McKinsey asks graduates to use AI chatbot in recruitment

McKinsey is now asking graduate job applicants in the US to use an internal AI tool called Lilli during parts of its recruitment process, according to CaseBasix, a company that supports candidates applying to consulting firms.

CaseBasix said some candidates in final-round interviews are required to use Lilli to complete practical consulting tasks, where they must prompt the AI, assess its output, and deliver structured responses.

The interviews focus on how candidates collaborate and reason with the tool, rather than on technical AI expertise.

These assessments are held alongside other tests for problem solving, structured thinking, personal impact, leadership, and values.

McKinsey, a global management consulting firm, did not comment on the change.

McKinsey’s CEO previously said the firm operates about 20,000 AI agents alongside its 40,000 staff.

🔗 Source: The Guardian

🧠 Food for thought

Implications, context, and why it matters.

McKinsey’s Lilli remains unproven despite a high-profile hiring test

McKinsey now tests some graduates on AI teamwork with Lilli, its in-house assistant. Public material gives no hard metrics like productivity lift or accuracy gains, so real-world maturity stays unclear.
The firm says it runs 20,000 AI agents with 40,000 staff. Without deployment detail or benchmarks, that could be a real shift or a staged recruiting story. Here, “AI agents” refers to software systems that autonomously or semi-autonomously perform tasks.
Only some candidates in final rounds use Lilli, which points to a limited pilot, not a firmwide norm. In those sessions, candidates prompt the tool, assess its output, then share structured answers.

Training vendors can turn AI collaboration into a hiring skill

Human Resources (HR) technology firms and enterprise learning platforms, software that delivers plus tracks employee training, can build assessments for AI teamwork that score prompt quality, output review, plus reasoning with tools 1 2.
TestGorilla and Mercer | Mettl already sell generative AI tests 2 3. These cover prompt structuring, ethics, AI delegation, plus technical skills, so buyers appear to want standard measures in hiring.
Hiring teams now weigh how candidates collaborate and reason with AI over pure technical depth 4. Products should target behavioral skills such as communication, judgment, critical thinking, plus soft skills integration.