Interview Prep AI: A Practical Workflow Before a Real Interview

The five-step AI interview prep workflow

Use this when the interview is days away, not months away.

1. 01

   Build the evidence list

   Pick three to five real work moments. For each one, write what happened, what you personally did, what changed, and what proof you could defend if asked. If you do not have enough raw material, start with a STAR story bank before opening an AI tool.
2. 02

   Translate the job description

   Copy the role requirements into a short list of likely interview themes: ownership, stakeholder management, customer judgment, technical depth, leadership, ambiguity, or execution under pressure. Do not ask AI to guess your experience. Ask it to identify what the role is likely to test.

3. 03

   Run a focused mock round

   Ask AI to act as an interviewer and ask one question at a time. Answer out loud. If the tool gives feedback before you speak, stop and reset the prompt. You need practice retrieving your own answer, not reading its version.

4. 04

   Ask for friction

   After each answer, ask for one follow-up that tests ownership, trade-offs, evidence, or result. Good interview prep AI should make the answer harder, not smoother.

5. 05

   Rewrite by hand

   Use the feedback to shorten setup, add missing context, or remove unsupported claims. Keep your own phrasing. If AI adds a metric, a stronger result, or a cleaner lesson you did not provide, delete it.

Copy this prep brief before prompting AI

Fill this in for one story at a time. A thinner brief with real facts is more useful than a long prompt with invented polish.

The prompt to use

Act as a realistic interviewer for this role: [target role].

Use the requirements below to ask interview questions:
[requirements to prove]

I want to practice from this real work moment:
[work moment, my action, result, constraint]

Rules:
1. Ask one question at a time.
2. Wait for my answer before giving feedback.
3. Ask one follow-up that probes detail, ownership, trade-offs, or result.
4. Tell me what sounded specific and what sounded vague.
5. Do not write the answer for me.
6. Do not add facts, metrics, achievements, or examples I did not provide.

Weak prompt vs stronger prompt

The weak version asks AI to perform. The stronger version gives it material to test. You may still answer poorly the first time. That is useful. It shows you which details are missing before a human interviewer finds them.

What feedback to trust

AI feedback can sound more certain than it deserves. Treat it like a sharp draft note, not a verdict.

This is where many candidates lose their voice. The tool tries to help by sanding down uncertainty. But interviews are often about judgment under constraints. A specific answer with a real trade-off is stronger than a spotless answer that sounds borrowed.

Use STAR after the facts, not before

So let AI help you check the shape:

Boundaries before the real interview

There is a practical line between preparation and misrepresentation. Using AI before the interview to generate practice questions, rehearse answers, and critique clarity is normal prep. Using undisclosed live prompts during an interview is different, especially if the employer expects unaided answers.

Where Koru fits

Koru's point of view is that AI interview prep starts before you open the prompt box. The hard part is not asking AI for questions. The hard part is remembering enough real work to answer well.

A weekly career record gives you better inputs: projects, constraints, decisions, feedback, numbers, and awkward trade-offs captured while they are still fresh. Then AI has something useful to pressure-test.

That is the reason this workflow starts with evidence. If you bring vague memories, AI will usually make them smoother. If you bring real work, it can help you turn that work into answers that still sound like you.