No. SleekAI is configured to never ask about trauma history, never request details about what happened, and never push the conversation toward content the visitor has not offered. The system prompt enforces that boundary across every conversation regardless of how the visitor frames the question. The bot can describe modalities, list which therapists practise what, and help schedule a consultation, but the actual history conversation is reserved for the consultation itself with a licensed clinician who can pace it appropriately.

No. The system prompt refuses grounding exercises, coping techniques, and any form of in-chat processing work, even when the visitor explicitly asks for them. That work has to come from a therapist who can pace it safely and respond to what is happening in the moment, not from a chatbot working from text. The bot redirects calmly to a consultation and routes to 988 if the visitor describes distress that needs immediate support. The boundary is deliberate and the system prompt enforces it consistently.

988 (the US Suicide and Crisis Lifeline) routing fires before any other flow when distress signals appear. The bot is configured to recognise hopelessness, self-harm ideation, dissociation cues, or descriptions of crisis and to lead with 988 and the local emergency line first. Most trauma practices also add language about the Crisis Text Line (text HOME to 741741) and the nearest emergency department. Booking and practice information are secondary to safety, and the system prompt enforces that order on every conversation.

If your practice includes the routing, the system prompt can surface RAINN's National Sexual Assault Hotline at 1-800-656-HOPE before any other flow when sexual assault is disclosed. The bot does not engage with the disclosure beyond acknowledging it briefly and providing the hotline and offering to help book a consultation with a therapist who specialises in sexual trauma if the visitor wants to continue. Pacing is the priority, and the system prompt sets the order explicitly.

The system prompt routes to the National Domestic Violence Hotline at 1-800-799-SAFE (7233) and 911 for immediate danger when signs of ongoing abuse or controlling behaviour appear. The bot recognises indicators like fear of a partner, controlling dynamics, isolation, or descriptions of physical or emotional abuse, and leads with the hotline before any booking flow. Safety routing is the first priority in trauma intake, and the system prompt makes that order non-negotiable on every conversation.

Yes, when each therapist is tagged in custom fields. The bot reads modality and EMDRIA, IFS Institute, or SEP certification into the prompt and matches conversationally. Visitors often arrive with a preference (because a friend, therapist, or research has pointed them toward a specific approach), and the bot routes accordingly. Visitors with no preference get a brief framing from the practice's published copy, with the actual modality choice decided in consultation with the therapist who will do the work.

In your WordPress database, with retention set through the plugin. Most trauma practices use tight retention (7 to 14 days) given the extreme sensitivity of any content the visitor offers, even when the bot is scoped to never request trauma history. Conversations stay on your server, OpenAI or Anthropic calls use your own API key, and there is no SleekAI proxy. Most practices also forward conversation summaries to their EMR through a webhook and purge raw logs on a strict schedule that fits the privileged nature of the inquiries.

Yes. The 28+ display conditions let you scope the chatbot tightly to booking-intent pages (contact, fees, therapist bios, new clients) and hide it on the blog, resources, and educational pages. Visitors reading those pages are often in the help-seeking phase that itself is therapeutic, and pushing chat onto them can feel intrusive. For trauma practices specifically, most firms also hide the bot on the about page where the visitor is researching rather than booking, since the trust building happens through reading, not chat.