Natively-AI-assistant/natively-cluely-ai-assistant

What it solves

Natively is an AI-powered interview copilot and meeting assistant designed to provide real-time, undetectable assistance during high-stakes meetings and coding interviews. It eliminates the need for expensive subscriptions to similar tools while ensuring user privacy by storing data locally rather than on cloud servers.

How it works

The tool operates as a native application for macOS and Windows that captures audio and screen content in real-time. It uses a Rust-based audio capture system for ultra-low latency (<500ms) and on-device speech-to-text (STT) using optimized ONNX models (such as Whisper and Moonshine). It integrates with various LLMs (GPT, Claude, Gemini, Llama) via a bring-your-own-key (BYOK) model or runs entirely offline using Ollama.

Key technical features include:

• Dual-Channel Intelligence: Separates system audio (the interviewer) from the user's microphone.
• Stealth Mode: Hides the application from the dock and disguises processes to avoid detection by proctoring software.
• Local RAG: Uses SQLite vector search to allow users to query past meeting history.
• Browser Companion: A Chrome extension that pulls current webpage content into the AI's context.

Who it’s for

Job seekers preparing for technical or behavioral interviews, professionals in sales or recruiting, and students attending lectures or seminars who need real-time transcription and AI-generated guidance.

Highlights

• Privacy-First: Local data storage and source-available license to prevent data breaches.
• On-Device STT: Local Whisper models for zero cloud fees and increased privacy.
• Versatile LLM Support: Compatible with any major LLM or fully offline via Ollama.
• Context-Aware: Supports reference files (PDF, DOCX, TXT), custom personas, and a rolling memory window of the conversation.
• Undetectable: Hardened stealth mode for use during monitored interviews.