1
Assign Persona & Jurisdiction
Establish the precise clinical, legal, or pedagogical identity of the language model.
2
Reasoning Methodology
Dictate the exact cognitive framework the model must use to process the data before generating text.
3
Raw Incident Data
Input the unstructured, messy data (e.g., an email from an ex, a school incident report, or a brain dump of morning chaos).
4
Enforce Strategic Guardrails
Crucial: Explicitly ban generic, harmful, or legally vulnerable responses that the AI defaults to.
5
Formatting Mandate
Force the model to return data in highly actionable, structured formats rather than dense paragraphs.