An AI interview assistant handles a multi-part question by accumulating the caption while the interviewer talks, so "tell me about a project, what went wrong, and what you would do differently" arrives as one caption and the structured answer that follows about 2 seconds after the pause takes the three parts in order, key points first. If the interviewer adds a part while you are already answering, you press Ctrl+Shift+Q to start a new question (or a new question is detected automatically after a 20-second pause in speech), get a fresh answer for the new part, and keep speaking in the same order the parts were asked.
Multi-part questions are where candidates most often lose a thread, because by the time they finish part one they have forgotten part three. A caption on your own screen is a direct fix for that, independent of whether you read the suggested answer at all. This page walks through how the capture, the answer and your own speaking fit together.
How a multi-part question arrives in the app
The live captions do not reset between sentences. As long as the interviewer keeps talking, the caption grows. When they pause, auto-answer can trigger, or you press ⚡ Answer (Ctrl+Shift+A), and the answer is built from everything captured since the last new question. For a question in three parts, that means the answer sees all three parts and addresses them in sequence.
The structure of the answer helps here: key points first, then detail. For a three-part question the key points typically map onto the three parts, so a glance at the top of the window tells you the order before you start. If the interviewer asked the parts in a confusing order, the answer still follows the order they used, which is also the order they expect to hear.
Reading a long answer quickly
A three-part answer is longer than a one-part answer, and you do not have time to read all of it before you speak. Three controls help:
- Bullet points. The bullet-points button reformats the answer as short lines, which is easier to scan while talking.
- Shorter and longer. The shorter-answer button compresses to the essentials; the longer-answer button expands when the interviewer clearly wants depth on each part.
- Scroll without clicking. Ctrl+Shift+/ scrolls the answer while your cursor stays in the meeting window, and the text size can be enlarged in Options so the key points fit in one glance.
The way to use this is to read the key points, start speaking on part one from your own memory, and glance down for part two while you finish part one. You are reading an outline, not a script.
When the interviewer adds a part mid-answer
This is the common pattern: you are halfway through part two and the interviewer says, "and also, how did the team react?" Two things can happen in the app.
If you do nothing, the new words are captured, and after a 20-second pause in speech a new question is detected automatically. That is fine when you are about to finish anyway. If you want the new part answered right away, press Ctrl+Shift+Q to start a new question the moment they add it, then Ctrl+Shift+A when they stop. The new answer appears in about 2 seconds and covers only the added part, which is exactly what you need because you have already handled the earlier parts.
A third option: if the added part changes the whole question ("actually, tell me about a different project where it went badly"), press Ctrl+Shift+Q and then Ctrl+Shift+R to redo the whole answer against the new framing. Our page on real-time interview assistants for follow-up questions covers the chain of follow-ups that usually comes after.
Scenario, keystroke, and what to say
| Scenario | What you press | How you say it |
|---|---|---|
| Three parts asked up front, then a pause | Nothing (auto-answer) or Ctrl+Shift+A | "Three things there. First..." and take them in order |
| A part added while you are answering | Ctrl+Shift+Q when they add it, Ctrl+Shift+A when they stop | "I'll come to that right after this" |
| The added part changes the whole question | Ctrl+Shift+Q, then Ctrl+Shift+R | "Let me reframe for that case" |
| You forgot what part three was | Ctrl+Shift+/ to scroll back to the caption | Read it from the caption; no need to ask |
| Parts asked in a confusing order | Nothing; the answer follows their order | "Taking those in the order you asked..." |
| Interviewer wants depth on every part | Longer-answer button | Give each part its own example |
| Two interviewers each add a part | Ctrl+Shift+Q after each, Ctrl+Shift+A | Name who asked what: "To Priya's point first..." (example we wrote) |
Taking the parts in order out loud
The app gives you the order; signposting is what makes the interviewer hear it. Here is an example we wrote, for "Tell me about a project you led, what went wrong, and what you'd do differently." The suggested key points were: the project and your role; the specific failure and its cause; one concrete change. Spoken, that becomes: "Three parts to that. The project was a billing migration I led across two teams. What went wrong was a cut-over we scheduled too close to quarter end, which surfaced a data mismatch under load. What I'd do differently is run the dual-write period a week longer and put a hard go/no-go gate on the reconciliation numbers."
Notice the counting ("three parts"), the explicit labels ("what went wrong was..."), and the fact that every specific came from the speaker, not the app. The suggested answer supplied the skeleton; the pasted résumé pointed it at a real project; the candidate supplied the quarter-end detail. Our STAR examples page shows the same signposting across more questions, and behavioral interview help covers how to build the story bank these answers draw from.
Multi-part questions in a coding round
Coding interviews stack parts too: "write it, then tell me the complexity, then handle the empty case." With Coding mode on, the first part returns complete code in your chosen language with an example run. For the complexity question, press 🗣 Talk answer (Ctrl+Shift+E) and the next question is answered in words while Coding stays on. For the empty-case change, press Ctrl+Shift+Q and ⚡ Answer for revised code. If the interviewer typed the extra requirement into the shared editor rather than saying it, Read screen (Ctrl+Shift+S) captures what is shown and answers it. The real-time interview copilot page lists every one of these hotkeys in one place.
What the app cannot know
It cannot know which part the interviewer actually cares about. A three-part question often has one part that matters and two that are warm-up, and only the interviewer's tone tells you which. If you are unsure, ask: "Which of those would you like me to go deepest on?" That is a strong move in any interview and it works with or without an AI interview assistant on your screen. The app also cannot stop you from reading all three parts aloud in a monotone; practice the counting and labeling in the AI mock interview, which asks multi-part behavioral questions built from your own job description, until signposting is a habit. If you are choosing a tool, compare AI interview assistants on whether the caption stays visible after the answer appears, since that is what rescues a forgotten part three.
Follow the rules of the employer or assessment you're in — many people use it only to practice, which works on the free plan.
Practice three-part questions out loud
The AI mock interview builds multi-part behavioral questions from your own job description and scores each answer out of 10. Free plan, 15 minutes a day, no signup.
Start a mock interviewFAQ
Does the AI interview assistant answer all parts of a multi-part question or just the last one?
All of them, as long as they were asked before the pause. The caption accumulates while the interviewer talks, and the answer is built from everything captured since the last new question. The key points at the top usually map onto the parts in order.
What do I press when the interviewer adds a part while I am talking?
Press Ctrl+Shift+Q to start a new question when they add it, then Ctrl+Shift+A when they stop. The new answer covers only the added part. If you do nothing, a new question is detected automatically after a 20-second pause in speech.
How do I remember part three while answering part one?
Read it from the caption, which stays on your screen. Ctrl+Shift+/ scrolls without clicking into the window, and the text size can be enlarged in Options. Saying "three parts to that" out loud at the start also commits you to the structure.
Can it tell which part the interviewer cares most about?
No. Tone and emphasis do not come through the caption. If one part seems to matter more, ask the interviewer which they want you to go deepest on. That question reads as thoughtful, not as stalling.
Does this work for coding questions with several steps?
Yes. Coding mode returns code for the first step, Talk answer (Ctrl+Shift+E) handles a conceptual step such as complexity in words, and Ctrl+Shift+Q plus ⚡ Answer produces revised code for an added requirement. Read screen captures a requirement typed into the shared editor.
Is it better to use a shorter or longer answer for a three-part question?
Start with the default and switch to bullet points if you need to scan faster. Use the longer-answer button only when the interviewer clearly wants depth on each part. A shorter answer is useful when you already know two of the three parts cold.