AI Interview Copilot for Case Interviews: What Works and What Does Not

Last updated: August 2026

Quick Answer

For case interviews the honest answer has two halves. In an MBB live case, do not plan on using an on-screen assistant: McKinsey states that external apps, websites, calculators and prepared notes are not permitted during its interviews, and the other major firms run their live cases as proctored assessments. For those rounds, the tool that helps is mock practice: OphyAI Interview Practice runs case-format mock interviews with follow-up questions, a transcript and per-answer feedback. In the much larger population of case-style interviews outside MBB, corporate strategy teams, in-house consulting groups, product and analytics roles, and boutique firms that run a case on Zoom without a stated tool policy, OphyAI Interview Copilot can transcribe the prompt and keep your structure visible, if the employer’s rules allow it. Ask, then decide.

Situation	What actually helps
McKinsey, BCG or Bain live case	Case-format mock practice beforehand. Outside tools are not permitted in the round.
BCG Casey or a chatbot case	Practice with the firm’s own tool. Treat the assessment as proctored.
Corporate strategy or in-house consulting case	Ask about tool rules. If permitted, Copilot for transcription and structure.
Product, analytics or ops case on Zoom	Same: ask first, then use Copilot for the prompt and your framework prompts.
Any mock case	Copilot and Interview Practice both, with no ambiguity about the rules.

What Changes in This Situation

A case interview is not a question-and-answer round, and that changes what a real-time assistant can and cannot do for you.

It is interactive, and the interviewer drives. McKinsey runs interviewer-led cases where the interviewer sets each step and may share exhibits on screen; BCG and Bain have historically run more candidate-led cases where you propose the structure and drive. Either way the round is a conversation with turns every thirty seconds, not a monologue you can pre-plan. Reading suggested text mid-turn costs you the thing being scored: responsiveness.

Structure is graded, not recalled. Interviewers assess whether your issue tree fits this problem, not whether you can name a framework. A generic profitability tree pasted onto a market entry case scores worse than a rough structure you built from the prompt. That means a tool that supplies frameworks does not help much; a tool that helps you rehearse building them does.

Mental math is explicitly tested. Cases include arithmetic you are expected to do out loud, and at MBB you are not permitted a calculator. Practising with any computational aid trains the wrong reflex.

The formats are increasingly proctored or automated. BCG’s Casey chatbot case has replaced a live first round in most offices, and McKinsey’s Solve assessment includes an online case. In 2026, McKinsey has also piloted a final-round component in which candidates work with its internal AI platform as part of the assessment itself. When AI is inside the assessment, using your own AI alongside it is clearly outside the rules.

Tool rules vary a lot outside consulting. A product-sense case at a tech company or a strategy case with an in-house team is often just a Zoom call with a hiring manager who has no stated policy. That is where asking gets you a clear answer, and where a copilot is genuinely usable.

How Interview Copilot Helps (Step by Step)

Where the employer permits it, and in every mock case:

1. Load your case notes before the call. Put your own material into Copilot: the structures you actually use, the metrics you tend to forget, your two-sentence synthesis template, and the numbers you researched about the company. Copilot works from what you load beforehand rather than fetching things mid-round.

2. Let it transcribe the prompt. Case prompts are dense and delivered once, and the most common early mistake is answering a slightly different question. A live transcript of the prompt on screen means you can re-read the exact wording before you structure, instead of reconstructing it from memory.

3. Use the overlay to hold your structure, not to source it. The value during a case is keeping your own issue tree visible while you talk, so branch three does not disappear because branch one went well. Whisper Mode keeps that in a private overlay on your screen.

4. Keep the synthesis honest. Most candidates lose points at the recommendation: they summarise what they did rather than saying what the client should do and why. Having your synthesis template visible (“recommendation, two supporting reasons, one risk, next step”) is a small thing that fixes a common failure.

5. Review the transcript afterwards. The transcript is the part most candidates skip and the part that improves you. Read where you narrated arithmetic instead of stating a number, where you asked for data you already had, and where the interviewer redirected you.

6. Do the actual training in mock cases. OphyAI Interview Practice runs case-format mocks with follow-up questions and per-answer feedback, which is the closest available analogue to a live case and carries no rules ambiguity. This is where the improvement comes from. Start practicing →

What It Cannot Do

It cannot be used in an MBB live case. McKinsey’s published guidance for candidates prohibits external apps, websites, calculators and prepared notes during interviews. Treat BCG and Bain live rounds the same way unless a recruiter tells you otherwise in writing.

It cannot do case math for you. Nor should it. Arithmetic out loud is a scored dimension, and the interviewer will ask how you got the number.

It cannot build a structure that fits a case it has not seen. A case is defined by its specific prompt and constraints. A generic tree supplied mid-round is exactly the failure mode interviewers are trained to catch.

It cannot substitute for reps. Case performance is a motor skill: clarify, structure, prioritise, compute, synthesise, under time pressure. Nothing produces that except doing it repeatedly and hearing where it went wrong.

It cannot tell you an employer’s rules. Only the recruiter can. Ask.

Set-Up Checklist

• Ask about tools first. One line to the recruiter: “Are candidates permitted to use notes or any on-screen tools during the case?” A yes or no removes the whole question.
• If the answer is no, or you get no answer, do not use it. Prepare with mocks instead, and keep the round clean.
• If the answer is yes, create your account and open Interview Copilot at least ten minutes before the call. First session free for eligible new users, up to 15 minutes, no card.
• Load your material: your structures, your synthesis template, your researched facts about the company and the industry, and the two or three metrics you always forget.
• Test the platform you will be on. Copilot supports live interviews on Zoom, Microsoft Teams and Google Meet.
• Position the overlay below your camera so glancing at it looks like glancing at the interviewer.
• Keep paper and a pen anyway. Case interviews are structured on paper, and drawing your tree by hand is faster than reading one.
• Run one mock case first in Interview Practice so the format is not new on the day.

Frequently Asked Questions

Can I use an AI interview copilot during a McKinsey, BCG, or Bain case interview?

No. McKinsey’s candidate guidance states that external apps, websites, calculators and prepared notes are not permitted during interviews, and the other major firms run their live cases as proctored assessments with the same expectation. BCG’s Casey chatbot case and McKinsey’s Solve assessment are online assessments, which means they are monitored formats rather than informal conversations. Use mock practice for these rounds and sit the real case unassisted.

Where is a copilot actually appropriate for case-style interviews?

In case-style rounds outside the major consulting firms, where there is often no stated tool policy: corporate strategy teams, in-house consulting groups, product-sense and analytics cases at technology companies, operations cases, and some boutique firms. The rule is the same everywhere: ask your recruiter whether tools or notes are permitted, and follow the answer. It is also appropriate in every mock case, where there are no rules to breach and the feedback is the point.

Does Interview Copilot help with case math?

Not directly, and it should not. Case arithmetic is a scored part of the interview, done out loud, usually without a calculator, and the interviewer will ask you to walk through how you reached a number. What helps is drilling the arithmetic itself: percentages, breakevens, market sizing and ratio work under a timer, until estimation feels routine. Our consulting case interview frameworks guide includes math drills alongside the seven core structures.

What is the difference between Interview Practice and Interview Copilot for case prep?

Interview Practice is before the interview: voice or text mock cases with follow-up questions, a transcript and per-answer feedback on structure, prioritisation and synthesis. On the free plan a 5-credit grant covers one session capped at 15 minutes; paid plans use 10 credits per session. Interview Copilot is during a live interview on Zoom, Teams or Meet, where employer rules allow it, giving real-time transcription and a private overlay of the notes you loaded. For MBB preparation, Practice is the relevant product.

How many practice cases should I do before a consulting interview?

There is no published number, and the honest answer is that it depends on your starting point, but candidates who report doing well typically describe a range of roughly fifteen to thirty full cases, spread over several weeks, with at least a third of them live against another person rather than read from a book. Volume matters less than variety and feedback: cases across profitability, market entry, pricing, M&A and operations, each followed by a specific note on what to fix.

Will an interviewer be able to tell I am using an on-screen tool?

We make no claims about detection either way, and you should not choose a tool on that basis. The decision should be about the rules of the round you are in: if outside tools are prohibited, do not use them, and if they are permitted, use them openly and comfortably. In a case interview there is also a practical argument: reading from a screen slows your turns in a conversation where responsiveness is scored, which is a cost regardless of anything else.

What should I prepare instead if my case round prohibits tools?

Do full timed cases out loud, ideally with a partner who interrupts. Build a habit for the first ninety seconds: repeat the question back, ask two or three clarifying questions, state your structure before you dive into a branch. Drill mental math daily. Prepare a synthesis template you can deliver in thirty seconds. Then rehearse the whole loop in Interview Practice case format and read the transcripts for the steps you keep skipping. Our management consulting interview guide sets out the full preparation sequence.

Sources and verification notes

Checked August 2026. Firm-specific formats and rules change, so confirm the current process with your recruiter or the firm’s candidate materials before your round. Direct fetches of first-party pages were blocked in the environment this post was written in, so the statements below are cited to the pages that carry them, and each was cross-checked against at least two independent write-ups. Product facts about OphyAI reflect the current product specification.

• McKinsey, interviewing: candidate guidance on the interview process, including the prohibition on external apps, websites, calculators and prepared notes during interviews.
• Hacking the Case Interview, BCG online case and Casey chatbot: the chatbot case format, its length, and its adoption across BCG offices.
• StrategyCase, AI in consulting interviews and ManagementConsulted, McKinsey AI interview: reporting on AI components being built into firm assessments in 2026. Third-party reporting, treated as such.
• RoadToOffer, McKinsey case interview guide: interviewer-led versus candidate-led formats and the structure of live virtual cases.
• OphyAI Interview Copilot: supported platforms, live transcript, Whisper Mode overlay, language support, and free first session terms.