MemorisAI is an AI-powered personal knowledge assistant designed to provide personalized, context-aware conversations using persistent memory and Retrieval-Augmented Generation (RAG).
Users can create conversations, interact with an AI assistant, upload PDF documents, and ask questions based on their own knowledge base.
- User registration and JWT-based authentication
- Persistent multi-conversation chat system
- AI-powered conversations using Google Gemini
- Persistent user memory across conversations
- PDF document upload and management
- Semantic search across uploaded documents
- Retrieval-Augmented Generation (RAG)
- Context-aware responses based on conversation history, user memory, and documents
- Markdown-formatted AI responses
- Responsive and modern web interface
- React
- Vite
- React Router
- Axios
- Tailwind CSS
- React Hook Form
- Zod
- React Markdown
- React Hot Toast
- TanStack Query
- Lucide React
- Node.js
- Express.js
- MongoDB
- Mongoose
- JSON Web Tokens (JWT)
- Multer
- PDF Parse
- Google Gemini
- Gemini generative AI models
- Gemini embedding models
- Retrieval-Augmented Generation (RAG)
- Vector similarity search
MemorisAI combines three sources of context when generating responses:
- Conversation history for short-term context
- Persistent user memories for long-term personalization
- Relevant sections from uploaded documents using semantic vector search
Uploaded PDF documents are extracted, split into smaller chunks, converted into vector embeddings, and stored for later retrieval. When a user asks a question, the question is embedded and compared against stored document vectors to find the most relevant context. This context is then provided to the AI model when generating the response.
- Frontend: https://memorisai-1.onrender.com
- Backend API: https://memorisai.onrender.com