IAT reads the question — from your screen, from the conversation, or from your own voice — and puts a structured answer on screen in about two seconds. You stay in the conversation.
30 minutes of credit when you sign up. No card.
| HashMap | ConcurrentHashMap | |
|---|---|---|
| Thread-safe | No | Yes, lock-striped |
| Null keys | One allowed | Not allowed |
Most of an interview is spoken, some of it is on screen, and sometimes you want to ask something yourself. All three end up in the same place.
One key takes a screenshot and answers whatever question is on it — a coding pad, a shared document, a take-home task.
Listens to the call's audio and transcribes it as it goes. Press Answer and it works out which part was the question.
Ask out loud and get an answer back — useful when you want to think something through rather than wait to be asked.
Speech is transcribed continuously on your Mac. Nothing is sent anywhere until you ask for an answer — no dispatching on a pause, no answering a remark that was never a question.
Everything heard since the last press goes over as one question. The model separates the question from the small talk around it, and shows you both — so you can see what it thought you asked.
A sentence you can say out loud, then the detail: bullets with the why, tables for comparisons, diagrams for anything with a shape, and code with its complexity. Not a wall of prose.
↻ re-asks the same question for a different angle. ⌥F asks a follow-up on the answer you're looking at, keeping the context.
First words on screen in roughly two seconds, streaming as they're written. You read while it's still finishing.
Transcription runs locally with a bundled speech model. Audio is never uploaded — only the question text, once you ask for an answer.
A floating panel that never takes focus, excluded from screen capture, adjustable transparency, and one key to hide it instantly.
Add the company, the role, the job description and your résumé once. Answers reference your actual experience instead of speaking generically.
Tell it how you want answers — brute force then optimised, comments on every non-obvious line — and it follows that over its own defaults.
Each question, answer and follow-up saved and searchable afterwards, exportable as Markdown. Good practice material.
No subscription and no seats. Your balance is an amount of time, and it is spent only while a session is actually running.
Start a session before the call, end it after. The time it ran comes off your balance — nothing is charged while you are not in one.
Every new account starts with 30 minutes, no card. Enough for a real interview before deciding anything.
If your balance runs out during a session, it keeps going for a grace period rather than stopping on you halfway through a question.
Nothing leaves your Mac. Speech is transcribed locally by a model bundled with the app; no audio is uploaded, stored, or sent to us. When you press Answer, the transcribed text of the question is sent to get an answer — the recording never is.
It captures your Mac's audio output, so it works with anything that plays sound — Zoom, Teams, Meet, a phone on speaker. It doesn't integrate with any of them, which is why there's nothing to install on the other side.
The overlay asks macOS to exclude it from screen capture, and it never takes keyboard focus. We've measured this working on macOS 15, but no application can promise it for every version and every conferencing tool — so test it on your own setup with the built-in self-test before relying on it.
No. Answers come through the Claude Code CLI you already have installed, on your existing subscription.
Your email, your credit balance, and how long each session ran. Your résumé, interview context and session transcripts stay on your Mac. Questions are only shared beyond that if you explicitly turn it on in your account.
Not yet. macOS 14 or later for now; a Windows version is planned.
30 minutes of credit, no card, about a minute to set up.