AivahHelp CenterGo to Aivah

Use Playground

Start a private conversation, send text or a file, create an image, use voice, and test Avatar Chat or a presentation.

Who this is forAnyone privately testing or using an Aivah AI employee

You will achieveStart a clean conversation with the intended employee and test its supported text, file, image, voice, appearance, and presentation experiences.

Before you beginReview the requirements before changing anything.

Last reviewed 10 July 2026Owner: Product EducationCurrent guide
On this page
  1. Before you begin
  2. Steps
  3. 1. Start a clean conversation
  4. 2. Send a useful first message
  5. 3. Attach one supported file
  6. 4. Create an image when the option is available
  7. 5. Use voice
  8. 6. Use Avatar Chat when appearance matters
  9. 7. Test presentation controls when applicable
  10. 8. Start fresh after changing employees
  11. Expected result
  12. Status meanings
  13. Usage credit impact
  14. Information stored or shared
  15. Limits and permanent actions
  16. Common problems and recovery
  17. Next step

Use this guide to work privately with an AI employee or test it before publishing. You will start with the intended employee and use only the conversation features needed for the task.

Before you begin

Confirm that the employee's preparation is completed. Prepare a focused first message and any supported file you need. For voice, check that your browser can use the intended microphone and speaker.

Steps

1. Start a clean conversation

Choose New chat. Select the employee from Agent, the available response option from Model, and a Voice when voice is needed. Confirm the appearance or Character where available, then wait for every choice to finish loading.

New conversation composer with employee and conversation choices
Select the employee and conversation preferences before sending the first message.

2. Send a useful first message

In the message box, include the audience, desired result, required format, and important boundaries. Press Enter to send. Use a new line when a longer request needs structure.

Private text conversation with an AI employee
A text conversation keeps the request and response together for review.

3. Attach one supported file

Use the attachment control and choose the supported file. Add a written instruction explaining exactly what the employee should do with it. For the most reliable result, send one file at a time and wait for the response before adding another.

Playground attachment controls for adding photos and files
Use the attachment control to add a supported photo or file.

4. Create an image when the option is available

Choose the image option and describe the subject, setting, audience, orientation, mood, style, and any text that must appear exactly.

Image creation request in the Playground composer
Describe the image and its intended use in the message box.

Wait for generation to finish before starting a duplicate. Generated work may also appear later in AI Drive.

Generated image result shown in a Playground conversation
The generated result appears in the conversation when it is ready.

5. Use voice

Choose the microphone or voice control and allow microphone access when asked. Speak clearly, wait for the employee to finish, and use headphones if speaker sound causes echo.

Active microphone recording in the Playground composer
The message composer shows when microphone recording is active.

Close or mute another browser tab or application that is already using the microphone.

6. Use Avatar Chat when appearance matters

Avatar Chat combines an animated companion, scene, voice, and conversation panel. Talk or type, show or hide the transcript, and change appearance or scene in Preferences. The first visit may open Preferences if no appearance is selected.

Avatar Chat with an image Character and visible transcript
Avatar Chat can show the Character with a conversation transcript.

7. Test presentation controls when applicable

A presentation employee can guide slides or video while answering questions. Move through the material with the displayed presentation controls, show the transcript, and open related Quiz content when available.

8. Start fresh after changing employees

Changing Agent changes the selected employee for the conversation. For important testing, start a new conversation after switching so the employee, knowledge, generated content, and voice session are clearly separated.

Expected result

The conversation uses the intended Agent, response option, voice, and appearance. Messages and supported attachments receive a response, media finishes before another request begins, and presentation or Avatar Chat controls work as expected.

Status meanings

Loading, generating, or processing means the current response or media item is not ready. Ready or completed means you can review the result and continue testing. Failed or an error means the request needs attention; check the selected Agent, attachment, permission, and connection before trying again.

Usage credit impact

Conversations, voice, image creation, and other supported Playground activity may use usage credits. The guide does not provide a fixed per-action formula. Review the balance and any notices shown in your workspace before a long test session.

Information stored or shared

The conversation appears in your conversation history. Supported generated work may also appear in AI Drive. If the employee uses an approved connected action, the outside service can receive the information needed for that action. Use non-sensitive test data whenever possible.

Limits and permanent actions

Feature availability, response choices, voices, file guidance, and conversation controls can vary by account. A microphone, camera, or screen feature needs browser permission. Start a separate conversation after changing employees for important tests.

Deleting a conversation is permanent in the current workspace; manage deletion from the history only after confirming the name.

Common problems and recovery

  • The wrong employee answers: start New chat, select the intended Agent, and resend the test.
  • A file request fails: send one supported file with a written instruction and wait before adding another.
  • The microphone does not work: allow browser access, close other apps using it, check the selected input, and refresh.
  • Voice connects but no sound plays: check speaker volume, browser sound permission, the selected Voice, and other active audio sessions.
  • The appearance is missing: open Preferences, choose an appearance and scene, and wait for loading.
  • Generated work seems missing: check the conversation, the Productivity gallery, and AI Drive before generating a duplicate.

Next step

Learn to manage conversation history, or use Manage, test, and improve to record a full quality test.