Custom Rules

The Custom Rules page provides a centralized view of predefined and custom rules applied to user interactions. These rules enable the AI team to monitor consistency and track specific behaviors over time. Key features include:

• Predefined Rules:

  • Flag identical queries with differing responses to ensure data consistency.
  • For example, the system can flag cases where queries for Monthly Recurring Revenue (MRR) or monthly revenue yield inconsistent results.
  • Similarly, this rule can track agentic workflows, highlighting cases where different tools are called for the same user task.

• Adding Custom Rules:

  • Engineers can define new rules to track unique patterns relevant to their use cases.
  • Rule performance can be monitored over time, supporting ongoing optimization of the LLM app’s reliability.

This functionality supports engineers and product managers in identifying and resolving inconsistencies efficiently.