Ethyx v1.4: Compare, privacy wedge, and cognitive format

v1.4 adds three pillars on top of the v1.3 line: compare chat, a clearer privacy wedge, and cognitive format—information architecture preferences for replies that are easier to follow.

Multi-model compare chat

Run the same text prompt against 2–3 catalog models in parallel, stream side-by-side responses, pick a winner, and continue the thread on that model. Compare is available in the text composer on desktop (hidden in image mode). Each variant counts toward usage limits.

Privacy and security wedge

• Threat-model documentation for operators — what providers always receive vs what Ethyx pseudonymizes or withholds (docs/operations/provider-privacy-threat-model.md).
• Security settings panel — plain-language summary of encryption at rest, identity pseudonymization, and provider retention opt-outs, plus a link to our privacy policy.
• Account context tool opt-out — disable get_user_account_context when you do not want the assistant fetching a subscription/profile snapshot via tool call.
• Marketing copy — features and hero text now lead with pseudonymized identity at providers alongside encryption at rest (honest limits: live inference is not end-to-end encrypted).

Cognitive format (neurodivergency MVP)

A new chat-style dimension shapes how answers are structured, not tone or length. Pick a default or override per conversation or project:

• Standard — scannable markdown (default)
• Timeline — one merged chronological plan (e.g. combine multiple recipes into a single cooking timeline)
• Checklist — actionable checkbox items grouped by phase
• Step-by-step — numbered sequence with one active step emphasized
• Single screen — ultra-compact layout for one scannable view

Set defaults in Account → Preferences, override in the composer options panel, or scope per project.

Thank you

v1.4 is the release that makes Ethyx’s privacy story visible in-product and adds executive-function-friendly reply formats. Questions or feedback? Contact us.