AI is making smaller, highly capable teams an economic necessity, not merely aspirational. The opportunity here is to design the whole human-and-AI business rather than hastily cutting teams down.
Meta has repeatedly declared in unusually plain language the direction that many businesses are confronting. It has described becoming leaner, flattening management layers, cancelling duplicative work and using AI as a non-negotiable productivity accelerant.
Meta is an imperfect guide for almost every other business. Its scale and technical capability are hardly typical. Its approach to organisational change has not been without glaring missteps. But it has made the destination visible: an ambitious company can expect to pursue more of its goals with a smaller, more concentrated and increasingly AI-savvy workforce.
That expectation is spreading. As AI makes useful capability cheaper and more continuously available, boards compare the organisation they have with an imagined alternative: a much leaner human team directing abundant AI capability while producing the same result, or something better.
At first, that lean alternative looks like opportunity. But once competitors all do the same, opportunity becomes obligation. A business carrying layers, roles and repeated human effort that others no longer require will find its cost structure progressively uncompetitive.
We need language for where this pressure is all heading. Call it SToHP: a Small Team of High Performers. SToHP is fast becoming more than economically feasible. For many businesses, some version of it will become economically necessary.
SToHP is not a collection of individually ranked stars. High performance belongs to the configuration of the team. A SToHP combines deep domain expertise with local knowledge of the business: how it makes decisions, where its risks sit, what its customers value and how to survive a challenging week. Its people also have the AI fluency and support required to turn that knowledge into effective direction, delegation and control.
The pressure towards SToHP intersects directly with the Deduplication of Labour. When AI can supply a capability wherever it is needed, a business may no longer need to reproduce the same human capability across every team, shift or layer.
Human overlap in the context of AI can look on the surface like waste. That overlap may provide local knowledge, independent challenge, customer continuity, development capacity or recovery when somebody is absent. That perceived overlap can allow a function amplify and deliver much more than it conventionally could. A crude team contraction sees a duplicated task and removes a duplicated person. A more serious examination asks why the overlap existed and whether AI changes all of those functions or only the most visible one.
An ill-conceived SToHP programme can quickly become regrettable. Salary, headcount and direct output are easy to model and compelling to cut when the pressure is on. Institutional knowledge, exception handling, informal coordination, morale and resilience are all on the line. The costs of shortsighted cuts become painful and dangerous shortly after the cuts are made.
The resulting business may have a small team of impressive people. But if that team is overloaded, essential judgement has no backup, if AI operates without sufficient context or nobody can recover the work when a key person leaves, the team is not high-performing. It is a concentrated dependency with a superficially attractive cost base.
The answer is not to preserve the inherited organisation and it's conventional structure of human capital indefinitely. But economic necessity does not make a crude subtraction of staff anything other than a shortsighted move. A SToHP transition should begin with the outcomes the business must produce and the capabilities those outcomes require. It can then determine which work should remain human, which can become AI labour, which can disappear and where people and AI must work together.
Releasing human capacity is not a goal, and it must be considered more broadly than just a saving. It may be invested to support faster service, wider coverage, deeper expertise or delivering work that the old operating model could not afford. The goal is not minimum headcount. It is the sustainable configuration that lets the business remain prosperous as the economics of work change.
Meta’s signal has arrived early enough to be useful. Leaders can wait until competitive pressure turns contraction into an emergency, then cut towards an imagined smaller company. Or they can use the time to design the thriving AI business they increasingly need to become.
Mara Venn
Published: 30 August 2026
Publication: Field Notes
Category: AI Business 2.0
Reading time: Approximately 3 minutes
Topics: AI operating models, organisational design, knowledge work, workforce strategy
Related reading: The Deduplication of Labour