How Clarmo's AI Works

Last updated: October 8, 2026.

You are talking to an artificial intelligence

In Clarmo, Clarmo is an AI, not a person. We remind you at the start of every chat conversation. No person writes the replies.

What it does

The chat: you can write to it whenever you like, including from SOS (“Talk”) in a moment of urge.
Weekly notes on your stats: every Monday it leaves a short reading of how the week went, and once a month one about your year.

What the AI does not do: SOS tool recommendations, your streak and your achievements are worked out by the app with fixed rules, not by the AI.

Which model it uses

Replies are generated by Claude, a model by Anthropic. Depending on how much the chat is being used, a lighter Claude model may answer.

What the AI sees

To answer with context, the AI receives:

what you write in the conversation and the earlier messages of that conversation;
your check-ins (day type, mood, urges, feelings, what was going on or what helped, and what you wrote under “Other”), your journal and your SOS sessions from the last 30 days;
your goal (full commitment or gradual reduction), your first name, the time and day on your phone, since when you have used Clarmo, your current and best streak and your clean days, and whether you have taken the PIED test (only whether, never the result);
if you open the chat from Stats, a summary of your history (up to two years) worked out by the app, plus what you wrote in your journal and under “Other” on your hardest days of the last 90 days.

It does not receive your email, your payment details or anything that identifies your account outside the app. Stats notes are written from numbers and counts only, never from your journal text or what you write under “Other”.

What happens to that data

What is sent to Anthropic is used only to generate the reply. Anthropic does not use it to train its models. It may keep it for a limited time for safety and abuse prevention: it deletes it within 30 days, unless its automated systems flag a breach of its usage policy, in which case it may keep it for up to 2 years.
Your conversations and notes are stored in your account so you can come back to them. You can delete each conversation from the chat history, and everything is deleted if you delete your account.
More in the Privacy Policy (clarmo.app/privacy).

It is optional and needs your permission

The AI only works if you accept it, with a separate, unticked box, before subscribing. If you don’t, the app works the same, without the chat and the notes. You can turn it on or withdraw it at any time in Profile → My data & consent, or from the chat itself.

Its limits

It can be wrong. Do not take its answers as facts or as medical or psychological advice. Clarmo is a personal support tool and is not a substitute for professional help.
It does not diagnose. If something points to a physical cause or strong distress, it will suggest talking to a professional.
Fair use: if it is used a lot in a short time, replies may get shorter or it may take a while to be available again. If you come from SOS in a moment of crisis, the chat never shuts: at most it answers more simply, or with a fixed message pointing you to the SOS exercises and help lines.

If you are in danger

If you are thinking about harming yourself, don’t wait for the AI: call the emergency number in your country.

How Clarmo's AI works