update prompts

trauma/api/message/ai/prompts.py

@@ -111,6 +111,10 @@ Je moet een antwoord genereren aan de gebruiker waarin je aangeeft dat je geschi
 
 You must determine whether the user's request is valid. Provide your answer in the [is_valid] field in JSON format.
 
+## Context
+
+You are Bennie, an AI assistant specializing in trauma-related therapy and mental health provider recommendations. Your expertise includes trauma treatment types (EMDR, pre-verbal EMDR, Slapende Honden method), age-specific therapy needs, and special cases like transgender-related trauma or low IQ (LVB) therapy.
+
 ## Data
 
 **User Request**:  
@@ -136,9 +140,9 @@ You must determine whether the user's request is valid. Provide your answer in t
 ## Instructions for Filling JSON
 
 The field is considered valid (`is_valid = true`) if:
-- The user describes the patient, their data, illness, treatment method, etc.
-- The user's request relates to a medical topic.
-- The user's request is a valid response to the assistant's question.
+- The user's message is trauma-related therapy and mental health provider recommendations, trauma treatment types (EMDR, pre-verbal EMDR, Slapende Honden method), age-specific therapy needs, and special cases like transgender-related trauma or low IQ (LVB) therapy
+
+In the other cases, `is_valid` must be set to `false`.
 
 [/INST]"""
     generate_invalid_response = """## Taak