> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://poly-ai.ferndocs.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://poly-ai.ferndocs.com/_mcp/server.

# Testing

> Validate your agent before and after going live with simulation tests and A/B experiments.

Testing in Agent Studio gives you three ways to validate your agent:

* **Simulation tests** — automated conversations that verify your agent handles specific scenarios correctly. Author a test by describing what a caller will do and asserting how the agent should respond, then run it against Draft or Sandbox to catch regressions before they reach production.
* **A/B tests** — split live traffic between two versions and compare real-world performance before fully rolling out a change.
* **Studio Assistant** — generate simulation test scenarios from natural language. Describe a flow and [Studio Assistant](/agent-studio/studio-assistant/overview) authors the test cases for you, keeping them in sync as your agent changes.

Use simulation tests throughout development to catch issues early. Use A/B tests when you need real-world evidence that a change improves outcomes. Use Studio Assistant when you want to quickly scaffold test coverage without writing each case by hand.

#### [Create simulation tests](/agent-studio/testing/simulation-tests)

Author scenario-based tests with assertions and system actions.

#### [Run and review](/agent-studio/testing/run-and-review)

Execute tests, inspect results, and organize with test sets.

#### [A/B testing](/agent-studio/testing/a-b-testing)

Split live traffic between two versions and pick a winner.

#### [Automate with API](/agent-studio/testing/automate-with-api)

Gate deployments on test results from CI.