The Partnership Pyramid, Generalized
The claim¶
Law firms and management consultancies have run the same staffing model for a century. A small number of partners own client relationships and outcomes. A much larger group of associates does the analysis and execution underneath them. This bundle's central, most contested claim is that this structure, a partnership pyramid, is not unique to professional services. It generalizes to any organization, once AI agents can supply the staffing multiplier that used to require an army of junior humans.
The professional-services precedent¶
One partner directing several associates produces far more billable, accountable output than that partner working alone, without needing several more partners. That associate-to-partner ratio is the whole mechanism. Current data on major US law firms shows how wide this ratio already runs even within one industry. Chambers Associate's 2026 rankings show Latham & Watkins running roughly 2.5 associates for every partner. Davis Polk & Wardwell, one of the most staff-heavy firms in the market, runs closer to 4.4 associates per partner (Chambers Associate, 2026). A firm does not add partners in proportion to its workload; it adds associates underneath the partners it already has. That is the entire structural insight this page argues generalizes: the scarce, accountable role stays small while the executing layer beneath it scales.
The counter-narrative¶
The strongest objection comes from inside the industry running this staffing model today, and it is more fundamental than "two pyramids side by side." PwC's own flagship report on redesigning the workforce for agentic AI argues the pyramid does not survive at all, in any number. Instead, the winning shape is an hourglass: "with a strong, capable base and a forward-looking leadership tier, connected by a lean, high-performing middle, your org chart looks like an hourglass" (PwC, 2026) — a shape PwC says it is building in-house at itself.
On this view, agents do not simply replace the associate tier underneath a small partner group. They hollow out the middle of the organization, the layer that used to do most of the analysis and execution, while both ends stay populated with humans: senior strategists at the top, and a wide base of humans doing the new entry-level skill of directing agents. PwC has separately described that base and the leadership tier as functioning almost like two workforces under one roof, "one is a pyramid of humans, and one is a pyramid of agents" (PwC, 2025). Either way, PwC's own picture keeps a lot of humans in the building. It does not shrink to a small elite directing a mostly agentic base.
Why it generalizes anyway¶
PwC's hourglass and this page's partnership pyramid agree on the part that matters most: agents absorb the analysis and execution work a much larger tier of humans used to do. They disagree on what happens to the humans that work displaces, and that disagreement traces to a concern specific to firms like PwC, not to work itself.
Professional-services firms need a wide human base for a reason beyond getting today's work done: they train tomorrow's partners. An up-or-out career ladder only works if juniors keep entering at the bottom and rising through it, so PwC has a structural reason to keep humans in that base even where the work itself no longer requires them, and to redefine their job as agent orchestration rather than eliminate the tier outright. Most organizations outside this specific model carry no equivalent obligation to cultivate a pipeline of future leaders from a wide junior base.
Where that obligation is absent, the evidence on how agentic teams actually run today looks like a small elite directing a mostly agentic base, exactly what this page argues generalizes. McKinsey's research on the agentic organization describes agentic teams already operating this way: "a human team of two to five people can already supervise an agent factory of 50 to 100 specialized agents running an end-to-end process such as onboarding a customer, launching a product, or closing the books" (McKinsey, 2025). That ratio runs even wider than anything professional services achieves with human associates. It is the partnership pyramid's underlying mechanic, a small accountable layer directing a much larger executing layer, transferred cleanly to a domain with no talent-pipeline reason to keep that layer human.
Where it does not generalize¶
The partnership pyramid depends on two conditions PwC's hourglass shows are not universal: the executing layer must be genuinely substitutable, and the organization must have no independent reason to keep humans in it anyway. Professional services fails the second condition by design. Plenty of other organizations fail the first.
Gallup's own 2026 workplace data shows how widespread that first failure already is, and it is not a hypothetical edge case.
Among employees with AI access who do not use it, 39% doubt AI can help with their own work (Gallup, 2026). Another 43% call themselves ethically opposed to using it at all (Gallup, 2026). That is a large share of the workforce judging their own work as the kind an agent-staffed pyramid cannot absorb, whether because the work depends on human judgment, relationship trust, or values they do not want automated.
The honest scope of this claim, then, is narrower than "every organization becomes a partnership pyramid." Wherever work is specifiable enough to hand to an agent, and the organization has no independent reason to keep a human doing it anyway, the partnership pyramid's staffing mechanic applies. A meaningful share of work, and a meaningful class of organization, is not there yet.
Sources¶
- Chambers Associate — America's Leading Law Firms Uncovered — Partner-Associate Leverage, 2026. This basic ratio showing how much partner exposure each associate gets... gives a basic idea of how a firm staffs its teams. View source · verified 2026-07-03 · primary
- PwC — No More Pyramids: Rethinking Your Workforce for the Agentic AI Era, 2026. With a strong, capable base and a forward-looking leadership tier, connected by a lean, high-performing middle, your org chart looks like an hourglass. View source · verified 2026-07-03 · primary
- PwC (Prasun Shah, Partner, Global CTO & AI Leader, Workforce Consulting) — PwC on an agentic AI future: Your people will become the 'fundamental differentiator' (UNLEASH), 2025. one is a pyramid of humans, and one is a pyramid of agents – this will be the future operating model. View source · verified 2026-07-03 · primary
- McKinsey & Company — The Agentic Organization: Contours of the Next Paradigm for the AI Era, 2025. In our experience, a human team of two to five people can already supervise an agent factory of 50 to 100 specialized agents running an end-to-end process such as onboarding a customer, launching a product, or closing the books. View source · verified 2026-07-03 · primary
- Gallup — AI in the Workplace: What Separates Adopters and Holdouts, 2026. 39% of non-users say they do not believe AI can assist with the work they do. View source · verified 2026-07-03 · primary