A React and Vite web application for recording study sessions, lecture videos, or slide presentations, and interacting with an AI assistant that answers questions about the recorded content.
ReplayRAG Frontend is the client interface for the ReplayRAG platform. It lets users capture their screen and audio in real time, manage recording sessions, and query a local retrieval-augmented generation (RAG) assistant powered by Ollama (Llama 3.2). The assistant answers questions with references to exact timestamps in the recording.
- Screen and audio capture with start, pause, resume, and stop controls
- Session-based organization of recordings
- Chat interface to ask questions about a session's content
- Timestamp-referenced answers from the RAG assistant
- Persistent Q&A history per session
- React
- Vite
- JavaScript
- Node.js 18+
- The ReplayRAG Backend running locally
- Ollama installed and running with the
llama3.2model pulled
-
Install dependencies:
npm install -
Start the development server:
npm run dev -
Open the app at
http://localhost:5173. Ensure the backend is running athttp://localhost:8000.
- Create a new session by entering a title.
- Start recording and select the tab or window to capture, enabling audio sharing.
- Pause or resume the recording as needed; segments are processed in the background.
- Ask questions about the session content through the chat interface.
- Review past sessions and their Q&A history at any time.
- ReplayRAG Backend — FastAPI service handling transcription, keyframe extraction, and the RAG pipeline.
Not specified.