AgentBrains Synthetic QA icon

AgentBrains Synthetic QA

Runs a synthetic QA evaluation test on your agent and returns scoring results.

AgentBrains Synthetic QA

AI-generated

Summary

Launches a synthetic conversation test with a configured AI persona, evaluates it against selected quality criteria, and returns scoring results and a structured report.

Inputs

  • syntheticUserId (required) — Select a Synthetic QA (persona) for this test run from the list of available users, or specify an ID using an expression.
  • runs (required) — The number of conversations to run for the synthetic test.
  • scoring (required) — A list of behaviors to evaluate and score, such as 'Human-Free Issue Handling', 'Customers Mood Change', 'Objection Handling', and others.
  • manualGoals — Goals to use across all conversations. If empty, goals are auto-generated based on the persona.
  • negotiation — Promotion or negotiation text the synthetic user will apply during conversations.

Output shape

A single item containing a JSON object with scoring results and a parsed report.

The node polls the AgentBrains API for up to 15 minutes (900 seconds) to retrieve the completed test results. The report is parsed into a structure with a title, sections, bullets, and optional body text.

Examples

Example 1: Run a standard quality check on a customer support persona

Select a persona, set runs to 5, and score 'Human-Free Issue Handling' and 'Objection Handling'.

Example 2: Test a sales agent with a specific promotion

Select a sales persona, set runs to 3, add a negotiation text, and score 'Making a Sale'.

Links

Discussion