AI and the Corporate Pyramid: Who Stays, Who Goes?
The End of Middle Management or the Birth of a New Corporate Model?
Most of the talk around AI is on efficiency, productivity and skills. But what about organizational structures?
If you’ve been following Work3 for a while, you know we’ve talked about modern and innovative models like Holacracy (which revolutes around self-managing teams and no hierarchy) and Web3 / Blockchain based concepts.
The question for today is: will the new kid on the block (AI) also change how organizations structure and manage their workforce?
Let’s dive in!
Flattening Organizations
It’s impossible to have a perfect number, but management makes up an average of 10-20% of the total workforce. Unsurprisingly, cost is disproportionate, with high variations but likely starting at 30% of the total cost.
CEO-to-worker compensation comparison has skyrocketed from the 90s, and though it’s had its ups and downs, still remains at 300x. Even though this is for C-level, I would expect a similar curve (though not as steep) for middle management.
It seems like disparity is increasing further and further, and this has a huge problem hidden in plain sight: connection to reality. Many C-levels at big organizations, either because of generational differences or because of distance from operations, use a completely different map of the territory to navigate decision. Whilst there are instances where this is useful, it has a blind spot towards what really happens at the company and in the market, often resulting in long and costly crusades - anyone remember about Meta’s investment/gamble of just a few $bn in the Metaverse?
Another effect, is on the level of emotional and cultural distance. How can a CEO earning 300x more than one of her employees, think that they could have the same level of buy-in, or even worse, a sense of belonging?
So in these delicate times (likely to get even worse soon, I’m afraid) we see companies raking in billions in revenues slashing 5-15% in workforce, while increasing bonuses by 200% (again looking at you, Meta).
Others, are starting to cut middle-management (PwC, Unilever a while back) including Amazon, making waves with its plan to cut 14,000 managerial roles by early 2025, aiming to save up to $3.6 billion annually. CEO Andy Jassy wants to increase the ratio of individual contributors to managers by at least 15%, pushing the company toward a leaner, faster, and more agile structure. The strategy is framed as fostering a culture of urgency, ownership, and collaboration—essentially operating like "the world's largest startup."
Unsurprisingly, this restructuring has raised many concerns among employees about limited career advancement opportunities and growing dissatisfaction with Amazon's work environment. While Jassy highlights the benefits of flattening hierarchies and reducing bureaucracy, the move coincides with layoffs and a shift back to full-time office work, leaving many wondering whether the cost of efficiency is coming at the expense of employee morale.
Is all of this tied back to the rise of AI or is it result of macro-economic and market trends?
It’s impossile to say precisely, but if estimates (with their huuuuge ranges and doubts in terms of calculation methods) give anywhere a potential ‘efficiency’ improvement of 10-40%, I would expect CEOs looking at that sort of replacement / hiring avoidance. Of course this is incredibly simplistic and doesn’t take into account any kind of risks in timing, birth of potential new roles and so and so on - but that’s how we roll and what is being incentivized as an approach (just look at DOGE).
Also, by just having a look at this graph - I have a hunch that Amazon is trying to catch-up to the game, and trying to ‘reverse engineer’ how they improve their revenue/employee.
Anyway, this will deserve a piece on its own - as time will tell which factors plays the most, and what results come out of this wave of macho-capitalism (anyone coined this already? I’ll take credit if not).
Hackman Authority Matrix
Hackman’s Authority Matrix (1986) provides a great lens to understand how workforce dynamics are evolving. It outlines a spectrum of team structures, from manager-led to self-managing, self-designing, and fully self-governing teams. Companies like Amazon are moving toward the latter, shifting authority from traditional management to teams themselves. While this concept isn’t new—Spotify and Stripe tried similar approaches years ago—they struggled to execute effectively.
AI enables these shifts and makes self-organizing teams more viable than ever. For example, departments like R&D might operate as self-governing teams, while Finance could function as self-managing units—all within the same organization. These structures can coexist, creating flexibility across business units.
It’s important to note that reducing the Management Index (MI)—the ratio of individual contributors to managers—isn’t the goal in itself. A lower MI only adds value when it aligns with improved processes and value generation. Flattening hierarchies without rethinking workflows or operations won’t yield meaningful results. This is where AI hopefully shines: it should empower teams with decision-support systems, reduces time to insight, and facilitates collaboration by cutting through complexity.
So, does this mean middle-management is doomed? Not really.
Human-AI augmentation will impact individual contributors, but also management tasks. If less time is spent on admin, more strategic work should take its place, something that requires creativity - not efficiency (check out my recent piece on this and the distinction with productivity).
Human-AI Collaboration - A new type of management will arise: managing the input / output of the AI itself, plus the interaction between human workforce and the AI agent workfoce. Again, this deserves a deep dive on its own, but it helps us segway into the next section of today’s topic.
New Paradigms
So how will organizations look like?
There’s various options and none mutually exclusive, but one thing is for certain: adaptability, agility and innovation are going to be fundamental in the age of AI (and should be measured). With almost no barriers to entry to technology, and less capital needed to start ideas (or clone them) the pace of change will be higher. Arguably, big corporations with large balance sheets will still have an edge given their access to resources, but it’s not always for given - because of the ‘Innovation Dilemma’ (just look at Google for example, struggling to adjust to AI Search because they need to protect their ad revenues):
The innovator's dilemma describes how successful companies, focused on serving existing customers and improving current products (sustaining innovations), often overlook disruptive technologies that initially underperform. These disruptions eventually surpass established offerings, causing market leaders to fail because they prioritized short-term gains over long-term, potentially transformative innovations. Essentially, it's the paradox of doing everything right, yet still failing.
Here’s three options though we should be considering:
Exponential - We’ve already seen how the latest and hottest startups/scaleups are achieving staggering revenue results with smaller headcounts. Full article on this kind of org here.
Holocratic - Out with the pyramids, power will be spread through ‘circles’ of influence and role mobility to adjust Full article here.
Web3 Based - One of the first Work3 articles was about DAOs (Decentralized Autonomous Organizations), because in the wake of the Web3 ‘summer’ in 2020/2021, many people in the space took building companies to the blockchain, and found new models to organize compensation, capital, and collaboration. Full article here.
Beyond internal restructuring, we will also be reshaping the very fabric of operations. By streamlining interactions between people and processes, starting with IT and now expanding into areas like HR and finance, AI is becoming a critical decision support system. Furthermore, it's impacting external operations, automating and augmenting complex coordination in areas like supply chain and sourcing. As organizations adapt, they must not only navigate internal shifts but also leverage AI to optimize their broader operational ecosystems, ensuring they remain agile and competitive.
We’ve seen how workforce, operations and processes are changing and how this can change organizational structures. The real question however, will be: who will be quick enough and brave enough to adapt?
My hunch is that larger corporations (see Amazon) may well go into ‘efficiency’ (or ‘Founder’) mode, but they may struggle to truly innovate, falling victim to the innovator's dilemma. In contrast, smaller, more agile organizations, embracing models like exponential growth, Holacracy, or DAOs, could leverage AI to achieve rapid phase transitions. They can more easily shift from a stable "franchise" state to one driven by "loonshot" innovations, fostering a dynamic equilibrium that balances efficiency with radical new ideas. The key will be understanding and managing these phase transitions – nurturing the "artists" alongside the "soldiers" – to unlock the full potential of AI and redefine the future of work.
Moreover, we're likely to see the rise of truly ‘AI Native’ organizations: companies built from the ground up with AI integrated into their core structure and operations. These organizations won't just adopt AI tools; they'll design their entire workflows and decision-making processes around AI, enabling maximum agility and innovation.
In this digital age, ‘designing work’ becomes key as we have many more choices from filling vacancies, to using contractors effectively, agencies, platform workers, AI agents etc.
We need to throw out the old Organisation Design rule book. (Spoiler: we plan to write much more about this in 2025). Who will design and build work, centralised teams like HR, or self-managed teams?
We have hypothesised about smaller organisations and a more liquid decentralised workforce with self-organised teams. It might be that AI that is the cover initially for layoffs, and then the catalyst for substantial new organisational shapes after.
Before jumping ahead too fast, some suggestions for work designers would be:
Experiment - no need to go ‘all in’ at the start.
Monitor progress - against intended goals
Speak to the people actually doing the work - feels silly to actually write this down, but you can write to me for examples!
Measure the impact of AI on your organisation - There’s no one-size fits all, but there are some good tools and frameworks out there (more to come on this soon)
Many questions remain, but questioning is progress.
Ciao,
Matteo
Amazing! Great to see someone writing about Web3/blockchain-based work models. I find this super interesting.