When we started building Nova, we faced a fundamental tension: to be truly useful, an AI assistant needs deep access to your digital life. But that access requires trust.
The Trust Problem
Think about what an AI assistant needs to know to be helpful: - Your calendar and schedule - Your emails and messages - Your smart home devices - Your preferences and habits
This is sensitive data. And the track record of tech companies with personal data isn't great.
Our Architecture
We designed Nova's architecture around a simple principle: minimize data exposure.
**Ephemeral Processing**: Your data is processed in memory and never written to disk. Once Nova completes an action, the data is gone.
**Zero-Knowledge Where Possible**: For many integrations, Nova uses delegated authentication. We never see your passwords.
**Local First**: Many operations happen on-device when possible, reducing what needs to travel to our servers.
What We've Built
Here's what's in place today: - AES-256-GCM encryption for all integration tokens - OAuth2 for secure third-party access - Strict Content Security Policy and HSTS headers - No training on your conversations — ever
Your Data, Your Control
You can export all your data anytime. You can delete everything with one click. No retention periods, no gotchas.
This isn't just a feature — it's a promise.