# AI Prompts This feature is [preview.](/appendix/feature-states.md) Settings | AI Prompts AI Prompts can be created to perform an action using AI on the transcription of the call. These actions can be triggered in multiple ways: 1\. Triggered manually by pressing the Generate Prompt button on the AI Prompts tab on the Call Details panel. When using this method, the response will be shown on the same panel. The response can also be shown in the call history panel if the setting Save Response to Local Call History is enabled. 2\. Triggered on chat callback reminder, email callback reminder, or chat consult transfer creation. When using this method, the response will be displayed in the consult transfer chat or callback reminder. The response can also be shown in the call history panel if the setting Save Response to Local Call History is enabled. 3\. Triggered automatically at end of call. When using this method, we recommend enabling the setting Save Response to Local Call History. The response will then be shown in the call history panel. 4\. Triggered manually by pressing a keyboard shortcut. When using this method, the response can be shown on the Call Details panel or the call history panel (if the setting Save Response to Local Call History is enabled). 5. Triggered by [Call Transfer Summary](/settings/transfer.md#call-transfer-summary). When using this method, the response can be sent along in the chat to the transfer recipient. {% hint style="info" %} If the AI Prompt is set to display on Call Details panel, the display order of the prompts can be changed. Select a shortcut from the list and use the up or down arrow to adjust its position. {% endhint %} {% hint style="info" %} In the transcription sent along with the AI Prompt, a star (\*) will denote responses from the attendant console user. Use this indicator to determine if the agent's transcription will be evaluated. {% endhint %} ### AI Prompt | Option | Description | | -------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | Display Name | This display name of the AI prompt. | | Description | Optional description of the AI Prompt. | | Prompt | This is the prompt that will be sent to AI. | | Display on Call Details | Displays this prompt as an option on the [AI Prompts tab on the Call Details panel](/daily-usage/call-details.md#ai-prompts-preview) | | Include as Option for Callback Reminders | Displays this prompt as an option on the [chat callback reminder, email callback reminder, chat consult transfer templates, and call transfer summary](/settings/transfer.md#call-back-messages) as a placeholder. | | Save Response to Local Call History | Save the AI prompt response to local history. This will allow the response to be viewed in the Call History panel. | | Triggers: End of Call | Triggers the AI prompt at the End of the call. When using this trigger, we recommend enabling Save Response to Local Call History setting. The response can then be viewed in the call history panel. | | Triggers: Keyboard Shortcut | Triggers the AI prompt by pressing the specified keyboard shortcut. When using this method, the response can be shown on the Call Details panel or the call history panel (if the setting Save Response to Local Call History is enabled). | | Triggers: Utterance ([Private Preview](/appendix/feature-states.md)) | Triggers the AI prompt when a keyword is mentioned in the call. The current phrase containing the keyword and the previous ten phrases will be sent in the prompt. Each new phrase will continue to trigger the prompt until the keyword is not included in the previous 10 phrases. | #### Adaptive Card Adaptive cards can be configured to be displayed as the AI prompt response. First, the AI prompt is required to return a JSON object. See example Below: ``` Summarize this call and return JSON object formatted like: { “Summary”: “CALLSUMMARYHERE” } ``` Second, the properties in the JSON object can be referenced in the adaptive card: ${PropertyName} Adaptive cards can be designed here: See example adaptive card below that can be used with the prompt above: ``` { "type": "AdaptiveCard", "body": [ { "type": "TextBlock", "size": "Medium", "weight": "Bolder", "text": "Call Summary" }, { "type": "TextBlock", "text": "${Summary}", "wrap": true } ], "$schema": "http://adaptivecards.io/schemas/adaptive-card.json", "version": "1.5" } ``` ### Advanced Settings Enable Real-Time Speech to Text will allow closed captions to function on the [Call Details Panel](/daily-usage/call-details.md) Save transcription to Local Call History. See [Call History](/daily-usage/call-history.md#closed-captions-preview) Enable AI Transfer Advisor ([Private Preview](/appendix/feature-states.md)) #### AI Connection (optional) **Additional permissions are required to connect to your own Azure OpenAI model. See** [**Permissions**](https://ac.docs.landis.cloud/settings/pages/MXfcsBnbFZNffCP8ReLj#https-cognitiveservices.azure.com-user_impersonation) These settings give you the ability to connect to your own Azure OpenAI model for the AI prompts. 1. Create or use an existing Azure OpenAI model. 2. Grant permissions to Foundry resource. 1. Grant the attendant console users the Cognitive Servises OpenAI User role. 3. Copy the Endpoint Target URI from the model. Use the instructions below to parse the URL and gather the required information to populate the setting fields. 1. Endpoint URL: The protocol and domain name of the URL 2. Deployment Name: The URL pathname after /deployments 3. API Version: The contents of the api-version URL parameter Example Endpoint Target URI: /openai/deployments/o4-mini/chat/completions?api-version=2025-01-01-preview Endpoint URL: Deployment Name: o4-mini API Version: 2025-01-01-preview --- # Agent Instructions: 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: ``` GET https://ac.docs.landis.cloud/settings/ai-prompts.md?ask= ``` The question should be specific, self-contained, and written in natural language. 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.