Privacy Protocol
Last Updated: November 2025
1. The Zero-Knowledge Principle
UMBRA is built on a philosophy of absolute privacy. We are a "Stateless" application regarding your voice data. When you speak to Umbra, your audio is streamed to our inference partners (Groq/Anthropic) for immediate processing and is not stored on our servers. Once the session ends or the browser tab closes, the conversation context is incinerated. We do not train models on your data.
2. Data We Do Collect
- Account Information: We use Clerk for authentication. We store your email and user ID to manage your access.
- Payment Data: All payments are processed by Stripe. Umbra does not see or store your credit card details.
- Local Preferences: Your "Identity Context" (Founder/Fear/Bias) is stored locally in your browser's
localStorage. It is sent to the AI only when you speak.
3. Third-Party Processors
To provide the Umbra experience, ephemeral data is transmitted to:
- Groq: For ultra-low latency Speech-to-Text.
- Anthropic: For the Socratic Intelligence engine.
- ElevenLabs: For voice synthesis.
These partners are used strictly for inference. We do not opt-in to data sharing for their model training.
4. Contact
For privacy concerns or data deletion requests (regarding your account metadata), contact:
support@umbramind.com