RecurseChat Product Information

RecurseChat is a powerful, fast, local-first AI client designed for offline use with local LLMs (such as ChatGPT-compatible and Claude-compatible models). It enables private, on-device AI conversations, multimodal chats, and seamless interaction with local documents while avoiding cloud data exposure. The app emphasizes security, speed, and customization, making it suitable for users who value privacy and performance on desktop platforms.


Key Capabilities

  • Local-first, offline-capable AI client with no mandatory cloud dependency
  • Chat with multiple AI models in one session
  • Multi-modal input: chat with images using an integrated LLaVA model
  • Chat with PDFs, Markdown, and text files locally with Retrieval Augmented Generation (RAG)
  • Full-text, ultra-fast search across thousands of messages
  • Import and resume ChatGPT history for continued local conversations
  • Customizable AI personality and appearance
  • Use your own GGUF models or models from Hugging Face; connect to custom OpenAI-compatible endpoints
  • Floating chat window (⌘ + ⇧ + Enter) always on top and accessible from anywhere

How It Works

  1. Load local models or connect to preferred local/remote endpoints.
  2. Interact via a fast, private chat interface that supports text and images.
  3. Use RAG with local embeddings and vector databases to search and reference your own documents without uploading data.
  4. Import existing ChatGPT history to pick up conversations where you left off, all locally.

Safety and Privacy Considerations

  • All data stays on your device; no mandatory cloud upload for normal operation when using local models.
  • Suitable for users who require private, offline AI experiences without external data transmission.

Core Features

  • Local-first AI client; offline capable
  • Chat with multiple models in one session
  • Multimodal chat: images via LLaVA model
  • Chat with PDFs, Markdown, and text locally
  • Retrieval Augmented Generation (RAG) using local embeddings and vector DBs
  • Fast full-text search across conversations
  • Import/continue ChatGPT history locally
  • Customize AI personality and appearance
  • Use GGUF models or models from Hugging Face; support for OpenAI-compatible endpoints
  • Floating, always-on-top chat window for quick access