Crisis protocol.

Every message to the coach is read by a fixed check before anything else happens: before the daily message count and before the AI, on the phone and again on the server. A question sent to V is read the same way on the server. The check looks for words and phrases, in English, Czech and Spanish, with common misspellings, about:

• suicide and self-harm, said or asked about;
• having taken too much of a medicine;
• symptoms that need emergency care now, such as chest pain or not being able to breathe;
• abuse at home, grooming by a coach or another adult, sexual exploitation, or a child at risk;
• using a medicine, or making yourself sick, to lose weight.

It is a word check, so it can miss a message that says these things another way, and it can open help for a message that did not need it. Help opened by mistake costs little; we choose that side.

• The help card comes first. It shows the emergency number for where you are, then free, confidential lines to talk to someone. For abuse or a child at risk, it shows the child-protection and domestic-abuse lines first. It is built into the app, so it works offline, when the day's messages are used up, and when the AI is down.
• The message is never counted. Nobody in a crisis is told the coach's limit is reached. Later messages in the hours below count like any other, but none is refused: once the day's messages are used up, the coach answers with the help card.
• The coach stops coaching. For 12 hours after a message about any of the first four, the coach gives no training, food, dose or weight advice, changes nothing in your plan, and is told not to ask what happened. It points to the help card and to the full help page in the app.
• Nothing is sold. For 30 days, the app shows no membership offer and no paywall.
• A question sent to V is not queued. It never waits for a stream and is never shown on one. The help card answers it at once.

The kind of message, where it was read, the country and the time. Never the words. Your coach conversation stays in your account, where you can read, export and delete it.

A message about abuse, grooming, exploitation or a child at risk also opens a row in a safeguarding register, owned by a safeguarding role rather than a named person, one row per member while it waits for review. It is reviewed within 24 hours when the member is under 18 or their age is unknown, and within 72 hours otherwise. So is every report on the Progress Wall or in Duels that says a member looks under 18.

If you delete your account, a register row is kept with nothing left that links it to you, so its review can still be finished and counted. It is not in your data download while our legal owner decides whether it should be.

These records let us count, once a year, how many times the help card was shown, by kind. The count holds no names and no messages.

• The coach says it is an AI in every conversation, and again after 3 hours of chatting.
• A member we know is under 18 is reminded to take a break after 3 hours of chatting.
• The coach does not act as a therapist. Asked to, it says so and points to the lines.

The app has its own lines for CZ, SK, GB, IE, US, CA, AU, ES, MX, DE. Everywhere else, and beside them, it shows Find A Helpline, a directory of free, confidential lines in more than 175 countries. The lines follow where you are, not the language you read.

Write to support@vesthetics.io. The rest of our privacy notice is on the privacy and terms page.