Redefining Org Charts in an AI-Forward World to Keep Incumbents Nimble

01The pyramid most incumbents run

Most large incumbents share the same shape. A small executive team sits at the top. Layers of middle management sit below it, translating strategy into plans and reporting progress back up. The broadest layer, the rank-and-file employees on the ground, serves customers directly. Each layer exists to coordinate the one below it, and each handoff between layers adds time to every decision.

That design made sense when information moved slowly and every task needed a person. AI changes both conditions. The question for incumbents is where in the pyramid to apply it.

02The default playbook: cut from the bottom

Conventional wisdom, shaped by incumbency bias and managerial inertia, points AI-driven cost cutting at the front line and part of middle management. The headline examples follow that pattern. Salesforce's CEO said the company cut its support headcount from about 9,000 to about 5,000 as AI agents took on customer conversations, with support costs down 17% since early 2025. Gartner predicts that through 2026, 20% of organizations will use AI to flatten their structure, eliminating more than half of current middle management positions.

Sources listed at the end of this note.

These savings are real, and they are short-lived as an advantage. Viewed per head, the pyramid inverts: compensation rises steeply with each layer, and so does the cost of coordination, measured in meetings, approvals and handoffs. Cutting at the bottom removes the least expensive roles and leaves the slowest part of the organization intact. It also uses tools that every competitor can buy. As rivals run the same program, the efficiency gain is competed away and shows up as lower prices for customers instead of durable margin.

The broader record supports the caution. McKinsey reports that more than 80% of companies using generative AI see no material contribution to earnings, and that fewer than 10% of function-specific use cases make it past the pilot stage. Widespread adoption without structural change produces plenty of activity and little lasting advantage.

03The uncomfortable question: designing from a blank page

The long-term benefit of AI is better realized by starting from first principles and asking an uncomfortable question: how would we design this org chart if we were starting afresh today, with every AI tool now at hand?

The honest answer is usually that the company would need less headcount at every level of the pyramid, the top and middle included. Related functions would be consolidated under fewer senior executives, each responsible for more of the business and trusted with more authority to decide. Fewer layers would sit between those executives and the customer. The org chart would resemble an obelisk more than a pyramid: narrow, tall enough to hold real expertise, with decisions made faster and accountability easier to see.

04An example: revenue management at a large public REIT

Consider a large public REIT where leasing, pricing and marketing each sit in a separate silo and report to different executives. Each function has its own managers, its own data and its own view of the customer. Every change to rents, concessions or lead generation requires coordination across three reporting lines, and accountability for revenue is shared so widely that it is hard to assign.

A truly AI-forward, headcount-light design brings all three revenue management functions under a single executive with fewer direct reports, each augmented by AI agents. The gain comes from faster, better-coordinated revenue decisions and one owner accountable for the result, in addition to lower headcount.

05The pace of change depends on the moat

This will play out more slowly in industries with embedded regulatory and structural barriers to entry. Licensing, capital intensity and long asset lives protect incumbents and give them time to adapt. Less protected businesses have less time. In go-to-market software, incumbents such as Salesforce now face challengers such as Clay, which raised its Series D at a $7.1 billion valuation after 4x revenue growth in 2025 and serves more than 17,000 customers. Challengers like these are built around lean, AI-forward structures from the start, and they improve every day.

For an exposed incumbent, the window to redesign on its own terms is shorter than it looks. For a protected one, the window is longer, and so is the cost of using it poorly.

06Why a fractional executive should lead the redesign

The clean-sheet question is uncomfortable for a reason. The executives best placed to answer it are often the ones whose domains would be consolidated. Each is right within their own mandate, and none is positioned to propose a design that removes their own layer or merges their function into a peer's. A strategic fractional executive brings the independence the question requires, along with the operating depth to make the answer work.

The mandate is narrow and time-bound: map how decisions are made today, design the target org chart, identify where AI agents take on routine work, and sequence the transition so the business keeps running. The fractional executive can also help the C-suite see which leaders are ready to take on bigger roles, an insight that is hard to gain from inside the current structure.

The cost of that mandate is small next to the value of an organization that decides faster than its competitors, and smaller still next to the cost of a challenger that already does.