Here are the main areas of the app and what lives in each.
The sidebar
- Your vaults, each with its colour. Clicking a vault opens it; if an unsent draft is sitting in the composer, the app warns you before switching (see Switching vaults).
- The conversations of the open vault.
- Preferences, at the bottom of the sidebar: interface language (French and English, applied live), AI providers and API keys, the local analysis module, telemetry.
The vault header
At the top of the window, for the open vault:
- Dictionary: the terms specific to this client that Trust6 always masks. See The custom dictionary.
- Audit log: the trace of everything that happened in the vault (analyses, sends, blocks, files), filterable by family and exportable as CSV. It also hosts the “Resolve a token” probe, the proof of isolation between vaults.
The composer
At the bottom, where you write:
- The paperclip attaches a file: the document is read locally and then accompanies your messages as context (see Attaching files).
- The Analyze button runs detection; Enter does too.
- The Send button only activates once the analysis is done and approved.
- The model selector picks which AI provider to talk to (see Choosing an AI model).
The analysis panel
It opens on the right when you analyze a text: counts by type, dismissible detection chips, the exact pseudonymized preview. It shrinks the chat without covering it. This is your review station before every send.
In the conversation thread
Every sent exchange carries its “Pseudonymized before sending” badge, the toggle between the original and the tokenized version, and access to the exact network payload. Nothing is cosmetic after the fact: everything can be verified.