> For the complete documentation index, see [llms.txt](https://dataqueue.gitbook.io/voicehub-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://dataqueue.gitbook.io/voicehub-docs/agent-design-and-usage/qa-testing.md).

# QA Testing

QA Testing in VoiceHub allows you to simulate conversational scenarios and evaluate how well an agent behaves according to its configuration and your business expectations.

### **What it does**

* Automatically simulates text-based conversations
* Uses your evaluation prompt to judge if the agent’s replies are correct, helpful, or aligned
* Produces pass/fail results per test scenario
* Lets you identify weaknesses before deploying to production

### **Why it matters**

This is a critical tool for:

* Continuous testing during development
* Comparing performance across prompt updates or model changes
* Ensuring agents follow business logic and tone
* Running regression tests at scale

Unlike real calls, QA tests are fast, cost-free, and focused entirely on logic accuracy.

### **How to use it**

1. Go to QA Tests in the sidebar
2. Click + Add Test
3. Fill in:
4. * Title (e.g., "Booking – success scenario")
   * Description (optional for internal clarity)
   * Evaluation Prompt – This is what the model uses to assess the response (e.g., "The agent should confirm the booking and provide appointment time")
   * Test Message(s) – Simulated user input(s)
5. Click Save

Run the test anytime, or batch-run all tests after an update.

#### **Example**

Title: Appointment Booking – Success Case Evaluation Prompt:

Evaluate if the agent confirms the booking and restates the appointment time in natural language. The answer must be accurate, clear, and helpful.

Test Message:

Hi, I want to schedule a new appointment next Monday at 3pm.

#### **Result**

* ✅ PASS: Agent confirms the correct date/time with proper tone
* ❌ FAIL: Agent ignores time or responds vaguely

***

VoiceHub’s QA Testing ensures every conversation design is reliable before going live — saving time, cost, and reputation.