Modern AI Engineering

Lesson 14.3 · 25 min · video lesson

Evaluating AI Agents: Metrics and Methods That Work

Our refund agent gave the customer the right answer, but it called the wrong tool twice and nearly emailed someone else: did it pass?

In short: Evaluating an AI agent means judging not just its final answer but the whole multi-step process: whether the goal was reached in the real environment, which path it took, how it used tools, how well it planned, and what it cost. Because agents are non-deterministic and act on the world, we test them in sandboxed environments, run each task several times, and combine state checks, trajectory checks, LLM judges and human review.

What you will learn

  • What is an AI Agent?
  • What is AI Agent Evaluation?
  • Why do we need AI Agent Evaluation?
  • How is AI Agent Evaluation different from LLM Evaluation?
  • Types of AI Agent Evaluation
  • Outcome Evaluation
  • Trajectory Evaluation
  • Tool Use Evaluation
  • Planning Evaluation
  • Key Metrics for AI Agents
  • Agent Benchmarks
  • Methods to Evaluate AI Agents
  • Frameworks and Tools for AI Agent Evaluation
  • Challenges in AI Agent Evaluation
  • Best Practices

In this lesson

  1. What is an AI agent, and what is agent evaluation?
  2. Why agent evaluation is needed, and how it differs from LLM evaluation
  3. Outcome evaluation and trajectory evaluation
  4. Tool use evaluation and planning evaluation
  5. Key metrics for AI agents
  6. Agent benchmarks
  7. Methods, frameworks and tools
  8. Challenges and best practices
  9. Going one level deeper
  10. Practice: try it yourself

Key terms in this lesson

AI agent
An LLM-driven system that repeatedly chooses actions, often tool calls, observes results and continues until a goal is reached.
Trajectory
The full sequence of reasoning, actions and observations in one agent run.
Outcome evaluation
Checking whether the agent achieved the goal, ideally by inspecting the final environment state.
Tool-call accuracy
How correctly an agent chooses tools and fills in their arguments.
pass@k
Probability that at least one of k attempts at a task succeeds.
pass^k
Probability that all k attempts at a task succeed; a measure of reliability.
Sandbox
An isolated, resettable environment where an agent can act without real-world side effects.

This lesson is part of the Generative AI Engineering and Complete AI Engineer tracks. See plans and pricing to open the full lesson, its video, labs and quiz, or try the free lesson first (free account needed).

← 14.2 LLM-as-Judge: Automating Evaluation with Another Model · 14.4 Agent Observability: Traces, Spans, and Debug Signals →