Real-time answers while the interview is still happening.

It's a desktop application for Windows that provides real-time AI assistance during job interviews. It reads the screen or window you choose, optionally listens to the call, and streams an answer into a small overlay window while the interview is still happening.

BYOK stands for bring your own key. Instead of bundling AI usage into the subscription, you connect your own OpenAI or Anthropic API key, and each request goes straight from your machine to that provider — billed at their rate, with no markup added.

It's priced as a subscription — $8 a month, or $5 a month billed annually — with a $80 one-time option if you'd rather own it outright. Every plan has the same feature set; the only choice is how long you're paying for.

Everything the subscription has — continuous mode, meeting audio context, capture exclusion, your own provider key — plus every future update and the macOS build once it ships, all for one payment with no renewal.

So your AI usage is billed at the provider's own rate, with nothing added on top. The $8 a month or $80 once pays for the app itself; OpenAI or Anthropic bills you directly for the tokens you use.

That's set by your provider, not by us, and depends on the model you pick and how often you ask. A vision request on a fast, low-cost model is typically a small fraction of a cent to a few cents; check your provider's own pricing page for exact rates, and you can switch to a cheaper model any time to control spend.

OpenAI and Anthropic for the vision model that reads your screen. Meeting audio context is always transcribed by OpenAI Whisper, regardless of which provider you pick for analysis, since Anthropic has no audio-input API yet.

Yes. It reads the pixels of whatever window or display you select and listens to your microphone and system audio — there's no Zoom-specific integration to break when Zoom updates.

Yes, the same way as Zoom: screen and audio, not an API integration, so it works whether the call is running in the desktop app or a browser tab.

Yes. Any call software running in a window or on a display you can select works, including Google Meet, CoderPad, and HackerRank.

On Windows 10 build 2004 or later, this app's own window carries the OS's exclude-from-capture flag, so a screen share composites it out — the interviewer sees only what you intentionally share. It's a best-effort OS mechanism, not a guarantee, and it only ever covers this app's own window.

The same exclude-from-capture flag applies to local recorders that use the standard Windows capture APIs, not just live screen shares. Support depends on the recorder's own capture method, and it's still best-effort — it won't hide the app from a camera pointed at your monitor.

No sign-up inside the app. You install it and unlock it with the license key from your purchase — there's no username, password, or profile to create.

No frame or audio chunk is ever written to disk. The rolling context and any meeting transcript are text-only, held in memory, bounded in length, and cleared when you quit or clear them yourself.

Not yet. Windows is where capture exclusion is fully supported today; a macOS build is in progress and will ship into every plan, including lifetime, at no extra cost.

No — a ChatGPT Plus or Pro subscription is separate from an OpenAI API key. You'll need a key from platform.openai.com, which bills per token rather than as a monthly plan, even if you already pay for ChatGPT.

Answers stream token by token as the model writes them, so text starts appearing well before the reply is finished. The wait before the first token depends on the model you picked and your connection.