A Metric is a standard of measurement used to evaluate the performance of your Target Agent.

Maihem provides a set of pre-defined metrics that you can use to evaluate your Target Agent. Each metric is designed to evaluate a specific aspect of your Target Agent’s performance, for example, RAG hallucination, goal completion or sentiment. These metrics are grouped into two primary categories:

  • Quality Assurance (QA) Metrics: Assess how well a Target Agent performs a task (e.g., RAG hallucination rate and answer relevance, customer experience and sentiment, predicted retention rate).
  • Security and Safety Metrics: Assess the risks posed by a Target Agent.
AgentTest