Refly

What is Refly?

Refly is the world’s first open-source Vibe Workflow platform, providing users with powerful multi-threaded dialogue and content creation capabilities. It supports multi-threaded dialogue, allowing users to freely switch between different sessions to achieve complex task workflows. Refly integrates 13+ mainstream language models, supports multimodal file processing, covers various text and image formats, and can efficiently process and analyze data. Refly features an AI-driven skill system, supporting intelligent web search and knowledge retrieval functions to help users quickly obtain information. Its knowledge base engine can integrate multi-source heterogeneous data to build personalized knowledge systems. It supports one-click content collection from mainstream platforms, facilitating knowledge management and creation for users.

Refly’s main functions

• Multi-threaded dialogue system : Supports parallel management of multiple independent session contexts, allowing users to switch freely between different dialogue nodes, realizing complex agentic workflows and breaking through the limitations of traditional dialogue models.
• Multi-model integration framework : Integrates 13+ mainstream language models, supports mixed model scheduling and parallel processing, and has a flexible model switching mechanism and a unified dialogue interface.
• Multimodal processing capabilities : Supports 7+ file formats (such as PDF, DOCX, TXT, etc.) and mainstream image formats (such as PNG, JPG, etc.), and can perform intelligent batch processing.
• AI-driven skills system : Integrates capabilities such as Perplexity AI and Stanford Storm, providing functions such as intelligent full-network search, knowledge retrieval, question rewriting and recommendation.
• Context Management System : It can accurately build temporary knowledge bases, flexibly select nodes as contexts, and realize multi-dimensional context association.
• Knowledge base engine : Supports importing heterogeneous data from multiple sources, performs semantic retrieval based on the RAG architecture, and can intelligently construct knowledge graphs.
• Intelligent content collection : Supports one-click collection of content from mainstream platforms (such as Github, Medium, etc.) and automatically performs knowledge classification and tagging.

How to use Refly

• Registration and Login : Visit the Refly website, click “Get Started”, and log in with your Google or GitHub account.
• Create a canvas : After logging in, select the “Create Canvas” option on the main interface to start a new creation process.
• Engage in dialogue : Have multi-threaded conversations with AI on the canvas, inputting ideas, thoughts, or questions to get instant feedback and suggestions.
• Editing and modification : Based on AI feedback and suggestions, edit and modify the content to improve the creation.
• Export and share : After completing your creation, you can export the content as a document or share it with others for collaboration and review.
• Deployment method
• Cloud version : Access Refly Cloud to use all features without configuration.
• Docker self-deployment : Ensures the machine meets the minimum system requirements (CPU ≥ 2 cores, memory ≥ 4GB) and allows for rapid deployment via Docker.
• Local development : Refer to the CONTRIBUTING.md file to configure the local development environment.
• Enterprise Deployment : Contact the Refly team for private deployment solutions.
• Other functions
  • Import Data : Import multiple sources of data, such as papers and web pages, with one click, and quickly locate core reference content through AI intelligent search.
  • Save : Save any webpage content with one click to continuously build your personal knowledge base.
  • Research : Stimulate thinking through multi-threaded dialogue and intelligent context management to ensure precise research direction.
  • Writing : Use professional writing aids to freely choose contexts and create a personalized writing environment from conception to final draft.

Refly application scenarios

• Content creation : Provides a free canvas creation environment suitable for professional scenarios such as academic research and technical documents, and can quickly transform ideas into high-quality content.
• Knowledge Management : Through knowledge base integration and intelligent retrieval, we help users build a personalized knowledge system and create a “second brain”.
• Project planning : It can be used to create complex tasks such as detailed game project plans, and efficiently organizes and manages projects through multi-threaded dialogue and knowledge base support.