Lakr233/FlowDown

https://flowdown.ai/

What it solves

FlowDown provides a private, high-performance interface for interacting with various AI models on Apple devices. It solves the need for a native, privacy-focused client that allows users to connect to their own self-hosted models or OpenAI-compatible services without the developer collecting any personal data or usage telemetry.

How it works

Built natively with Swift and UIKit, the app acts as a bridge between your device and AI models. You can connect to local models using MLX, self-hosted servers like Ollama, or any OpenAI-compatible API. The app handles the chat interface, including Markdown rendering and file attachments, while syncing your data securely via iCloud. It is designed for users who prefer to manually configure their own AI setups.

Who it’s for

This app is for power users who want full control over their AI experience, value data privacy, and prefer a native, smooth interface on their iPhone, iPad, or Mac. It is ideal for those who already use or are willing to configure local inference servers or specific AI model providers.

Highlights

  • Privacy-first design with zero data collection and no telemetry.
  • Native Apple performance using Metal and MLX for efficient on-device inference.
  • Universal compatibility with OpenAI-standard APIs and local model servers.
  • Advanced features including vision support, audio attachments, and web search.
  • Deep system integration with iCloud sync, Shortcuts, and Live Activity support.
  • Open-source codebase available for audit under the AGPL-3.0 license.

Related

  • Project
  • Project
  • Project
  • Project