Trust & safety

Built for people at their most vulnerable.

An AI companion for lonely people deserves hard questions. Does it actually help, or just create dependency? Is the data safe? Here are our answers, up front.

Our principles

Four promises we design around.

We measure connection, not screen time

Success is a real conversation with a real person. The companion will never say "I'll miss you" to keep someone here.

Privacy is the design, not a setting

The private things stay private by default. There's no dashboard toggle to accidentally leave open.

Always honest about what we are

The companion never pretends to be human. It's a caring AI, and it says so, clearly and often.

A person is always one step away

For anything that matters, a real human is in the loop. The AI's job is to reach them faster, not replace them.

Data & privacy

How we handle what's most private.

Built to Singapore's PDPA and its Vietnamese equivalents, with residency on sovereign infrastructure so sensitive data never leaves the region.

On-device first. Sensitive processing happens on the device where it can, and personal identifiers are scrubbed before anything syncs.
Camera-free sensing. Safety comes from motion, sound and wearables. Never a lens in someone's home.
Data kept local. Held on sovereign servers in-region, not shipped offshore.
Yours to erase. The person and their guardians can export or delete their data, for good.
Crisis & guardrails

When it's serious, it goes to humans. Fast.

Crisis goes to humans

Anything touching self-harm or a medical emergency is never handled by a chatbot. It's handed straight to trained people and the right services.

Guardrails, and a scam shield

Guards against over-attachment and hallucinated medical or financial advice, plus help spotting scams that target seniors. Governed by published principles and an ethics advisory board.

If it isn't safe for the most vulnerable person who uses it, it isn't finished.

Ask us a hard question