TL;DR
McKinsey launched a free AI practice tool for interview case studies. It also tests candidates on how they work with its AI assistant, Lilli.
McKinsey launched a free tool to practice AI in April that gives candidates unlimited attempts at the quantitative case study they will face in their interview. The tool is available globally to applicants for entry-level business analyst and associate positions. The firm says it is designed to level the playing field for candidates who can’t afford expensive consulting advisors.
The consulting interview coaching industry charges between $200 and $500 per hour. McKinsey’s free tool allows candidates to practice the same quantitative scenarios they’ll encounter in the actual interview, as many times as they want. Marie Christine Padberg, global co-leader of talent attraction at McKinsey, told Business Insider that the tool also addresses nerves: “Doing quantitative things is one thing, but doing it while someone is watching you is another.“
The practice tool is one half of a broader integration of AI into McKinsey’s hiring process. The other half has more consequences. Since January, the company has been testing the use of its in-house AI assistant, Lilli, during final-round interviews for business school graduates.
Pilot candidates are asked to use Lilli to analyze a case study and refine their conclusions. Interviewers evaluate how applicants activate the system, evaluate its results, and apply them to a specific client scenario. The test measures curiosity and judgment, not rapid engineering.
McKinsey is not testing whether candidates can avoid AI. It is being tested whether they can work with it effectively. The distinction reflects how consulting work itself has changed. Consultants are now expected to go beyond the analysis that clients can do internally and move toward problem formulation, judgment, and implementation.
The scale of AI within McKinsey makes the shift in hiring logical. CEO Bob Sternfels said at CES in January that the company now has approximately 25,000 AI agents supporting its 60,000 human employees. Eighteen months ago, that number was 3,000. More than 75% of McKinsey employees use Lilli on a monthly basis.
McKinsey has also eliminated approximately 200 technology positions as AI automates non-customer-facing operations. The company reduced its overall workforce by more than 10% between 2023 and 2025. Entry-level roles have been hardest hit, precisely the positions the AI practice tool is designed to help candidates land.
The tension between AI creating and eliminating jobs is playing out across the hiring market. Leading engineering positions increase 19 times year on year. The Claude evangelists earn $240,000. Artificial intelligence chiefs earn almost $500,000. The jobs AI creates pay more and require different skills than the jobs it replaces.
McKinsey’s approach encodes that change in the interview itself. The company does not ask candidates if they can use AI. It’s making fluency in AI an entry condition. CaseBasix, an interview preparation consulting company, said BCG and Bain are likely to continue with similar AI interview components.
The broader pattern is consistent. Detroit automakers are cutting management staff and posting AI positions. Salesforce eliminated 4,000 support jobs after implementing AI agents. McKinsey is simultaneously reducing its workforce and redesigning its hiring process to select people who can work alongside technology that makes others redundant.
The quantitative component is especially important, Padberg stated, because “Even in an AI-enabled workplace, consultants still need to understand how the numbers connect and what they mean.“AI can generate analysis. It can’t yet determine whether the analysis is relevant to a specific client’s problem. That judgment gap is what the McKinsey interview is now designed to test.
The graduating class of 2025 and 2026 is entering the job market where AI fluency is no longer an additional skill. At McKinsey, it is now part of the entrance exam. The free practice tool makes preparation accessible. Lilli’s interview makes the standard clear: if you can’t collaborate with the AI under pressure, you won’t get the job.
