Privacy Policy
The Short Version
We don't collect, store, transmit, or sell your data. Sansara runs 100% on your device. What you write, say, and feel stays on your phone.
1. Zero Cloud Architecture
Sansara is structurally incapable of sending your personal data to our servers. All natural language processing, multimodal inference (voice prosody, facial expression analysis), and memory embeddings are generated and stored locally on your device hardware using hardware-accelerated backends like CoreML or Qualcomm AI Engine.
Because we don't have servers that process your inputs, we don't have servers that can be breached, subpoenaed, or monetized.
2. Local Data Storage & Encryption
Your journal entries, voice recordings, drawings, and all resulting emotional embeddings are stored in an on-device database. On Android, storage is AES-256 encrypted at rest. Full encryption on iOS is in active development and will ship in a future update.
Biometric telemetry and multimodal inputs (like camera or microphone data) are processed in real-time. Raw pixel data and audio recordings are purged from RAM instantly after embedding extraction. Only the resulting anonymous mathematical vectors survive.
3. On-Device Learning (LoRA)
Sansara personalizes itself to you over time through Low-Rank Adaptation (LoRA). On Android, this training runs natively on your device. iOS personalization is in active development. In both cases, the resulting model weights never leave your phone, and you can wipe your data and personalized adapter at any time from the app settings.
4. Waitlist & Email Communications
If you sign up for our waitlist, your email address is stored securely on a minimal infrastructure just to notify you on launch day. After the launch announcement, we will delete your email address. We will never sell your contact information.
5. Contact Us
If you have any questions about this Privacy Policy, please contact us at info@sansara.app.