Dave, I'm loving this conversation, even while fearing it. On the face of it--the "Pod" model that Coinbase is subscribing to sounds very reasonable. After all, AI can automate so much execution work, we don't really need people handling it. For us Generalists who have built careers orchestrating across streams, we get to keep our jobs!
But peel this back a layer and it gets less reasonable. First let's ask ourselves: How did we Generalists get our visibility across functions? Likely because we had to hash it out in specialty roles across the spectrum, making mistakes but learning a lot along the way. Who will succeed us as the ICs in the middle? And will we be paid more for our expanded roles? Probably--especially since others were let go to pay for it. But still we have a role mobility issue.
These are the issues that keep me up at night. I'll share with you a framework I'm developing that addresses these very issues. Thanks for helping me think.
Jory, you put into words something I’d been circling but hadn’t articulated cleanly—the generalist advantage in the pod model is real, but it’s borrowed. We developed cross-functional range by coming up through the mess of specialty work, making mistakes in contained environments, and gradually earning the visibility. The pod model skips that entire developmental layer.
So who fills the pipeline ten years from now? You can’t manufacture that range from the top down, and the consequences won’t show up on a quarterly earnings call.
Really looking forward to seeing that framework. These conversations need to happen before the pod model becomes the default—not after.
Like you, David, the timing of announcements such as this warrants a level of questioning, re. the legitimacy of the finger-pointing. Putting that aside, there are several points that these changes by Coinbase that should draw the attention and reflection of every leader.
The first is the failure to draw a distinction between intelligence and wisdom. AI is quickly commoditizing the former; it brings none of the latter. By reducing the workforce by 14%, Coinbase is significantly depleting its available wisdom.
While we continue to think of human work with time as a key metric, in fact, productivity is fueled by mental, physical, emotional, and spiritual energy. There is clear evidence already that "supervising" multiple AI agents is mentally exhausting. Creating pods of one can be compared to staffing air traffic control towers with a single person. Our wisdom is developed in community with other workers. Removing that community increases the risk of stagnation as well as burnout.
Flattening the organization is, from my perspective, a move in the right direction. It is a necessary, but may be an insufficient, move. You reference "fewer layers between decision-makers and the work itself." In most cases, the best decisions will be made by those with the most wisdom associated with that decision; that are the workers themselves, not others some number of layers above.
By its very nature, technology speeds things up. AI is certainly doing that. But speed comes with a price. It limits discussion. It leaves little if any space for innovation. It bypasses tension. It tends to ignore intuition along with wisdom. While it may be there in practice, I have seen nothing about the changes Coinbase is making that circumvent these challenges.
The pod model as described here sounds very much like a hybrid between a traditional organization and a networked one. GE Appliances is an example of an organization that has moved much more fully into the networked organization; Bayer is on the same path. While the nature of their businesses is different, the reality is the nature of their workforces is not very different. As they move forward, Coinbase may benefit from looking at these and other companies that have broken the traditional top-down, command-and-control hierarchy.
Finally, I have yet to see a substantive discussion regarding those decisions that Coinbase is keeping out of the realm of AI. This is a key element of successful AI governance in the longer-term.
I really appreciate this perspective, and the air traffic control analogy is one I’m going to be thinking about for a while—it’s a vivid and accurate way to frame what solo pod work could actually feel like under pressure.
The intelligence vs. wisdom distinction is where I’d push back slightly, or at least add some nuance. I don’t think Coinbase—or any company making this move—is unaware that wisdom walks out the door with headcount. I think they’re betting that enough of it stays concentrated in the ICs who remain, and that agents can compensate for the rest. Whether that bet pays off is genuinely unclear, and your point about wisdom developing in community is something the pod model has no good answer for yet.
The governance point at the end is the one I find most underexplored in almost every AI restructuring announcement. What stays human? What decisions are explicitly kept out of the agent loop? Nobody is talking about this loudly enough, and the absence of that conversation is itself a signal worth watching.
David, I agree that there is nuance on the intelligence vs. wisdom distinction. While Coinbase and other companies may be aware of the wisdom that is walking out the door with headcount, this is far from a universal perspective. As we listen to leaders who say that artificial intelligence can replace entire entry-level professions, or a tech CEO proclaiming that AI can even do his job, it sounds like the unconscious definition of intelligence is broad enough to include wisdom, and that the distinction between artificial intelligence and human wisdom is lost. With that in mind, I encourage more public discussions on this distinction and how leaders are addressing it.
Dave, I'm loving this conversation, even while fearing it. On the face of it--the "Pod" model that Coinbase is subscribing to sounds very reasonable. After all, AI can automate so much execution work, we don't really need people handling it. For us Generalists who have built careers orchestrating across streams, we get to keep our jobs!
But peel this back a layer and it gets less reasonable. First let's ask ourselves: How did we Generalists get our visibility across functions? Likely because we had to hash it out in specialty roles across the spectrum, making mistakes but learning a lot along the way. Who will succeed us as the ICs in the middle? And will we be paid more for our expanded roles? Probably--especially since others were let go to pay for it. But still we have a role mobility issue.
These are the issues that keep me up at night. I'll share with you a framework I'm developing that addresses these very issues. Thanks for helping me think.
Jory, you put into words something I’d been circling but hadn’t articulated cleanly—the generalist advantage in the pod model is real, but it’s borrowed. We developed cross-functional range by coming up through the mess of specialty work, making mistakes in contained environments, and gradually earning the visibility. The pod model skips that entire developmental layer.
So who fills the pipeline ten years from now? You can’t manufacture that range from the top down, and the consequences won’t show up on a quarterly earnings call.
Really looking forward to seeing that framework. These conversations need to happen before the pod model becomes the default—not after.
Like you, David, the timing of announcements such as this warrants a level of questioning, re. the legitimacy of the finger-pointing. Putting that aside, there are several points that these changes by Coinbase that should draw the attention and reflection of every leader.
The first is the failure to draw a distinction between intelligence and wisdom. AI is quickly commoditizing the former; it brings none of the latter. By reducing the workforce by 14%, Coinbase is significantly depleting its available wisdom.
While we continue to think of human work with time as a key metric, in fact, productivity is fueled by mental, physical, emotional, and spiritual energy. There is clear evidence already that "supervising" multiple AI agents is mentally exhausting. Creating pods of one can be compared to staffing air traffic control towers with a single person. Our wisdom is developed in community with other workers. Removing that community increases the risk of stagnation as well as burnout.
Flattening the organization is, from my perspective, a move in the right direction. It is a necessary, but may be an insufficient, move. You reference "fewer layers between decision-makers and the work itself." In most cases, the best decisions will be made by those with the most wisdom associated with that decision; that are the workers themselves, not others some number of layers above.
By its very nature, technology speeds things up. AI is certainly doing that. But speed comes with a price. It limits discussion. It leaves little if any space for innovation. It bypasses tension. It tends to ignore intuition along with wisdom. While it may be there in practice, I have seen nothing about the changes Coinbase is making that circumvent these challenges.
The pod model as described here sounds very much like a hybrid between a traditional organization and a networked one. GE Appliances is an example of an organization that has moved much more fully into the networked organization; Bayer is on the same path. While the nature of their businesses is different, the reality is the nature of their workforces is not very different. As they move forward, Coinbase may benefit from looking at these and other companies that have broken the traditional top-down, command-and-control hierarchy.
Finally, I have yet to see a substantive discussion regarding those decisions that Coinbase is keeping out of the realm of AI. This is a key element of successful AI governance in the longer-term.
Brian,
I really appreciate this perspective, and the air traffic control analogy is one I’m going to be thinking about for a while—it’s a vivid and accurate way to frame what solo pod work could actually feel like under pressure.
The intelligence vs. wisdom distinction is where I’d push back slightly, or at least add some nuance. I don’t think Coinbase—or any company making this move—is unaware that wisdom walks out the door with headcount. I think they’re betting that enough of it stays concentrated in the ICs who remain, and that agents can compensate for the rest. Whether that bet pays off is genuinely unclear, and your point about wisdom developing in community is something the pod model has no good answer for yet.
The governance point at the end is the one I find most underexplored in almost every AI restructuring announcement. What stays human? What decisions are explicitly kept out of the agent loop? Nobody is talking about this loudly enough, and the absence of that conversation is itself a signal worth watching.
Thanks for adding real depth here.
David, I agree that there is nuance on the intelligence vs. wisdom distinction. While Coinbase and other companies may be aware of the wisdom that is walking out the door with headcount, this is far from a universal perspective. As we listen to leaders who say that artificial intelligence can replace entire entry-level professions, or a tech CEO proclaiming that AI can even do his job, it sounds like the unconscious definition of intelligence is broad enough to include wisdom, and that the distinction between artificial intelligence and human wisdom is lost. With that in mind, I encourage more public discussions on this distinction and how leaders are addressing it.