Such an incredibly important post, thank you for laying this out so clearly.
I have been talking about this for the past 18 months on my podcast (and almost everywhere else).
Individual productivity became a new vanity metric, and told us nothing about how teams, departments, or organizations were doing. One (human) cog going 10x while others go 1-2x means something is going to break.
You have to be measuring AI readiness and maturity with different attributes as you go from individual to institution.
Only two things I would add are 1) we need responsibility and human-centricity by design to mitigate risk and build things properly the first time, and 2) this also points to the importance of capitalizing on the institutional collective intelligence for making better decisions and architecting strategic work.
:) I’m solo nowadays but over 25 years of witnessing enterprise transformation challenges. AI is the disruptive wake-up call as well as the catalyst to reimagine work itself.
Thank you for this article. What stood out most for me and how I interpret this is that even more important than an AI strategy, companies need an infrastructure that drives, sustains and ensures consistent adoption of AI. I view this as part of what normally would be a change management strategy. Also, thank you for including the critical component of bias! I feel like the topic used to be a central part of most conversations around AI but it has slowly disappeared as the race to proliferate has taken center stage.
To add to that, in the software value chain, AI is squeezing the layers between talent, skills, capabilities and end business outcome.
Since we can now deliver software at a software speed, re-wiring the organization is the key institutional change required, and this is primarily required for non-software businesses.
Good article and spot on about institutional redesign. The irony for me and one thing missing here, is the obvious - that institutions rarely lead these shifts. Individuals usually adopt the tools first and therefore yes, create chaos and the drag the Institutions kicking and screaming into adoption. I went from Private sector to public sector recently where it’s all too apparent that individuals discover the capability, then force institutions industrialize it by causing this chaos of jevons paradox…
Huge difference: electricity could not change the production process' organisation but AI can do that. AI brings not only new technology as electricity did but also new organisation of production process. The challenge: would it be adopted by the human-led production system or must it be changed to AI-lead system?
Great post, and I agree. As someone who also looks at history for parallels, I quite appreciated the electricity example. I run into similar issues that you highlight (individual AI) when it comes to using AI in product marketing work. The current org designs will evolve if we’re to see AI scale outcomes beyond individual time-saving (and marginal quality gains, which is often debatable).
I really enjoyed reading your article, George. Your perspective is very spot on. I’ve been working on this very issue and now have an Agentic AI digital twin of myself that I routinely collaborate with. Human Bob plus “Bot-Bot” make a very powerful team. Here is a 10 minute example of a human interacting with my digital twin: https://vimeo.com/1120703818
Additionally, I just posted a Substack article in which I ask Bot-Bob to share his thoughts on nuclear war as a response to Reid Hoffman’s article about AI always escalating a nuclear war scenario. There is definitely something interesting here.
This is timely and important. Individual amplification is great for the solopreneur, but at an enterprise level it is insufficient, and in fact likely to amplify inefficiencies and misalignment. I’d be interested in your view on how trust is built at scale, I’m grappling with this problem https://buildingandexploring.substack.com/p/can-intent-and-context-scale?r=7lou4&utm_medium=ios and would love to hear how people are thinking about this and building for this problem
I work for a large B2B engineering software developer, and so many of your points resonate with discussions we're having with clients that are on the bleeding edge. Well done.
The textile mill analogy is spot on. From an audit and compliance perspective, I see this tension constantly — organizations adopt AI tools at the individual level but haven't redesigned their governance and controls around it.
The "AI auditor" and "AI compliance" use cases you mention under Bias are particularly compelling. In assurance, the value isn't just finding errors faster — it's having a system that can surface risks that no one thought to look for, unprompted. That's Pillar 7 in action.
The ESG reporting space is a perfect example of where Institutional AI could create real value. Right now, individual analysts use AI to draft disclosures faster — but the institutional challenge is consistency, auditability, and cross-entity comparability. That requires exactly the kind of deterministic, signal-finding architecture you describe.
Great framing. The factories that redesigned the floor will win.
Awesome post! One angle that I think is under-discussed in the institutional vs individual AI debate is that these two forces aren’t really competing, rather they’re forming a new stack.
Institutions are likely to dominate the capital-intensive layers of AI (compute, model training, data infrastructure). That part looks increasingly like utilities.
But the application layer is fragmenting rapidly. With foundation models and agent frameworks becoming more accessible, individuals and very small teams now have the ability to build software that previously required entire companies.
In previous computing waves the pattern was similar: large institutions built the infrastructure, while the most interesting innovation happened at the edge.
The real question might not be institutional AI vs individual AI, but how much leverage a single person can extract from institutional infrastructure.
If that leverage keeps increasing, we may see the most productive era for small teams and solo builders in the history of software.
I’m commenting from way up before I finish. I know it’s frowned upon, but screw it.
It’s our perception that gets in the way. With industry, we had to rework everything from the top down. It looks like we’re getting to see a similar threshold now. Because so much attention was drawn away from the production line by digital reforms in the 90s-2010s…. People refused to believe something could be so quickly outdated.
I speak from experience in the digital photography sector. The bosses kept saying “nothing beats film.” So did the salesman at the counter. And so did the pro and hobby photographer.
But that entire industry died, was dismantled, parts were auctioned to pay debts or raise money for the next venture. It left the factory work without a boat.
I have conflicted views on the matter. I miss working with real film on an SLR that didn’t even need batteries. At the same time, the camera phone lets me get better pictures of my life in progress, without the framing and scaling required for old SLR. Each frame was a guess. And a lot of the time it was absolutely shooting blind. Until you developed it, anything could’ve happened to mess up the photo.
Ten years ago, China took the road less traveled and committed to embedded AI. The investment is now paying off in the form of reduced inflation/increased productivity, and it will continue to pay off for some years to come.
Such an incredibly important post, thank you for laying this out so clearly.
I have been talking about this for the past 18 months on my podcast (and almost everywhere else).
Individual productivity became a new vanity metric, and told us nothing about how teams, departments, or organizations were doing. One (human) cog going 10x while others go 1-2x means something is going to break.
You have to be measuring AI readiness and maturity with different attributes as you go from individual to institution.
Only two things I would add are 1) we need responsibility and human-centricity by design to mitigate risk and build things properly the first time, and 2) this also points to the importance of capitalizing on the institutional collective intelligence for making better decisions and architecting strategic work.
Thanks Rob…agree. Most of the alternate arguments for AI in knowledge work today are made by solo practitioners…
:) I’m solo nowadays but over 25 years of witnessing enterprise transformation challenges. AI is the disruptive wake-up call as well as the catalyst to reimagine work itself.
Thank you for this article. What stood out most for me and how I interpret this is that even more important than an AI strategy, companies need an infrastructure that drives, sustains and ensures consistent adoption of AI. I view this as part of what normally would be a change management strategy. Also, thank you for including the critical component of bias! I feel like the topic used to be a central part of most conversations around AI but it has slowly disappeared as the race to proliferate has taken center stage.
Very good insights.
To add to that, in the software value chain, AI is squeezing the layers between talent, skills, capabilities and end business outcome.
Since we can now deliver software at a software speed, re-wiring the organization is the key institutional change required, and this is primarily required for non-software businesses.
Good article and spot on about institutional redesign. The irony for me and one thing missing here, is the obvious - that institutions rarely lead these shifts. Individuals usually adopt the tools first and therefore yes, create chaos and the drag the Institutions kicking and screaming into adoption. I went from Private sector to public sector recently where it’s all too apparent that individuals discover the capability, then force institutions industrialize it by causing this chaos of jevons paradox…
Huge difference: electricity could not change the production process' organisation but AI can do that. AI brings not only new technology as electricity did but also new organisation of production process. The challenge: would it be adopted by the human-led production system or must it be changed to AI-lead system?
Great post, and I agree. As someone who also looks at history for parallels, I quite appreciated the electricity example. I run into similar issues that you highlight (individual AI) when it comes to using AI in product marketing work. The current org designs will evolve if we’re to see AI scale outcomes beyond individual time-saving (and marginal quality gains, which is often debatable).
How about we increase curiosity and let productivity be the subtext? Because tell you what: harping about productivity doesn't increase it
I really enjoyed reading your article, George. Your perspective is very spot on. I’ve been working on this very issue and now have an Agentic AI digital twin of myself that I routinely collaborate with. Human Bob plus “Bot-Bot” make a very powerful team. Here is a 10 minute example of a human interacting with my digital twin: https://vimeo.com/1120703818
Additionally, I just posted a Substack article in which I ask Bot-Bob to share his thoughts on nuclear war as a response to Reid Hoffman’s article about AI always escalating a nuclear war scenario. There is definitely something interesting here.
https://substack.com/@bobdanna/note/c-226420896?r=3ac1vb&utm_medium=ios&utm_source=notes-share-action
Enjoyed the read!
This is timely and important. Individual amplification is great for the solopreneur, but at an enterprise level it is insufficient, and in fact likely to amplify inefficiencies and misalignment. I’d be interested in your view on how trust is built at scale, I’m grappling with this problem https://buildingandexploring.substack.com/p/can-intent-and-context-scale?r=7lou4&utm_medium=ios and would love to hear how people are thinking about this and building for this problem
I work for a large B2B engineering software developer, and so many of your points resonate with discussions we're having with clients that are on the bleeding edge. Well done.
The textile mill analogy is spot on. From an audit and compliance perspective, I see this tension constantly — organizations adopt AI tools at the individual level but haven't redesigned their governance and controls around it.
The "AI auditor" and "AI compliance" use cases you mention under Bias are particularly compelling. In assurance, the value isn't just finding errors faster — it's having a system that can surface risks that no one thought to look for, unprompted. That's Pillar 7 in action.
The ESG reporting space is a perfect example of where Institutional AI could create real value. Right now, individual analysts use AI to draft disclosures faster — but the institutional challenge is consistency, auditability, and cross-entity comparability. That requires exactly the kind of deterministic, signal-finding architecture you describe.
Great framing. The factories that redesigned the floor will win.
Amazing post written with such clarity.. I hope most of the people understand the value of this blog .. “PURE GOLD” ❤️
Awesome post! One angle that I think is under-discussed in the institutional vs individual AI debate is that these two forces aren’t really competing, rather they’re forming a new stack.
Institutions are likely to dominate the capital-intensive layers of AI (compute, model training, data infrastructure). That part looks increasingly like utilities.
But the application layer is fragmenting rapidly. With foundation models and agent frameworks becoming more accessible, individuals and very small teams now have the ability to build software that previously required entire companies.
In previous computing waves the pattern was similar: large institutions built the infrastructure, while the most interesting innovation happened at the edge.
The real question might not be institutional AI vs individual AI, but how much leverage a single person can extract from institutional infrastructure.
If that leverage keeps increasing, we may see the most productive era for small teams and solo builders in the history of software.
I’m commenting from way up before I finish. I know it’s frowned upon, but screw it.
It’s our perception that gets in the way. With industry, we had to rework everything from the top down. It looks like we’re getting to see a similar threshold now. Because so much attention was drawn away from the production line by digital reforms in the 90s-2010s…. People refused to believe something could be so quickly outdated.
I speak from experience in the digital photography sector. The bosses kept saying “nothing beats film.” So did the salesman at the counter. And so did the pro and hobby photographer.
But that entire industry died, was dismantled, parts were auctioned to pay debts or raise money for the next venture. It left the factory work without a boat.
I have conflicted views on the matter. I miss working with real film on an SLR that didn’t even need batteries. At the same time, the camera phone lets me get better pictures of my life in progress, without the framing and scaling required for old SLR. Each frame was a guess. And a lot of the time it was absolutely shooting blind. Until you developed it, anything could’ve happened to mess up the photo.
Ten years ago, China took the road less traveled and committed to embedded AI. The investment is now paying off in the form of reduced inflation/increased productivity, and it will continue to pay off for some years to come.