About Supplemental Instructions in an Oracle Analytics AI Agent

Supplemental instructions play a crucial role in shaping how an AI agent interprets user questions and behaves within a specific business context.

Datasets and knowledge documents provide factual information. Supplemental instructions define the logic, expectations, and interpretation rules that guide the Oracle Analytics AI Assistant’s reasoning. They act as the agent’s operating manual by reducing ambiguity, minimizing hallucinations, and ensuring that responses align with your organization’s terminology, standards, and domain logic rather than generic AI assumptions.

Custom supplemental instructions currently support up to 6,000 characters per agent.

Oracle recommends using the following structure for your supplemental instructions.

1. Intent / Objective

   Describe when the rule applies or what it is intended to achieve. For example, Apply this rule when a user asks about Important Orders or Regular Orders.

2. Definition / Use Case Description

   Specify how the agent should interpret or classify the concept. For example, Use the following mapping: ‘Low, Medium, Not Specified’ = Regular Orders; ‘High, Critical’ = Important Orders.

3. Examples / Rules / Conditions

   List the triggers, keywords, or conditions that determine when the rule should be executed. For example When terms like ‘important,’ ‘regular,’ or ‘priority comparison’ appear, apply the above Order Priority mapping.

Using this structure ensures your instructions are unambiguous and makes it easier for the AI agent to understand business intent and deliver accurate, context-aware responses.