Knowledge Stopped Being the Moat. Here's What Replaced It.

Joe Hudson coaches Sam Altman. He also spends much of his time inside OpenAI's research team, working with the people actually building the technology everyone else is reacting to. So when he describes a CEO whose team cut 40% of its active projects and grew revenue per employee within six months, the interesting part isn't the outcome. It's what unlocked it: not a new tool, not a reorg, not more AI, but resolving an unconscious fear of disappointing people.

That's the anecdote Hudson leads with, and I want to be clear about what kind of evidence it is before I say anything else about it. It's a single coaching engagement, described by the person who ran it, in a piece written for a coaching-and-leadership audience. It's not a study. But Hudson's vantage point is genuinely rare – few people get to watch, up close and over years, what actually separates the people thriving inside the most AI-forward organizations on earth from the people struggling – and his argument is worth taking seriously on its own terms, not because it's proven, but because of who's making it and from where.

The argument: effort and knowledge just got cheap

Hudson's framing is straightforward. For most of modern work, the two things that got you ahead were effort and knowledge – put in the hours, know more than the next person. AI is commoditizing both, fast. What's left, in his account, isn't a technical skill at all. It's what he calls emotional clarity: the capacity to feel what you're feeling without being run by it, to stay in a hard conversation instead of deflecting or shutting down, to keep going after a visible failure instead of protecting yourself from the next one.

He's built a rough taxonomy around this – discernment, the capacity for productive conflict, willingness to fail, and quieting negative self-talk – which he calls a "wisdom stack." I'd flag clearly that this is Hudson's own framework, not an externally validated model with peer-reviewed backing. His own organization, the Art of Accomplishment, reports tracking one of its programs for seven years and finding a full standard deviation of improvement in negative self-talk among participants – a real number, but a vendor's own measurement of its own program, not independent research. Treat the wisdom stack as a practitioner's working theory that's shaped how a very influential coach operates inside very influential companies, not as settled science.

Why teams are starting to look like rosters instead of factories

The individual-skill argument sits inside a broader shift Hudson calls the "NBA-ification" of teams: as AI amplifies what one person can do, organizations flatten, and headcount concentrates more capital and consequence onto fewer people. Fast Company reported forecasts that roughly one in five companies will use AI to significantly shrink middle management by the end of 2026 – a projection, not a completed fact – and pointed to Anthropic, Amazon, Shopify, Coinbase, and Block as companies already visibly experimenting with flatter structures and "player-coach" roles.

I'll keep that part brief here on purpose, because the org-structure mechanics – what happens to the coordination work a manager used to do once the layer is gone – is a big enough question that I've given it its own post. This piece has a narrower job: assuming the flattening happens, what makes any one person on that flatter team actually valuable. On an NBA roster, nobody wins by having memorized the most plays. They win by making the right read with two seconds on the clock, staying composed when it gets physical, and making the players around them better. Hudson's argument is that AI-amplified teams are starting to reward the same thing, and knowledge was never really the scarce resource there to begin with.

The parallel evidence: individual contributors are already being paid like this is true

Separately, and worth treating as its own data point rather than something Hudson said, Elena Verna has documented individual-contributor roles increasingly commanding management-level pay without any direct reports – because one AI-amplified person can now complete end-to-end projects that used to require a full team. Verna describes her own move into an IC role at Lovable, spending "about 90% of my time on the parts I actually enjoy – building," rather than managing. That's a personal account, not a market-wide statistic, but it's an independent confirmation of the same underlying shift from a different vantage point: the org chart is flattening in a way that concentrates both output and reward on individuals, not just on titles.

The part that should worry you if you're early-career

There's an uncomfortable implication here for anyone who was planning to build emotional clarity the traditional way – by surviving a decade of hard conversations, failed projects, and disagreements with people more senior than them, in roles junior enough that the mistakes were affordable. That path assumed a pipeline of junior work where the stakes were low enough to practice on. I've written before about what happens to that pipeline once AI takes the easy, junior-level reps – the same reps that used to be where this kind of judgment actually got built, slowly, through repetition and recoverable failure. If Hudson's right that emotional clarity is the scarce skill now, and the entry-level work that used to train it is disappearing at the same time, that's not two separate trends. It's one trend colliding with itself, and I don't think most organizations flattening their structure right now have thought about where the next generation of people with this skill is actually supposed to come from.

What I'd actually do with this

The lazy version of this argument turns into an HR platitude – "hire for EQ" – and stops there. I don't think that's useful, and I don't think it's what the evidence actually supports either. The useful version is a diagnostic question, not a hiring filter: on a team that's already flattening, who do you actually trust with more autonomy as the coordination layer thins out? Not who has the most domain knowledge – that's compressing toward zero as a differentiator, per Hudson's whole argument – but who stays functional and honest in the room when a call goes badly, who can sit in a disagreement without either caving or digging in, and who doesn't need the win to justify trying again.

That's a harder thing to evaluate than a resume, and it's exactly the kind of judgment that gets sloppier, not easier, as teams get flatter and faster. Product strategy built for a world where the tools just execute has to account for who's actually making the calls once the org has fewer layers to catch a bad one – and if Hudson's right about what predicts that, most teams aren't evaluating for it yet. They're still hiring and promoting for the thing that just got cheap.