No. SleekAI is configured to refuse diagnosis and refuse any attempt to assess a child based on parent description. Diagnosis is what the evaluation is for, and the bot is not a clinician. The system prompt enforces the refusal across every conversation regardless of how the parent phrases the question. The bot can describe what an evaluation involves, who on the team conducts which assessments, and how the report is delivered, but any determination about a specific child has to come from the evaluating psychologist.

No. The system prompt refuses parenting advice, behaviour scripts, and coaching on how to respond to a specific child's struggles. That work has to come from a psychologist who has met the child and understands the family system, not from a chatbot working from one parent's description. The bot redirects calmly and offers to book either an evaluation or a parent consult, depending on what the parent describes wanting. The boundary holds even when the parent explicitly asks for tips.

Your system prompt should surface the Childhelp National Child Abuse Hotline at 1-800-4-A-CHILD before any other flow when any indication of abuse or neglect appears. The bot recognises signals (physical signs, controlling caregiver behaviour, fear of going home, unexplained injuries) and leads with the hotline and 911 for immediate danger. Booking is secondary to safety. The bot does not improvise on this point. Most practices also add language directing the visitor to local child protective services if the parent is reporting a concern about another household.

Distress signals (self-harm ideation, hopelessness, or descriptions of crisis, whether reported by the parent or expressed by an adolescent reaching out directly) route to 988 (the US Suicide and Crisis Lifeline) before any other flow. The system prompt sets that order explicitly. Most practices also add language about the nearest pediatric emergency room and the local mobile crisis team, since 988 is often only one piece of the immediate-response picture for a child in active crisis.

Yes, when each assessment type is described on your practice's published pages and each psychologist is tagged in custom fields with the assessments they conduct. The bot routes accordingly: a parent asking about an autism evaluation gets matched to the autism specialist, a parent asking about a school-related learning difference gets matched to the psychoeducational assessor. The bot does not improvise on which assessment is appropriate, since that determination belongs to the clinician at intake.

Assessment pricing is typically bundled across multiple sessions (parent intake, child sessions, rating scales, written report) and runs in the low thousands, which is different from per-session therapy billing. The bot quotes your published bundled pricing accurately, explains which insurance plans cover the diagnostic component versus which require out-of-pocket payment with a superbill, and notes that verification happens at intake. The transparency is important because the cost difference between assessment and therapy is a common source of confusion.

Yes. SleekAI is multibot, so the assessment page can run a chat scoped to evaluation intake (what's involved, how long, pricing, when the report is delivered) while the therapy page runs a separate bot with its own framing. This matters because the two services have almost no overlap in intake pattern, and forcing them into a single prompt produces compromised answers for both audiences. Each bot has its own tone and routing path.

Yes, in your WordPress database. Retention is set through the plugin, typically 7 to 30 days for child psychology practices given the sensitivity of conversations involving minors. Conversations stay on your server, and OpenAI or Anthropic calls use your own API key directly without a SleekAI proxy. Most practices forward conversation summaries to their EMR through a webhook and purge raw logs on a strict schedule, particularly when the conversation references the child by name or describes specific clinical concerns.