The Conversations tab
What you can see of a patient's Pippa AI chat: the summary, known facts, flagged turns, crisis flags, and the escalation email you get when a conversation trips a self-harm flag.
The Conversations tab shows what Pippa's AI companion has learned and stored about a consented patient, plus any content that tripped the safety filter. It is not a live chat window and you cannot message the patient from here (for that, see Messages). It is a window into the patient's use of the in-app AI, gated behind their signed consent.
What you can see
- Summary. A rolling, plain-language summary of what the patient has been talking about with Pippa.
- Known facts. Short factoids Pippa has retained, for example a trigger the patient named or a goal they mentioned.
- Transcript. The most recent turns, with any flagged turn highlighted so it stands out.
If nothing has been logged yet, the tab says so. Chat content only appears for patients who have granted consent; consent explicitly covers AI chat content. For how the patient experiences this chat, see the patient-app Pippa chat page.
Crisis flags and their tiers
Every message the patient sends to Pippa (and every reply Pippa gives) is screened. Flagged content is recorded on the tab in a highlighted panel with the category and an excerpt. Not every flag is treated the same:
Flagged content is always recorded so you can review it, but only the crisis (self-harm) subset sends an escalation email. Other flagged categories are logged without paging the care team, to avoid alert fatigue. A flag also fires if Pippa's own reply raises self-harm when the patient's message did not.
The patient app itself layers its own crisis handling on top, with distinct tiers ordered self-harm, violence, medical eating-disorder emergency, then eating-disorder distress. The first tiers route to hotlines (988 and, for medical emergencies, urgent care or 911), while the eating-disorder tier routes to ANAD at 888-375-7767. The dashboard sees the result: a recorded flag, and for self-harm, an email.
The escalation email and what to do with it
When a conversation trips a self-harm flag, Pippa emails every clinician who holds a granted consent for that patient. The subject is Pippa risk alert, and the body names the patient, gives the flagged categories, and links straight to their patient page.
The email link takes you to the patient's page. Go to the Conversations tab to read the flagged content in context.
Act on it per your clinical judgment and the crisis playbook. Review or update the safety plan if warranted.
Acknowledge the related concern on the patient's overview so the care team knows it has been seen and handled. Your name and the time are recorded.
Even before a patient consents, a self-harm flag in their chat still alerts you so no one falls through the cracks. You get the safety signal only, not their records.