Bob Sternfels: McKinsey Adds 25,000 AI Agents

McKinsey says it has effectively expanded its workforce by adding tens of thousands of AI agents—technology it wants working alongside every employee. The shift shows how quickly AI is changing what consultants deliver, how they work day-to-day, and how firms aim to compete.

McKinsey’s “60,000-person” workforce

McKinsey CEO Bob Sternfels said on Harvard Business Review’s IdeaCast that the firm’s latest tally puts its workforce at about 60,000, made up of roughly 40,000 people and 20,000 AI agents. He later said at CES in Las Vegas that the AI agent number is closer to 25,000, and a McKinsey spokesperson confirmed that figure as the most accurate. Sternfels also noted that about 18 months earlier the firm was using only a few thousand agents, and the goal for the next 18 months is for every employee to be “enabled” by at least one agent.

In practical terms, the company is treating these agents as part of the operating workforce, not as a side experiment. Sternfels has framed the ambition as one AI agent working alongside each human employee, signaling that “agent + consultant” is becoming the default unit of work at the firm.

What AI agents change in consulting

AI agents are typically described as virtual assistants that can complete tasks autonomously, including breaking down problems, planning steps, and executing actions without constant user prompting. McKinsey’s rapid rollout reflects a wider industry push to embed generative AI into everyday consulting workflows, not just use it for drafting text or speeding up slide creation.

Across the consulting market, firms are increasingly moving beyond classic advisory engagements toward multi-year, AI-driven transformation programs. In that environment, competitive advantage can come from how quickly teams translate “desired outcomes” into shipped tools, redesigned processes, and measurable business performance—work that AI agents are designed to accelerate.

QuantumBlack’s role and the talent shift

McKinsey’s AI push is driven through QuantumBlack (about 1,700 people), which leads the firm’s AI initiatives. Alex Singla, a senior partner who co-leads QuantumBlack, said these AI initiatives now account for about 40% of McKinsey’s work.

That shift is also influencing who consulting firms want to hire and develop. Singla said the firm is looking for candidates who can move between a traditional consultant skill set and an engineering mindset—and collaborate effectively with AI systems as part of delivery.

Other firms are making similar moves, including Boston Consulting Group’s use of “forward-deployed” consultants building AI tools directly on client projects (often framed as fast, iterative tool-building rather than only analysis and recommendations). Taken together, the message is that “consulting talent” is being redefined to include more builder-style capabilities, with AI agents handling a growing share of repeatable research, drafting, and workflow steps.

A new model: fees tied to outcomes

Sternfels has also emphasized that AI is reshaping McKinsey’s business model, not just its internal productivity. Instead of relying only on a classic fee-for-service approach linked to the scope of work, the firm is moving toward a model where it aligns with clients on joint business cases and then helps “underwrite” the outcomes of those cases.

This direction matches what McKinsey leaders have described elsewhere: more performance-based arrangements where fees are increasingly contingent on results rather than purely on time and team size. McKinsey has said roughly a quarter of its global fees now come from outcomes-based pricing, and leaders expect that share to rise as clients pursue complex, multi-year transformation programs.

For clients, the appeal is straightforward: large transformations are risky, board-visible bets, and vendors who share accountability can feel more like partners than suppliers. For consulting firms, the model can create upside if the work delivers—and pressure to prove impact in measurable terms.