The reported avatar pilot is closer to a client conversation than to working through a traditional case from start to finish. You speak to an AI-generated stakeholder, and a McKinsey assessor observes the interaction.

The client may ask for clarification, raise a concern, or introduce information that calls for a thoughtful response. McKinsey has not published a full question bank, detailed rubric, or a firm statement about how much quantitative analysis is included.

As practice, imagine recommending a new product launch and hearing that a distributor is worried about shelf space. Rather than launching into a memorized framework, acknowledge the concern and ask which commercial constraint matters most. That is a CaseBasix-style illustration, not a real interview question.

Practise concise, clear spoken responses and be comfortable adjusting your thinking when the facts warrant it. Avoid anyone promising exact avatar prompts or claiming access to a private scoring guide.

For a deeper explanation, see the AI avatar format and example questions guide for reported details and illustrative scenarios.

You can also follow the discussion asking whether anyone has actually experienced the pilot, but treat unverified personal accounts cautiously.

The part I was missing was whether you need to run a full case on the laptop. This sounds different.

A short role-play with a friend might be more useful than doing extra mental math for this specific part.

Has anyone seen whether the assessor asks a follow-up after the avatar finishes?