A local-first platform for building and running private AI apps on your computer.
Visit AppJuice · User guide · Browse the Juicery
With AI becoming part of everyday coding for software engineers, I found myself spending more and more time building apps for myself and my family, working with private personal data on my own computer. But I kept running into the same chores: configuring AI providers, storing credentials, and figuring out how to deploy each app. Even a small change meant opening a terminal, switching to the project directory, starting a coding agent, then deploying again. AI had made writing code much faster, but the whole process still felt fiddly.
I wanted one place where I could build an app, use it right away, and change it without friction—with changes taking effect as I made them, no deployment needed. I wanted those apps to run on my own computer and keep their data there, so I could work with private information comfortably, without having to upload it to a third-party server just to run an app.
I wanted a shared set of AI providers and credentials I could manage once for all my apps. I also wanted to choose whichever providers I liked and make AI features and agents easy to add.
That’s why I built AppJuice.
What makes AppJuice different
- Build apps that run from the first line of code. AppJuice provides the app runtime, so your app runs locally as it is built, with its own interface and persistent data—no deployment needed.
- Use and tweak. Run the app, then summon the builder agent in Tweak mode to make changes as your needs evolve. See changes immediately in the running app.
- Agents inside apps. Give a workflow its own agents, tools, skills and context. AppJuice provides a complete Agent SDK and ready-to-use UI components for conversations, model selection, approvals, attachments and generated artifacts. Code-first multi-agent orchestration lets you define agent flows and coordination directly in code.
- Bring your own AI. Choose from 50+ AI provider integrations, including OpenAI, Anthropic, Google, DeepSeek and MiniMax, or use supported local models through Ollama or LM Studio. Configure your preferred providers and choose models for each capability.
- Built-in AI capabilities. Text generation, image understanding and generation, transcription, speech synthesis, music and video generation, web search and more. Configure providers once; apps and agents can use the corresponding capabilities through the platform’s shared APIs and tools.
- Share and remix. Install apps from the Juicery or share an app package separately from private runtime data and provider credentials.
- Access your apps remotely. Use Tailscale to securely access apps running on your computer from your other devices.
- Publish with a simple request. Ask the builder agent to deploy a static site or an app with a backend as a public website (with a cloud host configured)—all in one sentence.
Apps run locally by default. Cloud AI requests send the required context to your selected provider. File and tool access is governed by platform permissions and your approvals.
Try it
Available in Preview for macOS 13.5+, on Apple Silicon and Intel. See the AppJuice website for current downloads.