Short answer: McKinsey says its workforce now includes 20,000 AI agents alongside 40,000 people. Global managing partner Bob Sternfels gave the figures on Harvard Business Review's IdeaCast in January 2026, and the number grew from about 3,000 agents roughly 18 months earlier, a rise of more than 500% (Fortune). A Motley Fool analysis published October 21, 2026, frames the agents as the new junior layer of consulting, doing research and first-draft work once billed by the hour.

Key Takeaways

• 60,000 workers: 40,000 humans and 20,000 AI agents, per Sternfels.
• 3,000 to 20,000 agents in about 18 months.
• From about 45,000 to 40,000 people since 2023, according to The Motley Fool.
• Outcome-based fees: Sternfels said McKinsey is moving away from pure advisory and fee-for-service work.
• "Every employee" will be enabled by 1 or more agents within about 18 months, Sternfels predicted.

• Human employees: Figure: About 40,000; Source: Bob Sternfels, HBR IdeaCast, January 2026
• AI agents: Figure: About 20,000; Source: Bob Sternfels, HBR IdeaCast, January 2026
• AI agents about 18 months earlier: Figure: About 3,000; Source: Fortune, January 2026
• Growth in agent count: Figure: More than 500%; Source: Fortune, January 2026
• Human headcount in 2023: Figure: About 45,000; Source: The Motley Fool, October 2026

What Are McKinsey's Agents Doing?

McKinsey has not published a breakdown of its agents' tasks. The Motley Fool's reading is that they take over the research, analysis and first drafts that junior consultants used to produce. The firm's internal AI assistant, Lilli, is also entering hiring: McKinsey has piloted a final-round assessment in North America where candidates work with Lilli, according to Financial Times reporting cited by HRD. Sternfels also said the firm will look more at liberal arts graduates for creativity and judgment, skills he said AI cannot replace.

Why the Business Model Matters More Than the Headcount

Hourly and team-size billing assume that work scales with people. When agents produce a growing share of the work, that link breaks. Sternfels' answer is to tie fees to client outcomes. The Motley Fool applies the same logic across professional services: it rates research and staffing businesses such as Gartner and Robert Half as more exposed, and legal software and data providers such as RELX and Intapp as better protected. Those ratings are the author's analysis, not established facts.

• Who does research and first drafts: Old consulting model: Junior consultants; Agent-era model: AI agents, reviewed by people
• How fees are set: Old consulting model: Hours and team size; Agent-era model: Outcomes and impact
• What clients buy: Old consulting model: Advice and reports; Agent-era model: Results and working systems
• What hiring looks for: Old consulting model: Analytical horsepower; Agent-era model: Judgment, creativity and AI fluency

What Should Consulting Buyers Do?

1. Ask how agents are used on your engagement. Which tasks are automated, which are reviewed by people, and how quality is checked.
2. Negotiate on outcomes. If the work costs less to produce, price it on results, not on team size.
3. Own what is built. Agree that useful agents, prompts and workflows created for you are handed over with documentation.
4. Check data handling. Confirm where your data goes when consultants run it through internal AI tools.
5. Build your own agent capability. The same tools that make consultants faster can make your team less dependent on them.

For market context, see AI consulting market data for 2026, questions to ask an AI consulting firm and the Gartner IT Symposium 2026 highlights.

Frequently Asked Questions

How many AI agents does McKinsey have?

McKinsey global managing partner Bob Sternfels said on Harvard Business Review's IdeaCast in January 2026 that the firm has 60,000 workers: 40,000 humans and 20,000 AI agents.

How fast did McKinsey's agent count grow?

About 18 months earlier the firm had roughly 3,000 agents, so the count grew more than 500%, according to Fortune.

Is McKinsey changing how it charges clients?

Sternfels said McKinsey is moving away from pure advisory and fee-for-service work toward an outcomes-based model that ties fees to client impact.

What does this mean for companies buying consulting?

Research and first-draft work is becoming cheaper to produce, so buyers can push for outcome-based pricing, ask how agents are used on their engagement, and require that useful agents and workflows are handed over.

Planning an AI agent or consulting project? Book a consultation with Deployed Labs.