changes: the Industrial Revolution precipitated the exploitation of workers and its counteraction in Marxism. Both dehumanized societies.

The Church’s response was Rerum Novarum (1891), an encyclical that promoted both the rights of workers and free markets. It offered a sensible middle path between anything-goes capitalism and atheistic socialism, and set the foundations for the Catholic Church’s modern social positions.

In choosing the name Leo XIV, the current Pope hinted at continuing this work. It’s certainly needed: digital technologies — and AI in particular — are at least as transformative as industrial machinery. The central question is, how will we use this technology? Will it serve the common good or dominate and exploit people?

These aren’t technical questions: they’re moral and spiritual. The Catholic Church has grappled with such issues for centuries, and Magnifica Humanitas offers an overview of that history, warts and all. The outcome is a comprehensive social framework based on several key principles:

• The common good. Work toward social well-being collectively, not by aggregating individual interests.
• The universal destination of goods. Don’t monopolize wealth and resources; steward the goods that create well-being for all.
• Subsidiarity. Delegate decision-making authority to the smallest possible unit.
• Solidarity. Recognize that our destinies are bound together and we have obligations toward each other.
• Social justice. Be fair and don’t exploit or discriminate against people. (Again, everyone has inherent worth.)
• Integral human development. Elevate people so they can participate with dignity in society.

After recapping the Church’s Social Doctrine, the encyclical applies these principles to AI and the digital economy. We can choose how to design these technologies. The Pope describes two possible approaches by using architectural metaphors drawn from scripture. One is that of the Tower of Babel: a construct meant to glorify its dominator-builders and homogenize people for the sake of power and efficiency. The other is exemplified by the lesser-known story of the rebuilding of the walls of Jerusalem:

Nehemiah, a Jew in the service of the Persian King Artaxerxes, received news of the disastrous state of his ancestral city. Before taking action, he fasted, prayed and interceded for the people. He then asked the king for permission to return to Jerusalem and, upon arriving, examined the destroyed areas in silence. He did not impose solutions from above. He convened the families, assigned each of them a section of the wall to rebuild, listened to their concerns, coordinated their efforts and addressed any opposition.

Which is to say, an organic, consultative process with the people affected — as opposed to a top-down technocratic initiative. The former responds to the needs of members of the community. The latter imposes “solutions” from above, eroding their agency. Upholding human dignity is the encyclical’s foundational principle, and it’s under threat:

It is important to ensure that this growth in appreciation of human dignity is not obscured by the pressure of new ideologies or very powerful interests in today’s world. Among these ideologies, I consider particularly insidious the one that suggests that every person must earn or justify his or her own worth, to the point of attributing greater value to those who are more efficient or effective. From this perspective, persons end up being reduced to a means of achieving results, a resource to be used and exploited, and are no longer recognized as a proper end in themselves who should never be instrumentalized. The value of persons, however, does not depend on what they achieve or produce. There are rights that apply to everyone simply by virtue of being human, and no human power can legitimately deny or arbitrarily limit them.

Like all technologies, digital systems are shaped by the values of the socioeconomic systems that create them. Systems that prioritize domination and personal gain over the common good pose serious risks, including growing inequality, weakened democracy, increased social unrest, environmental degradation, the emergence of new forms of slavery, and more.

And of course, there’s the ever-present risk of violence. Autonomous weapons threaten new levels of destruction. War takes a new valence when the systems deciding on the path of death aren’t human:

Artificial intelligences do not undergo experiences, do not possess a body, do not feel joy or pain, do not mature through relationships and do not know from within what love, work, friendship or responsibility mean. Nor do they have a moral conscience, since they do not judge good and evil, grasp the ultimate meaning of situations, or bear responsibility for consequences. They may imitate language, behavior and analytical skills, or even simulate empathy and understanding, but they do not understand what they produce, for they lack the affective, relational and spiritual perspective through which human beings grow in wisdom.

But Magnifica Humanitas is less an AI manifesto than a commentary on (and corrective of) the socioeconomic systems under which it’s emerged. Instead of systems that treat people as means to particular ends (e.g., Marxism, extreme forms of capitalism), the Pope calls for building a civilization of love grounded on justice, human rights, and basic decency.

In issuing this encyclical, the Church stakes a clear moral position. It isn’t an argument against technology, private property, or free markets. Instead, it’s a reasonable call to treat humans and our labor with dignity:

In the era of artificial intelligence, when human dignity is threatened by new forms of dehumanization, ours is the pressing duty to remain profoundly human.

You needn’t be Catholic to recognize the importance of this message. Ultimately, the question of how we’ll put AI to use isn’t technical, but moral. The stakes — human dignity, democratic participation, the distribution of power — couldn’t be higher.

We do that a lot, lie to ourselves.

I don’t like making predictions, but I’ll make one now. I’m confident it’ll pan out: organizations cutting staff “because AI” will come to regret that choice.

That’s not to say I believe that’s why they’re actually laying people off. Many orgs were likely overstaffed. AI is just the excuse leaders are giving Wall Street. Perhaps some believe it. If so, they’re lying to themselves. And there will be consequences.

As the class “fat kid,” I was bullied. It affected my personality. I became simultaneously shy and viciously sardonic. My grades suffered. I’m still paying the price for the fridge “easy fix” decades ago. Near-term pleasure, long-term suffering.

Orgs cutting staff are being rewarded with a near-term financial boost. What might be some long-term consequences?

I see at least three:

1. Reduced trust. The people who remain still have work to do. That includes deploying AI, a technology they now viscerally perceive as a threat. They’re damned if they do and damned if they don’t. And their leaders have made clear what they value.

2. Brand erosion. Wall Street isn’t the only audience: customers are listening too. And they’re likely (and rightly) concerned that the org seeks to replace human judgment. We’ve been burned before: that’s why the word ‘enshittification’ exists.

3. Loss of culture and knowledge. Much of how organizations produce value isn’t formally codified. In many ways, an org is its people — their shared culture and knowledge, most of which is tacit. The data AI needs about how orgs tick doesn’t exist.

Humans aren’t mere resources. Organizations are complex adaptive systems where people play critical roles that extend well beyond their job descriptions. For many orgs, interpersonal relations are the golden egg-laying goose.

Current AI can’t bridge these gaps. Leaders who believe it can are either deluded about its capabilities, the true nature and complexity of the business, or both. Or they’re lying to themselves.

My mother was right. You may think you’re cleverly misdirecting others, but you’re only misdirecting yourself. And the consequences will haunt you.

I’ve been pretty glum on this note, but that’s because I don’t like what I’m seeing. So I’ll leave you with the positive flipside of my prediction: The organizations that thrive in the AI era will be those who augment and empower their people — not those who cut them.

The technology is capable, but it will require lots of work. The challenge is architecture, not headcount.

You want to make better decisions to act more skillfully. But how? Seeing clearly is essential. It’s hard enough to get a good read on your context, but getting a good read on yourself is even harder. When it comes to self-assessing, we all have blind spots.

In episode 36 of Traction Heroes, Harry brought a short reading that explores this idea. It’s from Phyl Terry’s Never Search Alone, a book that has helped people I know with their job search. But our conversation didn’t focus on job hunting. Instead, we explored what Terry calls Kahneman’s conundrum, after celebrated psychologist Daniel Kahneman:

Kahneman says that he himself has no idea when he’s making any of the cognitive mistakes that he spent a lifetime identifying: ‘my intuitive thinking is just as prone to overconfidence, extreme predictions, and the planning fallacy as it was before I made the study of these issues.’

Whoops! Even the most prominent researcher on cognitive biases can’t effectively self-assess them. Terry continues,

Might there be a way for leaders to at least know when they’re about to make a mistake based on a bias? Kahneman says no. We would all like to have a warning bell that rings loudly whenever we’re about to make a serious error, but no such bell is available.

Bottom line:

The knowledge of these biases does not seem to matter. The expert knowledge gained from decades of research does not even help the co-founder of the discipline. We humans are destined to make the same kinds of mistakes repeatedly, to be blind about our blind spots. It is simple but true. We all know that we can easily see the errors in others while being frustratingly blind to our own.

A conundrum indeed! What can we do about it? I suggested getting others’ perspective, whether a trusted colleague, friend, or mentor. Harry offered the Johari window as a tool for understanding what others think about us that we don’t know ourselves.

Continuing on the practical vein, I mentioned my experiment with using AI to illuminate my blind spots. Some folks pushed back on this as a sort of modern-day astrology (i.e., the language invites us to read into it), but I found ChatGPT’s ‘reading’ insightful.

However you do it, include others’ candid feedback into your decision-making process. Self-awareness alone won’t cut it.

Traction Heroes episode 36: Blind Spots

Still, mindful leaders are encouraging their teams to embrace AI. Experiments proliferate. Overnight, teams find themselves grappling with dozens of “agents” that cover similar ground. There’s more competition than collaboration among team members.

There’s little coherence. Most of these efforts don’t solve strategic business or customer problems. At best, they’re compelling — yet ad-hoc — proofs-of-concept.

Worse, they’re emerging in a context of generalized anxiety. Team members are trapped in a “damned if they do/damned if they don’t” conundrum: Their jobs are at risk if they don’t adopt AI, but media constantly reminds them AI might replace their jobs.

The anxiety is understandable, but also a bit sad. There’s so much potential. AI is by far the most exciting tech development I’ve experienced in my three-decade career.

Even so, many people have bad expectations of what the technology can deliver now. These things take time. It took several years — and expensive mistakes — for businesses to learn where and how the web could be used most effectively.

It’s too early to expect efficiencies from AI. The goal right now should be learning, not optimizing.

The best we can do is run directed experiments: fast, small, iterative projects that explicitly aim to move the business forward while developing essential new skills. Not just bottom-up, but building toward a directed vision.

How do you define that vision? You consider the big picture. What’s the organization’s strategy? How does the business unit serve that strategy? What are its key information flows? Where are the bottlenecks? Which can be best addressed using AI?

Throwing agents against the wall won’t answer these questions. Real progress requires top-down direction and visibility: understanding the big picture well enough to determine how AI might best unlock new possibilities. But it also requires experimenting to learn how the technology can actually work within your particular context.

These things aren’t in tension. A mindful balance is called for.

Ultimately, it’s an architectural problem. The organizations that benefit most from AI won’t be the ones that burn the most cycles and tokens. Instead, it’ll be those who understand the big picture well enough to drive advantage by architecting intelligence.

When I was a kid, I tended to view the world in stark black-or-white terms. Things were either amazing or terrible. People were brilliant or bozos. My dad would caution me: right answers often land closer to the middle of a spectrum than its extremes.

Later in life, I read an essay called Axis Thinking, which organized these ideas into a coherent framework. Written by the musician, producer, and systems polymath Brian Eno, it appeared in his book A Year with Swollen Appendices.

Axis Thinking has influenced how I consult. I wanted to hear Harry’s take, so I brought it as the focused reading on episode 35 of Traction Heroes. How has it changed my work, specifically? Eno put it well:

Axial thinking doesn’t deny that it could be this or that, but suggests that it’s more likely to be somewhere between the two. As soon as that suggestion is in the air, it triggers an imaginative process, an attempt to locate and conceptualize the newly acknowledged greyscale positions.

I am interested in these transitions — these moments when a stable duality dissolves into a proliferating and unstable sea of hybrids. What happens at such times is that all sorts of things become possible: there is a tremendous energy release, a great burst of experimentation. Not only do the emerging possible positions on this new-born axis have to be discover­ ed and experienced and articulated; they have to be placed in context with other existing axes to see what new resonances appear.

That is, it’s not enough to look for solutions in the middle points of a spectrum. Sometimes, what’s called for is questioning the spectrum altogether. One or both poles might not go far enough — or go too far. And exploring completely different axes might prove fruitful.

Axial thinking is highly relevant today. See, for example, the barren discussions that frame AI as either all good or all bad. Most realistic scenarios fall somewhere in the middle — but opening an altogether different axis (e.g., replacement ↔ augmentation) can spur more generative discussions.

Gaining traction sometimes calls for dissolving stable dualities “into a proliferating and unstable sea of hybrids.” Nuance and imagination are more important than ever. This conversation offers pointers on how to move past entrenched polarities.

Traction Heroes episode 35: Axis Thinking

That’s where we’re at with AI. Companies are throwing stuff at LLMs to see what works. Right now, traction and speed matter more than efficiency. But that won’t always be the case. Eventually, we’ll shift to AI-powered systems that are both more efficient and controllable. What will they look like?

My bet: a combination of more carefully structured inputs (i.e., context engineering) and good, old-fashioned deterministic programming.

I like Simon Willison’s definition of agents: “An LLM agent runs tools in a loop to achieve a goal.” The most common way to do this is to give an agentic system (e.g., Claude Code) the ability to call deterministic tools — Unix CLI utilities, APIs, etc. It’s a powerful and flexible approach, but one that uses a lot of tokens and requires advanced models. It’s also unpredictable in ways that matter when you’re running a business.

But for many use cases, you can achieve similar results by inverting the hierarchy. Instead of giving an LLM the ability to use deterministic programs, you have a deterministic program call an LLM at specific points in its flow, using heavily curated context. It’s an information architecture challenge as much as a software challenge.

This isn’t as flashy as an autonomous agent improvising its way through a problem. But for many business tasks, it doesn’t need to be. The constrained approach produces more predictable outcomes, runs on smaller and cheaper models, and keeps your data under control.

The tradeoff: you must know what you’re building before you build it, since the bulk of the work happens in traditional software. But — and here’s the kicker — coding agents can help design and write that software. A bit of probabilistic work upfront to spin up a mostly-deterministic system.

I’m using such “agents” on my Mac. For example, one monitors a directory for new PDFs and fires a script that calls an LLM to transcribe my handwritten notes. About 80% of the work is classic if-then logic. The AI handles only the part that actually needs AI. Because those asks are tightly constrained, I use small models running locally — free to download, private by default, and using hardware and energy I’ve already accounted for.

This is what more mature AI adoption will look like for most organizations. Instead of open-ended agents improvising at scale, tightly-scoped systems that call on AI for the things AI does best. Heavy thinking upfront, with day-to-day operations using limited AI. Upshot: increased control, efficiency, and predictability.

Automation doesn’t just do work for us: it changes how we experience work, sometimes leading to unexpected consequences. Technology can make us complacent, leading us to make mistakes.

In episode 34 of Traction Heroes, Harry told of a time when he drifted off course by following GPS navigation instructions. Instead of the airport, he ended in a remote part of Virginia — and almost missed his flight. Things like this happen. Harry also cited a compelling story from Nicholas Carr’s The Glass Cage:

Automation complacency has been documented in many high-risk situations, from battlefields to industrial control rooms, to bridges of ships and submarines. In one classic case involving a 1,500-passenger ocean liner named The Royal Majesty, which in the spring of 1995 was sailing from Bermuda to Boston on the last leg of a one week cruise. The ship was outfitted with a state-of-the-art automated navigation system that used GPS signals to keep it on course. An hour into the voyage, the cable for the GPS antenna came loose, and the navigation system lost its bearings. It continued to give readings, but they were no longer accurate. For more than thirty hours, the ship slowly drifted off its appointed route. The captain and crew remained oblivious to the problem despite clear signs that the system had failed. At one point, a mate on watch was unable to spot an important location buoy that the ship was due to pass. He failed to report the fact. His trust in the navigation system was so complete, that he assumed the buoy was there and he just didn’t see it. Nearly twenty miles off course, the ship finally ran aground.

Like many technologies, GPS works well most of the time, so we don’t question it. But things can go wrong. Our trust in the technology’s capabilities can lead to a misplaced sense of security. The results can range from inconvenient to catastrophic.

This is obviously highly relevant in the age of AI. Not only is the technology fallible, but it also presents its outputs with high degree of self-confidence. So we must be especially vigilant. That said, always second-guessing results can cost valuable time and resources.

As I explained, we must use AI with a clear understanding of our own competence in the domain in question. We can be a bit less vigilant in domains where we have enough expertise to judge the quality of the output, but must be more skeptical when we don’t know what we don’t know.

Knowing where and when to apply critical thinking is key to avoiding setbacks. What’s required is literacy: understanding how the technology works under the hood. AI isn’t magical. If you understand how it can fail, you’ll be less likely to accept results uncritically — and when it’s ok to trust them.

Traction Heroes episode 34: Automation Complexity

Here’s a tricky situation: you start reading someone through a negative lens, which changes how you interact with them. They respond in kind, which seems to confirm your negative views. Cue vicious cycle.

In any situation, you are both observer and participant, whether you realize it or not. And often, you’re responding not just to the person in front of you, but to your story about them.

This mind-bending topic was the subject of episode 33 of Traction Heroes. Harry brought a reading from Nir Eyal’s Beyond Belief to set up the conversation. Here’s one of the key bits:

More often, it’s our brains creating problems because none exist. Since perception follows belief, we perceive the problems we look to find and if we can’t find them, our brain skews the data to fit the brief. If you believe your partner is constantly criticizing you, innocent comments transform into attacks. If you believe your boss doesn’t value you, any feedback becomes proof of your perceived inadequacy. This cycle becomes dangerous when it reinforces our negative beliefs, locking us into a belief-driven feedback loop that distorts reality and quietly builds a prison of our own making.

I’ve been there, and I’m sure you have too. You may have even unwittingly flipped someone’s “bozo bit,” leading to a strain in the relationship that can be hard to undo.

The question is: what can you do about it? As with so many other topics we’ve discussed in the podcast, it comes down to self-awareness: having the wherewithal to step back and realize you’re layering meaning onto situations.

Easier said than done! For one thing, you want to perceive clearly to avoid misreadings. But you don’t want to lapse into paranoia, which can also cast a negative valence.

Often, our misperceptions become obstacles to gaining traction. Surfacing them is a start, but we also explored practical suggestions in the podcast. Check it out:

Traction Heroes episode 33: Perceptions

Sure, AI can help you move faster. But are you moving in the right direction? How do you know? These are the key questions Jesse James Garrett, Peter Merholz, and I explored in episode 69 of their Finding Our Way podcast.

Leadership entails acting intelligently — i.e., moving in the right direction for the right reasons. This requires seeing clearly. Tools can help… or they can make it harder while seeming to help.

The question is, how do you do it? I’m a big fan of understanding the technology firsthand. But we must also understand how the technology changes the nature of the work.

AI calls for moving up the abstraction stack. It’s similar to what happened with computer programming, which went from flipping bits to assembly language to higher-level languages and now coding agents. The question before design and product leaders isn’t whether this shift will happen to design: it’s whether they’re ready to lead at the right level.

A bifurcation is coming. The organizations that figure out the role the technology plays in this shift will thrive. Those who do it poorly will crank out work faster — but it’ll be increasingly misaligned with the business’ needs.

As I said near the end of the episode, if you come out of any of these conversations feeling like you’ve got the answer, you’re probably wrong. The technology is changing too fast. What you can get is a clearer read on the context. Hopefully, this conversation helps.

Finding Our Way, Ep. 69: In a World of AI, What is the Work Really About? (ft. Jorge Arango)

I’m pleased to talk PKM in a city where Erasmus of Rotterdam — one of my intellectual heroes — spent some of his early years. This presentation has two purposes. First, I’ll give you a different frame to think about personal knowledge management. Then, I’ll share ways to use AI effectively in that context. I expect Erasmus would’ve been tickled!

Last year, Lou Rosenfeld sent me an email asking if I’d seen a post by Joan Westenberg titled I Deleted My Second Brain. Back in 2024, Lou published my book Duly Noted, which shows you why and how to build a PKM system. Westenberg’s post argued against doing just that. Naturally, Lou wondered what I thought.

I won’t recap the whole post. The TL;DR: the author built an elaborate PKM system. After some time, the system wasn’t producing the expected results, so they got rid of it. Rather than summarize further, I’ll cite a couple of representative passages:

In trying to remember everything, I outsourced the act of reflection. I didn’t revisit ideas. I didn’t interrogate them. I filed them away and trusted the structure. But a structure is not thinking. A tag is not an insight. And an idea not re-encountered might as well have never been had.

And:

When I first started using PKM tools, I believed I was solving a problem of forgetting. Later, I believed I was solving a problem of integration. Eventually, I realized I had created a new problem: deferral. The more my system grew, the more I deferred the work of thought to some future self who would sort, tag, distill, and extract the gold.

That self never arrived.

It’s a good and nuanced post, and you should read it. That said, it doesn’t cover new ground. Even when I was writing the book in 2022–23, there were already posts with titles like Note-taking Became a Full-time Job, so I Stopped, Personal Knowledge Management Is Exhausting, and — my favorite! — Personal Knowledge Management Is Bullshit.

They all trace a similar arc: lured by visions of increased productivity, the author builds an elaborate PKM system. They spend lots of time capturing notes, tagging, linking, organizing, etc. But the expected results never come. Eventually, the author decides the effort is futile, and gives up. Deleting the system brings a sense of relief and renewed agency, which they feel compelled to share.

I’m not here to diss these people. PKMs aren’t for everyone. If it’s not for you, the sooner you stop, the better — perhaps. But I also believe mindset influences the value you get from these systems. And unfortunately, the most common framing for PKMs sets the wrong mindset. It’s the metaphor in the title of Westenberg’s post: second brain.

Problems with the “second brain” metaphor

In Metaphors We Live By, Lakoff and Johnson explain that metaphors don’t just reveal how we talk about things; they also reveal and inform how we think about things, deep down. Which is to say, metaphors matter. And I’ve come to believe the “second brain” metaphor leads to bad thinking about PKMs.

Before proceeding, I’ll say upfront that I admire Tiago Forte’s work. His PARA taxonomy has influenced me. And on the upside, thinking of PKMs as a “second brain” has brought lots of people into the fold.

That said, I think the “second brain” metaphor has three problems:

1. It implies delegating cognition. The promised outcome is a prosthetic mind. That is, the system will relieve you of thinking and (especially!) long-term recall. (Westenberg: “I believed I was solving a problem of forgetting.”)
2. It sets expectations PKMs can’t meet. This isn’t a promise current PKMs — even with AI — can deliver. The system won’t “extract the gold,” at least not for a long time and after a lot of work on your part.
3. These are bad expectations to begin with. Even if PKMs could do this, you shouldn’t want this. If you want to think better, your goal shouldn’t be to delegate your thinking: It should be enabling your first brain to work better.

Enter the knowledge garden

A more fruitful metaphor for PKMs is that of a garden. Many of us already talk about our PKM systems as “places” where we do focused work. This is why I like the garden metaphor: it’s about building a context for you to think in rather than a thing that thinks for you. But it goes beyond that.

There are different kinds of gardens for different purposes. Some are for pleasure, while others are for growing food. Some are industrial; others artisanal. What they all have in common: things grow there. And it doesn’t happen overnight, but after much toil in the soil. For a garden to fulfill its purpose — whatever it might be — it must be stewarded over a long time.

Also, a garden’s structure can’t be rigidly top-down. While some structure is needed, the place’s form emerges over time as it meets real-world needs. Thinking about PKM as a productivity hack leads to overemphasizing upfront structures and workflows at the expense of the more patient approach required by organic processes.

Finally, for many gardeners, the fruit is only part of their garden’s value. Gardening is pleasurable per se. It’s not something they do just because they want to eat. After tall, it’s cheaper and easier to go to the supermarket. Instead, they garden because they find it fulfilling.

Many a garden’s ulterior purpose is providing the kind of groundedness that comes from putting your hand in the soil and nurturing living things. The fruit that comes from such a place tastes better than the one you buy from the store — even if (or perhaps because) you’ve put a lot of work into it.

A garden provides solace and recreation — the opposite of the anxiety that overhangs systems built as productivity hacks. My PKM system provides solace and recreation. So I call it my “knowledge garden,” riffing on the popular digital garden metaphor and Andy Matuschak’s evergreen notes, among others.

I approach my knowledge garden with Field Notes’s tagline in mind: “I’m not writing it down to remember it later, I’m writing it down to remember it now.” I don’t keep a PKM to remember things later, but because writing, structuring, and connecting ideas is how I think. That the words are there for recall later is a bonus, not the main attraction. Clearer thinking is the “gold,” the notes merely record it happened.

But if the point is creating a place for your first brain to work better, that raises an increasingly pressing question: what role should AI — which is being explicitly framed as a prosthetic mind — play there?

AI as amanuensis

To explain how to use AI in a knowledge garden, I’ll refer back to what I wrote in Duly Noted. While I wrote the bulk of the book before ChatGPT came out, the final chapter covers this topic. I’ll offer a quick summary here, but I’ve shared a longer post should you want to dive deeper.

When thinking about your relationship with AI in general, it helps to consider a spectrum. On one end, you reject the technology completely: you don’t want it anywhere near your notes. On the other end, the AI completely replaces you.

Neither extreme is desirable, so most approaches fall somewhere on the spectrum. Toward the “rejection” end, you use AI merely as a copy editor, correcting your spelling and grammar. Millions of people already use tools like Grammarly, and likely won’t object to using AI in this capacity.

Toward the other end, you use AI to write for you. I call this role the ghost writer, although for many people it’s become a ghost thinker.

I get most value around the middle of the spectrum.

There’s a historical precedent here. Some early modern scholars employed live-in secretaries to do various tasks for them: researching, indexing, archiving, retrieving, organizing, translating, summarizing, and running errands. While not as famous as their employers, these people were often seen more as collaborators than anonymous servants. They were called amanuenses.

I consider amanuensis to be the ideal role for AIs in your knowledge garden.

Robots in the garden

This sounds nice in theory, but how does it work in practice? To find out, I undertook a major personal project last year. It’s something I’d always wanted to do: read through the humanities — the major texts that have shaped civilization: Homer to Faulkner, Plato to Freud, the Book of Job to the Communist Manifesto.

Daunting, right? Of course, it’s impossible to do comprehensively in a year. I followed Ted Gioia’s excellent syllabus, which curates key texts, artworks, and musical masterpieces into 52 weeks with reasonable limits. (As you’ll see below, I added cinema to the mix.)

There were two goals to the project. Primarily, I aimed to learn about the main ideas that have shaped our world. But this was also an opportunity to explore how AI might help with such an undertaking. I blogged what I learned every week, including how I used AI.

It was a messy process. That’s what you do in a garden! And the outcome wasn’t an enthusiastic endorsement of AI. Instead, I landed at a map of roles and modalities for how AI can help at different points in the spectrum. Let’s look at nine of these roles.

1. Tutor

The simplest role for AI is as a tutor. You ask it to explain a difficult concept, clarify a confusing passage, translate jargon, etc. I mostly did this via the standard chat UI (although I created a ChatGPT project to preserve context for the course.)

Example:

While reading Freud’s The Interpretation of Dreams, I came across three unfamiliar German terms: es, ich, and über-ich. ChatGPT helpfully explained these are more commonly known as id, ego, and superego — three terms I already understood.

Suggested prompt:

I just read [PASSAGE]. I understand [X] but I’m confused about [Y]. Can you explain [Y] in plain terms, without assuming I have background in [FIELD]?

2. Validator

Another basic role for AI is validating your understanding. To do this, you ask it to review your notes for errors or gaps, do basic fact checking, or critique your reasoning. Again, you can do this via the chat interface, but I also experimented with passing my notes in Obsidian using the Copilot plugin and in Emacs using gptel.

Example:

After reading The Epic of Gilgamesh, I wrote a note in Obsidian summarizing its plot. When I asked ChatGPT to critique my summary, it pointed out that I’d given the central character a redemption arc that isn’t present in the text. I’m so accustomed to the standard hero’s journey, that I projected it onto the book — and an LLM helped me correct this ‘hallucination.’

Suggested prompt:

Here are my notes on [WORK]. What important ideas did I miss or underemphasize? Don’t rewrite my notes — just flag the gaps.

3. Connector

Here’s yet another role you can easily do via chat: identifying thematic, philosophical, or narrative parallels between works. Note I wrote “works” — it’s fun and illuminating to ask for connections across media, genre, time, etc.

Example:

I watched Francis Ford Coppola’s The Conversation on the same week I read Oedipus Rex. For fun, I asked ChatGPT for possible parallels between the two works. Its reply was enlightening: it pointed out how the protagonists of both stories undertook an obsessive investigation that uncovered terrible knowledge.

Suggested prompt:

I’ve been reading [WORK A] and [WORK B]. What philosophical or thematic threads connect them? I’m looking for non-obvious resonances, not surface similarities.

4. Orienter

This role is something of an inversion of the validator. Instead of asking for feedback on your notes after reading a text, here you ask the AI for guidance before reading. You’re looking for framing, historical context, high level outlines, etc. — ideally, without spoilers.

Example:

Before reading Nietzsche’s Beyond Good and Evil and Tolstoy’s The Death of Ivan Illych, I uploaded both books to NotebookLM, which created a podcast for me that explained their thematic contexts. Listening to this podcast in my daily walk helped me better understand the readings.

Suggested prompt:

I’m about to read [WORK] for the first time. Give me enough context to make sense of it — historical background, key arguments, things to watch for — but don’t spoil the experience of discovering it myself.

5. Recommender

This is a useful role for deepening your understanding of a subject: asking for related works that reflect similar themes. It’s also a use case where I noticed considerable improvements in LLM performance over 2025.

Example:

Early in 2025, I read Confucius’s Analects. Perplexity was ahead in web-backed interactions at the time, so I asked it for a list of classic Chinese movies that reflected Confucian values. It responded with five suggestions, some of which it hallucinated. But one of them, Spring in a Small Town, was a bona fide classic — and I likely wouldn’t have learned of it without an LLM. (Later in the year, other chatbots gained this ability and hallucinations dropped across the board.)

Suggested prompt:

I just finished [WORK]. Recommend three films that explore similar themes or ideas. Prioritize films with strong critical reputations — I’d rather have one great recommendation than five mediocre ones.

6. Adversary

Here’s a fun role: asking for an LLM to push back on your position or steelman the opposing point of view. The idea is to expand your understanding by bringing your assumptions to the surface and challenging them.

Example:

After watching Modern Times, I asked ChatGPT to correct my understanding of the movie as a work of Marxist propaganda. The LLM convinced me that the film is in fact more of a humanist statement than a political one. As a result of this interaction, I changed my mind on Chaplin’s work.

Suggested prompt:

Here are my notes on [TOPIC]. Please help me see it through the lens of someone who might be sympathetic to [OPPOSING POSITION] without fully realizing it. What could I improve? Where is my argument weakest? [paste notes]

7. Analyst

This role will also help you appreciate a work from a different perspective. It’s easy: you ask for the LLM to apply a specific critical lens to a reading. Common lenses include Freudian, Marxist, feminist, Girardian, etc.

Example:

The same week I read Freud, my son and I watched Predator, the 1980s sci fi film starring Arnold Schwarzenegger. For fun, I asked ChatGPT to analyze the film through a Freudian lens. The result was both enlightening and hilarious.

Suggested prompt:

Apply a [Marxist / feminist / postcolonial / Jungian] reading to [WORK]. What does this lens reveal that a neutral summary would miss?

8. Mapper

This one’s a bit more esoteric. Some people — me included — are primarily visual: diagrams and drawings aid our understanding. Concept maps can be especially helpful. I’ve built an Agent Skill to allow LLMs like Claude draw concept maps. (Download it from Github.)

Example:

I used this mapping skill to generate a concept map of Virginia Woolf’s To the Lighthouse. It’s not especially insightful, but more of a proof point of using LLMs in a more visual modality.

Suggested prompt:

(Note: install my LLMapper Skill before issuing this prompt)

Generate a concept map for [WORK] centered on the question: “How does the novel’s treatment of [THEME] illuminate [BROADER QUESTION]?”

9. Reflector

This final role is different. Whereas the others took as the object of inquiry a particular work — e.g., a novel or a movie — this last one takes as the object your knowledge garden itself. That is, you point the LLM to a series of notes to analyze patterns over time and suggest improvements.

Example:

I fed all 52 weekly posts from my humanities crash course to Claude Code, and asked it to identify the various roles in which I used AI for learning throughout the year. Its answers — with some curation from me — are the roles you just read.

Suggested prompt:

Here are my notes from [X weeks/months] of reading on [TOPIC]. What patterns do you notice in what I pay attention to? What do I seem to find most interesting, and what do I seem to avoid or underweight?

Takeaways

This list isn’t comprehensive. I’m still experimenting and would love to learn from your experiments as well.

To wind down, I’ll summarize with three key points:

• Don’t try to build a brain. Instead, grow a garden. Metaphors matter. Stop thinking of your PKM system as a prosthetic mind. Instead, think of it as a place you can go to think.
• Use AI to help you think and learn better. AI can help you think better. But use it intentionally. Aim to land somewhere between “replacement” and “rejection.”
• Think calm… and long-term. Learning to think better is a lifelong project. Your knowledge garden is where it happens.

This isn’t a productivity hack. Results won’t come in a year. Results might not come after seven years. Thinking in terms of “results” might be wrong altogether.

You’re building a place to think. And not just any place: a living place that changes and grows over time. It’ll be messy. Good. But if you work on it, it’ll grow more beautiful and fruitful over time. You’ll get peace and satisfaction, even in its imperfection. And the sooner you start, the more material you’ll have for the AIs to help.

I’ll close with this passage from Montaigne, which I hope captures the spirit of what I’ve told you today:

When I dance, I dance; when I sleep, I sleep; yes, and when I walk alone in a beautiful orchard, if my thoughts have been concerned with extraneous incidents for some part of the time, for some other part I lead them back again to the walk, to the orchard, to the sweetness of this solitude, and to myself. Nature has in motherly fashion observed this principle, that the actions she has enjoined on us for our need should also give us pleasure; and she invites us to them not only through reason, but also through appetite. It is wrong to infringe her laws.

When Harry and I first discussed the possibility of starting a podcast, I started an outline of traction-generating ideas I wanted to discuss with him. Episode 32 covers one of my favorites: the MacGuffin.

What is a MacGuffin? I quoted a couple of passages from Dan Hill’s book Dark Matter and Trojan Horses, where I first learned about MacGuffins:

The MacGuffin comes with a particular provenance. The phrase is attributed to Alfred Hitchcock, and has become associated with him ever since. The dictionary defines it as “an object, event, or character in a film or story that serves to set and keep the plot in motion despite usually lacking intrinsic importance.”

And in Hitchcock’s words:

“A MacGuffin you see in most films about spies. It’s the thing that the spies are after. In the days of Rudyard Kipling, it would be the plans of the fort on the Khyber Pass. It would be the plans of an airplane engine, and the plans of an atom bomb, anything you like. It’s always called the thing that the characters on the screen worry about but the audience don’t care… It is the mechanical element that usually crops up in any story.”

I’ve written about MacGuffins before, and won’t restate that here. The TL;DR: in business, they are projects or artifacts that further both strategic and tactical goals simultaneously. While they have tactical value, their real worth comes from the interactions that produce the MacGuffin.

For example, I cited a project where different teams — some of which were in tension with each other — collaborated to produce a user experience journey map. While the final artifact (the map) had value per se (i.e., it informed product design,) the gold came from alignment and better relations between the teams.

Both Harry and I admitted that we’ve participated in more MacGuffin projects unwittingly than by design. But as Hill describes in his book, they can be used as a strategic “play” to precipitate change. As such, a MacGuffin can be a valuable means to gaining traction.

Traction Heroes episode 32: MacGuffins