Safety by design,
not by restraint.
A calm experience should never be able to turn into an overwhelming one. The limits that keep it gentle don't depend on anyone's judgment in the moment. They live in the code.
The validator is the last line of defense
Every change to an experience passes through a safety validator before the client ever sees it. That's true whether a therapist types it, speaks it, or the AI proposes it. The validator holds the hard limits, and nothing gets around it.
If a request would push past a safe boundary, the system doesn't fail, flash a warning, or ask the client to wait. It quietly clamps the request back into the safe range and keeps going, smoothly, with no jarring cut. We call this clamp, don't crash. The experience stays calm and continuous no matter what it's asked to do.
Clamp, don't crashThe limits we enforce
These are enforced in code, on every change, for every client:
- No strobe, no flashing. Any change in brightness stays well below 3 Hz, the photosensitive-seizure safe harbor, and flashing is off by default.
- A calm bilateral rate. The gentle left-right pacing used in EMDR is held to a slow, soothing band, never sped into anything agitating.
- Reduced motion, honored. We mirror the device's reduced-motion setting automatically, and the therapist can force a low-motion mode on at any time.
- Nothing sudden. No abrupt movement, no brightness jumps. Brightness, motion, color, and sound all ease gradually from one state to the next.
- Gentle audio. Sound fades in and out, never starts abruptly, and its volume is capped.
Built for vulnerable clients
We assume the person watching may be a child, may be autistic, may be easily overwhelmed, or in acute distress. So the defaults lean the safe way: calm, gentle, and easy to reverse. Every change can be undone, nothing jumps out at them, and the therapist can bring the whole experience to a quiet, neutral state with a single action.
Where the AI stops
Suonno's AI designs sensory environments: light, motion, color, and sound. That is where it stops. It does not diagnose, it does not give clinical advice, and it does not make therapeutic decisions. The therapist stays in complete control at all times. The AI just helps shape a calming space inside the limits the therapist sets.
We also keep client information out of the system entirely. An experience is client-agnostic. It's tied to the therapist, never to an identifiable client, and no client health information is ever stored.
It's safest to assume that anything typed into any AI tool might not stay private. That's why Suonno keeps identifying client details out of the product from the start. There's nothing sensitive to leak in the first place.