I built RunWhale so an idea can become a small app without leaving your phone. Describe what you want, let the agent edit the project, and open the preview on the same device.
RunWhale is open source and built on DeepSeek Harness, Expo, and React Native.
What the workflow looks like
- Create a project or import a repository.
- Describe a change and let the agent work on the files.
- Review the changes in Git.
- Open the web or native preview, then keep iterating.
The part I wanted to bring together is that edit–review–preview loop on a phone. It is useful when you want to try a small idea and your laptop is not nearby.
What runs on the phone
Project files, Node.js, Git, development tools, and the preview server run on the device. Model inference uses a remote provider that you configure, so AI requests still need a network connection.
The current focus is Web and Expo projects. It is not a general Linux environment, and arbitrary native dependencies will not all work.
Source and setup: github.com/zhiqingchen/RunWhale
The original code is Apache-2.0 licensed. The ready-to-use iOS app is a paid download; model service costs are separate.
I would love feedback from people building with Expo or coding agents: what would make a phone useful for your own development workflow?
Top comments (0)