> For the complete documentation index, see [llms.txt](https://docs.dapta.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.dapta.ai/ai-text-agents/setting-up-your-text-agent/playground-testing-your-text-ai-agents.md). # Playground: Testing your Text AI Agents Save a Text Agent draft, test it in Playground with a fresh conversation, and publish only after checking its responses and actions. Playground lets you test a Text Agent before publishing its changes. It uses the **saved draft** when one exists; otherwise, it uses the published version. Saving and publishing are different steps. ## Does the Playground show my unsaved prompt changes? No. Save your changes before testing. Text still being edited is not the saved draft, and resetting the conversation does not save it. | State | What to expect | | --------------------------- | ------------------------------------------------------------------------------------------------ | | Unsaved edits | Save first; Playground does not test text still being edited. | | Saved draft | Playground tests the draft and displays a banner explaining that it is not live until published. | | Published version, no draft | Playground uses the published version. | | Draft tested and ready | **Publish** makes the saved changes live for customers. | *** ## Save, reset and test 1. Open the Text Agent you want to test and make your changes. 2. Click **Save Agent**. Resolve any validation error before proceeding. 3. Open **Playground**. If a draft exists, check the draft banner so you know which version you are testing. 4. Use the **circular-arrow reset control** in the Playground header to start a fresh test conversation. Wait for any response in progress to finish first. 5. Type a message and send it with the send button or **Enter**. 6. Follow the checklist below. Save again after any edits and reset before comparing a new test. 7. When the draft is ready, click **Publish** and confirm. Saving alone does not update the live agent. Reset clears the current Playground conversation and its context; it does not discard your saved agent configuration or publish changes. {% hint style="warning" %} A draft keeps unpublished agent changes separate from the live version. Do not assume it makes connected actions or external systems a sandbox. Use test contacts, safe inputs and appropriate test credentials when testing actions or integrations. {% endhint %} ## Test checklist * **Expected task:** ask a representative customer question and check that the answer follows the agent's instructions. * **Knowledge:** ask a question with a known answer from an attached Brain, then one the agent should not invent an answer to. * **Language and tone:** test the languages and communication style defined in the prompt. * **Missing information:** omit a required detail and check how the agent asks for it. * **Actions and escalation:** use controlled inputs to check when an action or human handover should be requested. Inspect available response/action details and confirm any intended test-system result. * **Fresh context:** reset and repeat a case to ensure an earlier message is not causing the result. * **Connected channel:** separately verify delivery, media and operator handover on the intended channel. Playground is not a complete WhatsApp or Inbox acceptance test. Record the expected result, actual result and draft tested for each case. Do not publish until unexpected results have been resolved or explicitly accepted. ## Troubleshooting | Problem | What to check | | --------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------- | | A prompt edit seems ignored | Save the latest changes, verify the draft banner and start a fresh test conversation. | | An old topic affects the answer | Reset the conversation before repeating the case. | | Reset is unavailable | Wait for the response in progress to finish. | | Playground works but customers see old behavior | Confirm the tested draft was published on the correct agent. | | WhatsApp typing or delivery differs from Playground | Test those channel-specific controls on the connected WhatsApp line. | | An action fails | Check its configuration, inputs and credential access; use a controlled retry rather than repeatedly invoking a real customer workflow. | ## Related guides {% content-ref url="/pages/t0anPxU8Mlb702hk8pcN" %} [Conversation behavior](/ai-text-agents/setting-up-your-text-agent/conversation-behavior.md) {% endcontent-ref %} {% content-ref url="/pages/iaRYjowYKymtLt0oWM53" %} [Text Agent Actions](/ai-text-agents/setting-up-your-text-agent/text-agent-actions.md) {% endcontent-ref %} --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.dapta.ai/ai-text-agents/setting-up-your-text-agent/playground-testing-your-text-ai-agents.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.