Helpful, personal — and never all-knowing

How a reply is made

When you send a message, your words and a bounded amount of relevant context are processed to produce a reply. That context may include preferences you chose, patterns you confirmed and a short recent summary needed for continuity.

What it remembers

Behind 6 does not silently turn an AI guess into a fact about you. Inferred patterns are shown for confirmation before they are kept. You can inspect, correct or remove confirmed understanding.

• You confirm inferred patterns
• You can remove what it knows
• Long histories are not sent with every message
• Personalisation can be changed

Where it can fail

AI can misunderstand context and produce inaccurate, incomplete, repetitive, inappropriate or unexpected output. A confident tone does not make a statement true. Important health, safety, legal or financial information should be checked with a qualified or authoritative source.

Rules the AI cannot change

Guided crisis flows and Stronger protocols use fixed product rules. The companion cannot write Bible verses; it can only point to references that the app replaces with verified database text. Suggested exercises and tasks do not save until you choose them.

Not a human relationship

The companion can offer support, but it is not a therapist, pastor, friend or emergency responder. It does not continuously monitor you, and no human is watching a conversation in real time.