FULL PROTOCOL
BREAKDOWN

The engine hooks directly into Windows' DirectX/DXGI rendering pipeline. Screen capture APIs — used by Zoom, Teams, proctoring software, and remote desktop tools — receive a blank or decoy frame. Your actual workspace renders exclusively on your physical display.

• Hooks BitBlt, PrintWindow, and DXGI capture chains
• Renders exclusively to physical display output
• Zero visual artifacts — observers see a clean desktop
• Operates at driver level — undetectable by userspace monitors
• Compatible with all major proctoring: Examity, Proctorio, ExamSoft, HonorLock

Fire a single query and receive responses from GPT-4o, Claude 3.5 Sonnet, and Gemini Pro simultaneously. The fastest, most accurate response surfaces first. BYOK (Bring Your Own Keys) ensures zero data passes through our infrastructure.

• Simultaneous routing to 3+ LLMs
• Client-side API key injection — we never see your keys
• Auto-selects optimal response by speed and relevance
• Context-aware prompt engineering built in
• Supports custom model endpoints and local LLMs

Capture questions via hotkey while staying focused on your primary task. The system queues captures, processes OCR, routes to AI, and stacks responses — all in the background. Review the full response stack when ready.

• Global hotkey capture — works across all applications
• Background OCR processing with high-accuracy extraction
• Asynchronous AI resolution queue
• Response stack with timestamped history
• Kill-switch hotkey instantly clears all traces

Real-time transcription of audio streams with instant AI-generated responses. Designed for live video calls — the interface is completely hidden from screen sharing while feeding you contextual answers in real-time.

• Live audio transcription via system audio capture
• Real-time AI response generation
• Hidden overlay — invisible to screen share
• Context memory across conversation turns
• Works with Zoom, Teams, Google Meet, and any WebRTC call

The executable makes zero outbound connections to our servers. All AI calls go directly from your machine to the model provider using your own API keys. No analytics, no tracking, no phone-home behavior. Verify it yourself with Wireshark.

• No outbound telemetry or analytics
• All API calls are direct client → provider
• No auto-update pings or license phone-home
• Offline-capable for cached operations
• Open to network traffic audit