Yes, you can use ChatGPT and similar general-purpose AI assistants to practise case interviews, and they're genuinely useful for generating prompts, explaining concepts, drilling maths and giving feedback on structure. Their main limitations are consistency, case realism, data integrity and the lack of a structured interviewer experience, which is where dedicated AI case practice tools and human mock interviews add value. Firms like McKinsey explicitly encourage using AI for preparation, but not during interviews or assessments.
Is it allowed?
For preparation, yes. McKinsey's careers site lists "practicing interview questions" and "asking AI to explain concepts or frameworks as part of your prep" as encouraged uses of AI. Using AI to generate real-time answers during an interview, or during assessments unless specifically permitted, is not allowed.
What does ChatGPT do well for case prep?
Explaining concepts
Ask it to explain contribution margin, net revenue retention or how a DCF works, and you'll usually get a clear explanation. You can ask follow-up questions until you understand.
Generating practice prompts
It can create unlimited case prompts by industry and type, and estimation questions to practise.
Drilling maths
You can ask for 20 percentage-change problems or large-number divisions and check your answers.
Feedback on written structures
Type your issue tree and ask whether it is MECE, what's missing and how it could be more tailored.
Behavioural story refinement
It can help you tighten a story's structure and suggest follow-up questions an interviewer might ask. You must keep the content truthful and your own.
Availability and cost
It's available at any time and free tiers exist.
Where does it fall short?
Consistency and realism
A general chatbot doesn't naturally run a case like an interviewer. It may hand you all the data at once, accept weak answers, wander off-script or change facts midway. You can mitigate this with careful prompting, but it takes effort and is inconsistent.
Data integrity
It may invent numbers that don't add up, or make arithmetic errors when "checking" your maths. You need to verify its calculations.
Calibrated feedback
Without a clear scoring rubric, feedback can be generic ("Good structure! Consider adding more detail") or overly positive. It may not know how firms actually evaluate candidates.
Spoken practice
Voice modes are improving, but the experience of being interrupted, probed and paced like a real interviewer is still hard to reproduce reliably.
Current-process knowledge
Its knowledge of recruiting processes may be outdated or wrong. Don't rely on it for facts like current McKinsey Solve formats or application deadlines. Check firm sites.
How do you get the most from ChatGPT for case practice?
Use a detailed prompt, for example:
"Act as a consulting case interviewer for an interviewer-led case. Give me only the prompt first. Wait for my clarifying questions and structure before revealing any data. Reveal exhibits one at a time as tables. Ask me to calculate one figure. Don't give hints unless I ask. At the end, score me 1–5 on structure, maths, insight and synthesis with specific feedback."
Then:
- Check its maths yourself.
- Ask it to stay consistent ("Don't change any numbers you've already given").
- Save good cases and reuse them.
When are dedicated AI case tools more useful?
Purpose-built AI case platforms aim to solve the realism and consistency issues by design: pre-built cases with internally consistent data, an interviewer flow that withholds information until you ask for it, and feedback against case-specific criteria. caize.ai is one such platform: an AI interviewer built specifically for consulting and finance case practice, so you spend your time practising rather than prompt-engineering. Like any tool, judge it on whether its feedback helps you improve. Try several options and keep what works.
Where do humans still matter most?
Human mock interviews, especially with former consultants or candidates who have received offers, remain the best test of:
- Rapport and presence
- How you respond to a real interviewer's tone and body language
- Subtle judgement calls on what's "good enough"
A sensible mix is general AI (ChatGPT and similar) for concepts and drills, dedicated AI case tools for structured volume, and humans for final calibration.
What are the risks?
- Over-reliance: if AI does your thinking during prep, you won't build the skill.
- False confidence: generous feedback can mislead. Calibrate with humans.
- Integrity: never use any AI tool during live interviews or assessments unless permitted.
FAQ
Is ChatGPT good enough on its own for case prep?
It's a useful supplement. Most candidates benefit from adding structured case practice and human mocks.
Can ChatGPT run a realistic candidate-led case?
With careful prompting it can approximate one, but consistency varies.
Will firms know I practised with AI?
Practising with AI is encouraged by firms like McKinsey. What matters is your own performance in the interview.
What's the biggest mistake when using ChatGPT for prep?
Trusting its numbers or recruiting facts without checking.
