Transparency

How Amorelle uses AI.

Amorelle is AI-native — but every AI feature is opt-in, per feature. We tell you exactly what's sent and to whom. We never train models on your data. We never sell or share it. Below is every AI feature in the app and what it sees.

Daily Spark

Groq

A short personalized question every morning, tuned to recent themes you've been talking about.

What the AI sees: Topic tags from the past week (no message text).

Letter Composer

Groq

Co-writer for love letters. Pick a tone, get a starting sketch, polish your draft.

What the AI sees: Only what you type into the helper. Never the recipient's chats.

Vow Builder

Groq

Drafts your wedding vows from the structured prompts you fill in.

What the AI sees: Only what you provide. Won't invent details about your partner.

Mediator

Claude (safety tier)

When a conversation gets tough, both write privately to a neutral AI that summarizes back to both of you.

What the AI sees: Each partner's private input for that session. Not chat history.

AI Counseling

Groq → Claude on crisis-flagged content

Warm, judgment-free relationship guidance whenever you need.

What the AI sees: Only the messages you send to the counseling thread.

Weekly Story

Groq

Sunday recap — favorite photo, longest streak day, the moment that made you both laugh.

What the AI sees: Aggregated metrics: photo IDs, streak count, message engagement scores. No message text.

Smart Gift Suggestions

Groq

Hints based on partner's wishlist + recent topics — only when both have opted in.

What the AI sees: Distilled topic summary, not raw messages.

Date Night Planner

Groq

Three options based on your location, time, budget — long-distance friendly.

What the AI sees: Inputs you provide for the planner. Not chat history.

Translation in chat

Groq

Inline translate-this pill for messages in your partner's language.

What the AI sees: Only the message you tap to translate.

Crisis safety

When the AI detects language suggesting self-harm or domestic violence, the response is routed to Claude (a safer model) and a curated list of local crisis resources is shown above the response. This isn't a substitute for professional help — it's a bridge to it.