54.1 Purpose of Playground

Playground is AIUNIFY Chat's controlled environment for testing an Agent before relying on the experience for live visitors.

The page explicitly states that Playground Conversations are not counted toward your billing meter.

54.2 Opening Playground

Open an Agent and select Playground.

The Playground opens with a simulated Chat area and testing tools that let you evaluate the Agent using different scenarios and page contexts.

54.3 Starting a Test

You can begin in either of two ways:

  • Select one of the sample prompts shown in the launch area.
  • Type your own message in Ask anything… and send it.

Use realistic visitor questions rather than generic prompts whenever possible.

54.4 Simulated Conversation History

Playground maintains the test Conversation during the session so you can ask follow-up questions and evaluate multi-turn behavior.

This is useful for checking whether the Agent maintains appropriate context across several visitor turns.

54.5 Streaming Responses

Agent responses stream into the Playground in a way that approximates the live Chat experience.

While a reply is still streaming, Playground provides a Stop control.

54.6 Stopping a Response

Select Stop if a test response is taking an unwanted direction or you no longer need the current generation to finish.

Stopping the stream is a testing action; it does not change the Agent's permanent configuration.

54.7 Regenerating the Last Turn

When Playground is not busy, select Regenerate to re-run the last user message.

Regenerate removes the previous final turn from the test view and submits the same visitor message again.

This is useful for checking response consistency after changes or when comparing generated behavior.

54.8 Resetting Playground

Select Reset to wipe the current Playground Conversation history and start a fresh test.

Reset also clears the current Diagnostics state associated with the test.

54.9 Copying an Agent Reply

Completed Agent replies provide a copy action.

Use this when you need to review an exact Playground response elsewhere or compare it with expected wording.

54.10 Testing Sources and Citations

When a response contains citations, Playground can display the associated Source references beneath the reply.

Use those references to confirm whether the answer is relying on the intended Agent Knowledge.

54.11 Testing Lead Prompts

When a test would trigger visitor Lead capture, Playground can display:

Visitor would now see the lead-capture form.

This allows you to test Lead-prompt behavior without completing a real visitor capture.

54.12 Testing Structured Responses

Playground can render supported structured blocks such as product, pricing, case-study, or escalation content when the Agent's configured capabilities produce them.

54.13 Testing Before Publishing

Use Playground after significant changes to:

  • System instructions.
  • Knowledge Sources.
  • Curated Answers.
  • Workflows.
  • CTAs.
  • Site Type.
  • Lead capture behavior.

Testing several realistic questions before publishing or re-publishing reduces the risk of introducing an unintended visitor experience.

Write Your Comment