Crisis Response Protocol

This protocol covers the Mendly app used without a therapist. It describes what happens when text a person writes in the app suggests they may be at risk of suicide or self-harm.

For clients linked to a therapist, the same language check runs on the text they write in a check-in, on the website or in the app. When it is triggered, the client is shown the 988 Lifeline and that day's check-in is flagged on their therapist's dashboard until the therapist marks it reviewed. A positive answer to the self-harm question on the PHQ-9 is flagged the same way. There is no real-time alert. Their check-ins are part of a clinical record held by a licensed professional, and that clinician's own obligations govern what happens next; Mendly itself contacts no one.

Used on its own, Mendly is a general wellness product. It is not a crisis service, not a medical device, and not a substitute for professional care. It cannot contact anyone on a user's behalf, and it makes no clinical assessment of anyone. For therapist-linked clients, standardized questionnaires are scored by their published rules as screening results for the treating professional to interpret; Mendly does not make clinical judgements about those clients either.

Every time a check-in is saved, the note and any written answers are screened for language indicating suicidal ideation, intent, or self-harm. This runs for everyone who uses the app, with or without Mendly Plus.

For people with Mendly Plus who have asked for a weekly read, the week's notes and written answers are screened again before the read is written.

The screen looks at two things: explicit statements of intent or self-harm, and sustained expressions of hopelessness, worthlessness, or feeling like a burden. A single ambiguous phrase does not trigger a response on its own. Low scores alone never do — having a hard month is not a crisis, and treating it as one would make the response meaningless. Typographic differences such as curly apostrophes are normalised first, but no keyword screen catches everything, and it can also be triggered by language that does not signal risk.

This screen is a language check. It is not a diagnosis, a risk score, or a clinical judgement, and Mendly does not present it as one.

• When a check-in is saved with explicit self-harm language, the app shows the 988 Suicide and Crisis Lifeline straight away, with options to call, text, or chat, and the usual celebration and offers are skipped. Sustained distress language on its own does not interrupt a check-in; the 988 line is always shown on the check-in screen
• When a weekly read is due and the week is flagged, no read is written by the model and nothing from that week is sent to it. The person instead gets a short fixed message: that it looks like a hard week, a suggestion to reach out to someone they trust, the 988 Lifeline (call or text), findahelpline.com for people outside the US, and a reminder that Mendly is not a crisis service. The 988 Lifeline is raised to the top of that read
• The weekly read records that it was flagged and a short reason code (explicit self-harm language, sustained distress language, or distress language alongside low ratings), so the raised Lifeline appears again whenever it is reopened. Nothing is recorded against the person for a single check-in
• The person is not told they were flagged. Being told your writing is scanned teaches people to write less honestly, and honest writing is the only thing in this product that could ever help them. The screening itself is disclosed in our privacy policies
• A count is recorded — the number of referrals issued on that date, with no user identifier of any kind attached

The 988 Lifeline is also shown on every report, under every check-in, and in the app's Settings, whether or not anything was detected, with a link to findahelpline.com for people outside the US. Nobody has to be in crisis to be offered help.

• Mendly does not contact anyone. Not emergency services, not family, not a therapist. The duty to warn applies to licensed psychotherapists; Mendly has no clinician and no such duty, and notifying a third party uninvited would be a privacy violation rather than a rescue. For therapist-linked clients, the flag on the therapist's dashboard is part of the service the therapist chose, not a message sent by Mendly
• Mendly does not suspend, restrict, or close accounts on the basis of this screen
• Mendly does not build a risk profile or a risk score. The only per-person record is the flag and reason code on a weekly read described above, which is used for nothing except showing the resource again, is included in the person's data export, and is deleted with their account
• Mendly does not share the result with advertisers, insurers, or anyone else

The model that writes weekly reads is instructed never to describe, hint at, or quote methods of suicide or self-harm, never to write anything that could encourage either, and to leave out entirely any note that mentions self-harm, suicide, wanting to die, or deep distress rather than quoting, counting, or analysing it. It is also instructed never to name a condition or assign a severity, and never to recommend or discourage treatment. Every read is checked before it is saved, and a read that contains crisis language or clinical terms is rejected and rewritten; if no acceptable weekly read can be written, a short fixed notice is saved instead. Reads describe patterns in what the person recorded, and nothing more.

Mendly records the number of crisis referrals issued each day. That figure contains no user identifiers and cannot be traced to a person. Beginning July 1, 2027, Mendly will report the annual figure to the California Office of Suicide Prevention as SB 243 requires, together with the protocol on this page.

This protocol is reviewed at least annually and whenever the report generator changes materially.

Mendly LLC · support@mendly.me

See also our consumer health data privacy policy and app terms.