Hey Reader,

Welcome to the 15th edition of the 3-2-1.

My last edition on six new thinking skills made possible by AI resulted in a ton of great feedback from readers like you (more than any other edition so far). This issue builds on that idea.

Today, we’re talking about what happens when AI stops being a tool you delegate to and starts being something you work with.

Let’s get into it.

3 Things for Work (in L&D)

1. As We May Work (Taylor Pearson)

• Taylor borrows from freestyle chess to describe what’s emerging: freestyle work, where the machine proposes and the human disposes. Not delegation. Not automation. Something new. Amateurs with machines beat grandmasters with machines because they had better process. (This is also a great primer on Claude Code, a tool that transformed how I run my business).
• 🏋 Effort ≈ 25 min read

2. How Do Workers Develop Good Judgment in the AI Era? (HBR)

• Here’s the catch: the messy, repetitive work that AI handles is the same work that builds judgment. David Duncan argues that AI amplifies existing expertise but doesn’t compensate for its absence. Junior employees who never do the hard stuff can’t evaluate the AI’s output. For L&D leaders, this is the development question of the decade.
• 🏋 Effort ≈ 15 min read

3. Skill Atrophy: Frictionless AI and Cognitive Debt (Cognitive World)

• Professor Jarrahi names what many of us are feeling: when people shift from performing work to selecting from AI outputs, their ability to judge quality deteriorates. He calls it “cognitive debt.” The organizational risk isn’t that AI fails. It’s that humans lose the ability to notice when it does. His fix: attempt-first defaults, reasoning checkpoints, and AI configured to challenge rather than confirm.
• 🏋 Effort ≈ 8 min read

2 Things for Life

1. The Cult of Done (No Boilerplate, YouTube)

• The Cult of Done Manifesto was written in 2009 by maker Bre Pettis and writer Kio Stark. 13 principles that reframe work around finishing, not perfecting. “Accept everything as draft.” “Laugh at perfection. It’s boring and keeps you from being done.” It was written for artists and hackers. In 2026, it’s the operating manual for anyone with an AI tool. The biggest thing holding most people back isn’t technical skill. It’s that they don’t see themselves as someone who builds things. This manifesto fixes that.

2. The Aadam Jacobs Collection (Internet Archive)

• One guy in Chicago spent decades recording concerts with a Sony cassette recorder in his pocket. 10,000 shows. 3,000 bands. He caught Nirvana’s first Chicago gig in 1989 when they were an unknown opening act. After his death, volunteers digitized the collection and made it all available for free. The bands nobody else recorded are the real treasure. That’s the value of showing up and capturing what’s in front of you before it’s gone.

1 Idea from Me

When companies become living organisms

Most of the conversation about AI and work stays at the individual level. New tools, new workflows, new ways of collaborating with machines. The three resources above capture that shift well.

But something bigger is happening at the organizational level. And almost nobody in L&D is talking about it.

In 2023, I published a piece about Arie de Geus’s research on companies that survived for a century or more. His finding was counterintuitive: the ones that lasted didn’t optimize their org charts. They behaved like living organisms. Distributed intelligence. Learning at the edges. Decisions made by people closest to the work.

I thought it was a compelling frame.

I didn’t expect someone to build it.

In March 2026, Jack Dorsey and Roelof Botha published “From Hierarchy to Intelligence.” The core argument: hierarchy exists because information routing had no better mechanism. The Roman military invented it. Industrial management systematized it. Every business school since 1960 has taught some version of it.

The org chart isn’t a design choice. It’s been an engineering constraint. AI changes the constraint.

Now, intelligence can exist centrally in real time and be accessible to everyone in the company at the same time.

At Block, they’re replacing the traditional org chart with three roles to take advantage of this:

1. Individual contributors who go deep as specialists
2. DRIs (directly responsible individuals) who own cross-cutting problems for defined periods
3. Player-coaches who build and develop people at the same time

The hierarchy collapses because AI handles what the hierarchy was actually doing: routing information, filtering signals, surfacing priorities.

Here’s where the individual shift and the structural shift converge.

When work becomes freestyle at the individual level, the capability you need is judgment. When the org chart flattens at the structural level, the capability you need is… also judgment. The ability to work on novel problems without a template. The capacity to build trust when authority is no longer signaled by a title.

These are not capabilities you get from a course library. And the cognitive debt problem from the research above makes it worse: the very work that used to build judgment is the work AI now handles.

Most L&D teams I talk to are still designing for the old org chart. Role-specific content. Manager tracks. Leadership pipelines by level. The structure those programs were designed for is being disassembled.

The real question

The real question isn't "are you using AI?" Almost everyone is.

The real question is "does your AI know more about you and your work than it did yesterday?"

I’m genuinely asking. Hit reply and tell me what you’re seeing.

That’s it for this week. Enjoy your Sunday!

I’ll be back in two weeks ✌️

Andrew

P.S. 👉 Want to talk about how your team develops judgment in an AI-augmented world? Reply to this email. I do a handful of free 30-minute strategy calls each month for L&D in the trenches with this stuff.