Overview
Key Features
How It Works
Tech Stack
Live Demo
Contact
Loraly is a personal project I built to preserve a person's unique way of writing. By learning from chat messages, the AI captures a loved one's style of conversation—their humor, their advice, their warmth—allowing you to continue chatting with a text-based twin long after they are gone.
By leveraging fine-tuned LoRA (Low-Rank Adaptation) models on personal chat data, Loraly delivers a high level of personalization and authenticity, moving beyond generic chatbots to create a true digital echo of a human being.
- Authentic AI Twins: Create precise digital replicas of any individual's conversational style using their personal chat history.
- Legacy Preservation: Offers a meaningful way to preserve the voice and personality of departed loved ones.
- Style-Consistent Replies: Generates responses that maintain the unique tone and logic of the original person.
- Secure & Private: Implements security measures for sensitive chat data uploads and storage.
- Intuitive Chat Interface: Clean, responsive GUI for seamless real-time interaction with AI twins.
- Sign Up: Users create a secure account.
- Upload Data: Upload exported one-on-one Whatsapp chat logs.
- Train Twin: The system fine-tunes an AI model on the uploaded conversational data.
- Interact: Engage with the AI twin through an intuitive chat interface and experience authentic, style-consistent responses.
- Frontend: React
- Backend & Database: Supabase (Authentication, PostgreSQL)
- AI Training & Inference:
- Training: LoRA (Low-Rank Adaptation) fine-tuning
- Infrastructure: GPU-accelerated cloud training via RunPod
- Deployment: GPU-optimized inference hosting via Modal
You can explore Loraly at the site I created:
https://loralydemo.netlify.app
For any inquiries about the project, technical questions, collaboration opportunities, or feedback, please feel free to reach out:
Ethan Brook
Email: ethanbrook314@gmail.com