🍄 Control your LLM privacy blast radius
The Big Models 👽 already know too much about you.
🕷️ Do you really want them crawling your Google Drive, Dropbox, email, and Socials?
- 📱 Local on-device model by default.
- ☁️ Connect to self-hosted open weights models in the cloud.
-
Opt-in proprietary models with privacy controls like traffic anonymizations. - Open source under Apache 2.0 license.
🧠 Own your LLM chats and memories
The Big Models 👽 keep your memories for themselves, and who knows what they do with them?
📀 Take your chats and memories back into your own hands, and keep them private and portable.
- 📲 Local on-device storage by default.
- ☁️ Optionally store memories on connected cloud platforms.
🔗 Link your LLM to your data via connectors
The Big Models 👽 can link their models to your data. So can tf-llama.
- 📱 None of your linked data ever leaves your device by default.
- 🔒 Fine-grained controls on each connector—allow the LLM into only the Google Drive folders you choose.
- Extend capabilities with community contributed data connectors.
⚙️ Put your Llama in beast mode with mini-apps
Can the Big Models 👽 even do this yet?
- 🦙 Mini-apps integrate with your chosen LLM.
- 🔗 Mini-apps also have access to your connected data sources.
- Choose from a wide array of community contributed mini-apps.
- 💃🏻 Vibe-code your own mini-app like a boss.
Do more with LLM powered mini-apps
Mini-apps can be written manually and contributed via the GitHub repository. Or they can be vibe-coded and tested directly from within the app.
Next time you visit a new city, get a personalized tour guide experience generated by an LLM.
Need a quick translation? Wrap your preferred LLM in a nice interface with a language pair picker.
