Introduction
Rocky is a voice-first AI Agent app for iPhone and iPad. OpenRocky is the open-source project behind it.
Rocky is not a mobile chat shell, nor a ported Linux container on a phone. It puts voice conversation as the primary entry point, organizing voice interaction, task execution, system bridging, and result review into an agent experience designed for iOS and iPadOS.
Why Voice-First?
Most AI apps on mobile are text-based chat interfaces. Rocky takes a fundamentally different approach:
- Voice is the main entry — Talk to Rocky like talking to a person. No typing needed.
- Text is a supplement — Available when you need precise input like code or URLs.
- Tasks, not just chat — Rocky doesn't just respond; it executes tasks and produces real results.
Core Principles
- Voice as primary input — The home screen is a voice interface, not a chat list.
- Task execution — Powered by ROS, the internal runtime that plans and runs tasks.
- iPhone & iPad native — Built with SwiftUI, using iOS native bridges and on-device execution.
- Open source — Transparent development, community-driven.
Platform Support
| Platform | Status |
|---|---|
| iOS (iPhone) | Supported |
| iPadOS (iPad) | Supported |
| macOS | Not planned |
| Android | Not planned |
Current Status
OpenRocky is in the documentation-first + early prototype phase. The repository contains:
- Product positioning, iOS architecture docs, and reference materials
- An early SwiftUI iOS prototype for direction validation
The prototype has verified three key integrations:
- SwiftOpenAI — Model access and OpenAI Realtime session bridging
- LanguageModelChatUI — Chat detail view mounted in the iOS prototype
- ios_system — Controlled local execution layer for runtime environment
Standard Naming
Rocky is the app. OpenRocky is the open-source project behind it.