EMPS

Two channels, kept separate

The thing that makes EMPS work is a line drawn down the middle of it. Some of what you record is shared with your partner. Some of it is yours and Amanda’s alone, and your partner is never shown it — nor told that it exists.

That separation is enforced in the database with row-level security rather than left to the app to remember, and the shared weekly session is checked before it is saved to make sure nothing private has bled into it. The Privacy Policy sets out, line by line, exactly who can read each thing you write — it is the first substantive section of the document for that reason.

What Amanda does

Every week she reads what you have both recorded and writes a session for the two of you to read together, plus private coaching for each of you alone. She keeps working notes between weeks, so the coaching builds on what came before instead of restarting from nothing each time.

She is a large language model, and the app says so plainly rather than implying a person. Coaching is generated automatically, is not reviewed by a human before you see it, and can be wrong. It is something to consider, not a determination about your relationship.

What it is not

Not therapy, not counselling, not medical advice, and not a safety service. Nobody reads your entries as they arrive, and nothing you write reaches a person who could act on it. EMPS surfaces crisis resources where coaching detects language that may indicate risk, but that detection is automated and incomplete and must never be relied on.

The Terms also prohibit using EMPS to pressure, monitor or control a partner, and we terminate accounts for it. An app two people share is open to that kind of misuse, and saying so in the agreement is what lets us act.