WeClone is a tool for creating AI digital avatars by fine-tuning language models with your chat history.
The tool addresses the problem of capturing someone's unique communication style and personality in a conversational AI system. It works by extracting chat data from messaging platforms, preprocessing that data to filter sensitive information, fine-tuning a language model on the processed conversations, and then deploying the resulting model as a chatbot. The approach keeps data processing and model training localized for privacy control, and supports multimodal input including images alongside text from chat logs.
WeClone suits anyone wanting to create a personalized chatbot that mimics their own conversational patterns or those of someone else. It works best for users with substantial chat history on supported platforms and who prioritize data privacy through local processing. The tool currently supports Telegram as a data source with image modal support, while WhatsApp, Discord, and Slack integrations are in development. The project includes end-to-end functionality spanning data export, preprocessing, model training, and deployment, distinguishing it from solutions that handle only individual stages of this pipeline.
The project receives issues from both core maintainers and external users, indicating real adoption beyond the development team without an overwhelming support burden. Maintainers typically respond to new issues and pull requests within hours. Work tracked in the issue system centers on bug fixes, feature requests, and user questions.
