Something is shifting in how the best work gets done.
Not slowly, in the background, the way most business shifts happen. Visibly. Quarter over quarter. In the numbers, in the org charts, in the stories founders and CTOs are telling each other at off-sites and in Slack threads and on earnings calls.
The story goes something like this: a project that would have taken a cross-functional team of six, a two-month timeline, and a project manager to keep it all together got done in three weeks by one person who knew how to work with AI.
This is not an isolated story. It is becoming a pattern. And the implications for how companies are built, structured, and staffed are only beginning to be understood.
For decades, the equation for scaling output was simple: more output requires more people. You needed a bigger team to run a bigger campaign. You needed more engineers to ship more features. You needed more analysts to process more data. Headcount and capability were essentially the same variable.
AI is breaking that equation, not by eliminating the need for people, but by changing what one person can do.
Zuckerberg described the internal shift plainly during the same earnings call: projects that used to require big teams are now being completed by single, highly skilled individuals. His response was to flatten Meta's organisational structure, elevate individual contributors, reduce management layers, and invest in AI-native tooling to extend what each person could do independently.
This is not a small operational tweak. It is a fundamental rethinking of the relationship between talent, tools, and output.
Before AI entered the picture, small teams already had structural advantages that large teams could not replicate. Fewer people means faster decisions. Less communication overhead. No alignment tax, the invisible cost paid whenever five people need to be on the same page before anything moves forward. Fewer layers between an idea and its execution.
Large organisations often solve coordination problems by adding more coordination. More managers, more processes, more meetings, more documentation. It works, but it is expensive in time and energy. A lot of what passes for productivity in large teams is actually coordination, which is necessary but not the same as output.
AI does not reduce the coordination overhead of large teams. What it does do, dramatically, is extend what a small team can produce. Staying lean is no longer a constraint to overcome on the way to scale. For the right kind of work, it is the strategy.
This is why AI-native startups are increasingly choosing to stay small by design. Not because they cannot afford to hire, but because small is genuinely faster and increasingly more capable than it used to be. The old tradeoff between team size and team capability has shifted.
The implications inside larger organisations are more complex and more interesting than they first appear.
The obvious read is that smaller teams mean fewer people, which means headcount reductions. And there is truth in that, eventually. But that is the lagging consequence, not the immediate opportunity.
The immediate opportunity is leverage. Every enterprise has work that currently sits queued behind resource constraints: projects that would create real value but cannot start because the right people are not available, or the team is too small, or the budget for a full project team does not exist. AI changes the calculus on all of those.
This is not theoretical. GitHub reported that developer activity on its platform, measured by pull requests and code commits, jumped by more than 20% year over year in 2025. The developer workforce did not grow by anything close to that rate. The delta is AI-assisted work. More is getting built, by roughly the same number of people.
The companies capturing this leverage are not doing it by cutting teams and hoping AI fills the gap. They are doing it by giving their best people better tools and genuinely harder problems to work on.
There is a concept in organisational design sometimes called the coordination tax: the fraction of everyone's time that goes toward keeping a large team aligned rather than producing output. Research on software teams has consistently found that as team size grows, per-person productivity tends to decline. Not because the individuals get worse, but because the overhead grows faster than the output does.
AI does not solve the coordination tax in large organisations. But it does something interesting: it makes the productivity gap between small, well-equipped teams and large, coordination-heavy ones more visible and more consequential.
A five-person team that moves fast, uses AI tools deeply, and makes decisions without layers of approval can now produce outcomes that previously required twenty people and a project management office. The gap between what these two structures can accomplish in a quarter is widening, not narrowing.
This is the dynamic reshaping how the best founders and engineering leaders think about team structure in 2026. Not as a constraint — how do we make our large team work better — but as a design choice: how small can we be and still achieve what we are trying to build?
The old unit of competitive advantage in knowledge work was the team: its size, its composition, its processes, its institutional knowledge. Scale was a moat. A bigger team with more experience and more resources could out-produce a smaller one, almost by definition.
AI is shifting that unit from the team to the individual. The new competitive advantage is the person who can hold a large problem in their head, deploy AI across the breadth of that problem, and execute without waiting for consensus or resourcing. This person is more valuable than they have ever been. And there are nowhere near enough of them to go around.
This is what Zuckerberg was pointing at when he said Meta is focused on being the best place for highly talented individuals to make a massive impact. The competition for people who genuinely know how to work with AI, not just use it but integrate it into how they think and build, is intensifying. These individuals are not a subcategory of knowledge workers. They are becoming the primary unit around which ambitious organisations are designing themselves.
The question for every enterprise leader is not whether this shift is happening. It is whether the organisation is structured to take advantage of it, or whether it is structured in a way that will slow these people down, bury them in coordination overhead, and gradually push them toward environments where they can move faster.
The organisations getting this right are doing a few things that are worth naming.
If your best engineer, or your best analyst, or your best strategist spent the next quarter working with AI as deeply and intentionally as the power users Zuckerberg described, what would they be able to produce that they cannot produce today?
And then: is your organisation structured to let them?
Most organisations are not. They are built for coordination at scale. The overhead that made sense when large teams were the only way to do large work has calcified into processes, approval chains, and team structures that now slow down the very people who could benefit most from the leverage AI offers.
The shift happening right now is not about headcount. It is about what you can build when the right person has the right tools and enough room to move.
The teams figuring that out are not the largest teams in their industries. They are increasingly not even close to the largest. They are just the ones that compound fastest.
If you're planning an automation rollout, map the exceptions first. Find the people who handle them. Then decide.
We've delivered $100M+ in business impact across IT services, healthcare, HR tech, and fintech.
Book a Scoping Call
