Stealth Mode FAQ

Everything you need to know about PrepPilot's real-time AI interview assistant. 20 questions answered.

General

What is PrepPilot Stealth Mode?

Stealth Mode is a floating overlay that sits on top of your screen during live interviews or calls. It listens to the conversation through real-time audio transcription and generates AI-powered suggested responses that only you can see. Think of it as a silent co-pilot for any live conversation.

How does Stealth Mode work technically?

Stealth Mode captures your system audio, streams it to Deepgram for real-time transcription, then sends the transcribed text to your configured AI model (Claude, GPT, or Gemini). The model generates a short, natural-sounding response that appears in the floating overlay within 1-2 seconds.

What AI models are available?

PrepPilot supports six AI models: Claude Sonnet 4.6, Claude Opus 4.6, Claude Haiku 4.5, GPT-5.3, and Google Gemini. You can switch between them at any time. Faster models like Haiku are ideal for real-time use, while more powerful models like Opus provide deeper responses.

Is PrepPilot open source?

Yes. PrepPilot is fully open source on GitHub at github.com/opeoluwagiwa/PrepPilot. You can review the code, verify privacy claims, audit the transcription pipeline, and contribute to the project.

Technical

Which video conferencing platforms does it work with?

Stealth Mode works with any platform that outputs system audio, including Zoom, Google Meet, Microsoft Teams, Webex, Slack Huddles, Discord, and even phone calls routed through your computer speakers. If you can hear it, Stealth Mode can transcribe it.

Which operating systems are supported?

Stealth Mode runs on Windows 10/11, macOS (Intel and Apple Silicon), and Linux. The desktop app is built with Tauri for native performance with minimal resource usage across all platforms.

How fast are the AI responses?

Responses typically appear within 1-2 seconds after the interviewer finishes speaking. Speed depends on your internet connection and the AI model selected. Claude Haiku and GPT-4o-mini deliver the fastest response times for real-time use.

Can I resize or reposition the overlay?

Yes. The floating overlay is fully draggable and resizable. Position it anywhere on your screen, adjust opacity, and toggle visibility with a keyboard shortcut. You can configure it to be as subtle or prominent as you prefer.

Does it work for phone interviews?

Yes. If you take the call on your computer via speakerphone, a VoIP app, or by routing phone audio through your speakers, Stealth Mode captures and transcribes the audio in real time.

Is there a mobile version?

Currently Stealth Mode is desktop-only (Windows, macOS, Linux) because it requires system-level audio capture. A mobile companion experience is being explored for a future release.

Privacy & Detection

Is Stealth Mode detectable during screen sharing?

The overlay is designed to not appear in screen-share feeds on most video conferencing platforms. It behaves like a system-level floating window that conferencing tools typically exclude from capture. Results may vary by platform and OS version.

Does Stealth Mode record my interviews?

No. Stealth Mode does not save, store, or record any audio. Audio is streamed in real time to Deepgram for transcription only. The transcription is processed locally and no recordings are kept on any server.

Is my data safe and private?

Yes. Audio goes to Deepgram for transcription only and is not stored. AI responses are generated via your own API key. PrepPilot does not store conversation data on its servers. The entire codebase is open source for verification.

Can my interviewer hear any audio from Stealth Mode?

No. Stealth Mode is purely visual. It displays text responses in the overlay and produces no audio output whatsoever. There is nothing for the interviewer to hear.

Compatibility & Features

What languages does Stealth Mode support?

Stealth Mode supports transcription in over 30 languages through Deepgram, including English, Spanish, French, German, Portuguese, Japanese, Korean, Mandarin, Hindi, and Arabic. AI responses can be generated in any language the underlying model supports.

Does Stealth Mode work during panel interviews?

Yes. Stealth Mode transcribes all speakers and generates responses regardless of how many people are on the call. It handles multi-speaker conversations effectively, identifying questions directed at you.

Can I customize the AI responses?

Absolutely. Adjust the system prompt, set your experience level, specify the role, and choose between AI models. Pro users can save custom prompt templates for different interview types such as behavioral, technical, or case interviews.

Pricing & Support

Do I need my own API keys?

Not to start. Every account gets 50 free credits that work without any API key. After that, upgrade to Pro ($19/month) for unlimited credits or bring your own keys from Anthropic, OpenAI, or Google at their standard API rates.

How many interviews can I do with 50 free credits?

On average, 50 credits cover 3-5 full-length interviews of 30-45 minutes each. Each AI-generated response uses approximately one credit. Shorter interviews or focused phone screens use fewer credits.

How do I get support if something goes wrong?

Reach the support team at support@mypreppilotapp.com or open a GitHub issue. Pro and Enterprise users receive priority support with faster response times. The community is also active on GitHub discussions.

Try Stealth Mode Free

50 free credits. No credit card. Windows + macOS.

Download PrepPilot