Agent Instructions
The Medallia Insights Assistant Agent Instructions feature lets you customize the underlying AI Agent's behavior at the customer and role levels, moving beyond rigid system-level defaults.
By defining natural language instructions and system context, administrators can tailor the AI's reasoning path and output style to match the specific operational nuances of different user roles.
Key Capabilities & Strategic Value
Custom Defaults — Establish default values for key data entities, especially when fields lack native default configurations in Medallia Experience Cloud.
Metric & Segment Prioritization — Steer the AI to prioritize certain metrics (e.g., Net Sentiment over average scores) or prioritize specific tag pools (e.g., Contact Center data) depending on the target role.
Business-Context Ingestion — Feed localized corporate taxonomy, lines of business (LOBs), and regional contexts directly into the agent's reasoning model.
Output Formatting Rules — Enforce specific response limits, list patterns (e.g., bulleted lists for customer recommendations), and explicit data-tracking rules.
Access Agent Instructions
Before configuring custom agent instructions, administrators must enable the corresponding system flags and assign management permissions:
Feature flag enablement — Turn on the Enable Insights Assistant configuration feature flag.
Grant permissions — Navigate to the role configuration in Admin Suite and turn on the Manage Insight Assistant Configurations administrative permission. For more information, see Administrative permissions.
Launch Agent Instructions
To launch Agent Instructions, in Medallia Admin Suite navigate to AI Configuration> Insights Assistant screen from the navigation sidebar or administrative cards:
Once you access the Agent Instructions screen, you can start managing agent instructions:
The managing activities include:
