What InterviewShadow does during an interview — and what it won't

Most descriptions of interview tools are vague about the thing you actually want to know: what appears on screen, when, and where it came from. Here is the plain version.

What it does

Short cues, only when you ask. Nothing appears unprompted. You ask, and you get a few lines — not a paragraph to read out. The format is deliberate: a paragraph gets read aloud and sounds like it, while three bullets get used.

Drawn only from your resume, your notes and the job description. This is the part that matters most. The cues are retrieved from material you supplied, not generated from nothing. When you are asked about a project, what comes back is the project you actually worked on, with the details you wrote down weeks ago and had forgotten.

It flags when your profile lacks the detail. If you ask about something your material does not cover, it says so rather than filling the gap. That sounds like a limitation, and it is the feature that makes the rest trustworthy — if it invented an answer once, you could never rely on any of them.

It shows how much time you have used. The meter is visible, and pauses stop the clock. You are not guessing whether you are about to run out.

You can pause, recover, or switch to typing. Interviews are not tidy. The controls assume that.

What it will not do

It will not speak, type or send anything for you. There is no automation sitting between you and the interviewer. Every word delivered is yours, out loud, in your voice.

It will not invent achievements you did not supply. If it is not in your material, it does not exist. This rules out the failure mode that ends interviews — a claim you cannot elaborate on when asked the obvious follow-up.

It will not solve live coding or technical tests. Those are assessments of whether you can do a thing. Sitting in the middle of one is not a grey area, and the product does not.

Why the limits are the product

A tool that would write an answer for you is easy to build and useless to rely on, because the interview does not stop when the answer does. The next question is always "tell me more about that", and the person who read out a generated paragraph has nowhere to go.

Cues from your own material do not have that problem. You can elaborate, because you were there.

The honest description of what this does is a memory aid, not a ghostwriter. The value is not that it knows things you don't. It is that under pressure you cannot retrieve things you do know, and it can.

Where it runs

Alongside your existing call — you do not move the interview anywhere, and the other side sees the same thing they would have seen anyway.

Where it does not help

Worth saying plainly. It will not help you if:

• the interview is a live technical assessment
• you have not given it any material to work from
• you have not rehearsed once, so you are reading cues cold

That last one catches people out. The cues are prompts, not scripts, and prompts only work if you have said the words at least once before. That is what rehearsal time is for, and why it is sold separately.

---

If the limits above are narrower than you expected, that is deliberate. A narrower tool you can trust in the moment beats a broader one you cannot.