AI Analysis
AI Analysis
A background process that transforms finished entries into connected, encrypted memory.
What is AI Analysis?
AI Analysis is a background process that turns your entries into structured memory.
It:
- Identifies meaningful moments in your writing
- Detects recurring threads
- Connects moments to related threads across time
Instead of remaining separate entries, your writing becomes a structured network of connected moments and threads.
When does AI Analysis happen?
AI Analysis runs when you mark an entry as finished.
It runs in the background while the app is open.
If you close the app, analysis stops.
It resumes the next time you open it.
Privacy and security
DeepJournal combines selective field-level E2EE for configured synced fields with Confidential AI.
Entry content is encrypted before synchronization. Operational metadata used for synchronization remains visible as described in End-to-end encrypted sync.
AI Analysis runs only while the app is open. Extraction payloads are held in memory between pipeline stages rather than persisted for later processing.
AI request and response bodies are encrypted between the desktop client and an attested trusted execution environment. The proxy can still see authentication, routing, timing, model, rate-limit, and usage metadata outside the encrypted body.
In practice, this means:
- The application proxy routes encrypted prompt and response bodies rather than parsing journal text
- Journal content is not used to train models under DeepJournal's current policy
- Plaintext exists inside the verified enclave during inference and inside the unlocked desktop application
This model reduces plaintext exposure without claiming that all metadata is hidden or that an unlocked compromised device is protected.