> ## Documentation Index
> Fetch the complete documentation index at: https://koreai.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.
# Agent AI Integration with NICE CXone Agent Desktop for Digital Chat
This guide explains how to integrate the Agent AI widget with NICE CXone Agent Desktop for digital chat. The integration supports the following Agent AI capabilities:
* Automation
* [Agent Coaching](/ai-for-service/agentai/agent-experience/agent-realtime-coaching)
* [Agent Playbook](/ai-for-service/agentai/agent-experience/playbook)
* [Sentiment Analysis](/ai-for-service/agentai/agent-experience/agent-assist-widget-v3)
* [Conversation transcript](/ai-for-service/agentai/agent-experience/agent-assist-widget-v3#transcript-tab)
* [End-of-conversation summary](/ai-for-service/agentai/agent-experience/agent-assist-widget-v3#generate-summary)
* Custom data passing, such as agent name and ID.
## Responsibilities
| Area | Details |
| ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Shared responsibilities | See [Shared Responsibilities](/ai-for-service/agentai/integration/nice-max/nice-max-desktop-chat-and-voice#shared-responsibilities). Follow only the responsibilities that apply to digital chat. |
| NICE CXOne Admin | See [NICE CXOne Admin](/ai-for-service/agentai/integration/nice-max/nice-max-desktop-chat-and-voice#nice-cxone-admin) for details. |
| Platform Admin | The [support team](https://support.kore.ai/) provisions the Studio script required for the NICE CXone digital chat integration. |
## Prerequisites
Before starting with this integration guide, verify that you have the following:
* A NICE CXone administrator account. The account must be allowed to create digital automatic call distribution (ACD) skills, configure digital points of contact, configure digital chat channels, import/edit studio scripts, and assign agents who have the necessary roles and permissions to accept digital chat interactions in Agent Desktop.
* NICE CXone Studio installed on a supported Windows computer.
* Access to Kore.ai Agent AI, with permission to create or configure an AI agent.
* The following information from the platform administrator or [support team](https://support.kore.ai/):
* **Agent AI widget URL:** The web address used to open the Agent AI panel.
* **Bot ID:** Identifies the AI agent that assists the human agent.
* **Client ID:** Identifies the NICE CXone integration.
* **Client Secret:** A private key used to authenticate the integration.
* **Digital chat Studio script:** The script that connects NICE CXone to Agent AI.
For the meanings of the terms used in this guide, see [Definitions](/ai-for-service/agentai/integration/nice-max/nice-max-desktop-chat-and-voice#definitions).
## Configuration Steps
1. Import and configure the digital chat script in NICE CXone Studio.
2. Create a digital ACD skill in NICE CXone.
3. Create a digital chat channel.
4. Connect the chat channel to the correct skill and studio script.
5. Test the integration from the customer chat entry point.
6. Verify that the Agent AI widget, transcript, custom data, and conversation summary work correctly.
Do not use this guide to configure voice-related settings such as **SIP URI** or **Audio Relay Endpoint**.
### Import and Configure the Digital Chat Script in NICE CXone Studio
The Studio script connects NICE CXone to Agent AI and defines the following:
* Which AI agent to use.
* How to authenticate with Agent AI.
* Which language to use.
* Which chat conversation to connect to.
* What additional information to send to Agent AI.
#### Import the script
1. Download the **[JSON Studio script](https://raw.githubusercontent.com/Koredotcom/docs-v2/2a6bb9632b62d9b0bc518df3cccba08bdff01c1d/ai-for-service/assets/nice-agent-desktop-digital-chat-messaging-sample.json)** Kore.ai provides for digital chat.
2. Open **NICE CXone Studio** on your Windows computer or access [Nice CXone Cloud](https://na1.nice-incontact.com/studio).
3. Select **File** > **Import from File**.
4. Select the downloaded JSON file.
5. Select the **Snippet** tile to open the script editor.
6. Update the following parameters:
* **Kore.ai parameters**: Update **Bot Client ID** (`clientId`), **Bot Client Secret** (`clientSecret`), **Bot ID** (`botId`), and **Agent AI Widget URL** (`baseUrl`). Also update **Bot language code** (`language`, default `en`); see [Supported Languages](/ai-for-service/agentai/agent-experience/supported-languages).
* **Custom data**: Use the custom data object to pass additional context to the AI agent. At run time, the data is available in the AI agent's **UserContext**. See [Access Custom Data in Agent AI Bot](/ai-for-service/agentai/agent-experience#access-custom-data) for instructions.
* Add one key-value pair per line to the custom data object, for example, `customdata.agentName="Agent"` and `customdata.agentID="1234"`.
```javascript theme={null}
ASSIGN global:clientSecret=""
ASSIGN global:clientId=""
ASSIGN global:botId=""
ASSIGN global:baseUrl="https://platform.kore.ai/koreagentassist-sdk-v3/UI/agentassist-iframe.html"
ASSIGN global:language="en"
// 'customdata' object can be utilized to pass additional information to Agent AI Bot for better context
DYNAMIC customdata
customdata.agentName="John Doe"
DYNAMIC scriptparams
scriptparams.clientId="{clientId}"
scriptparams.clientSecret="{clientSecret}"
scriptparams.botId="{botId}"
scriptparams.conversationId="{contactid}"
scriptparams.threadIdOnExternalPlatform="{__threadidonexternalplatform}"
scriptparams.channelId="{__channelid}"
scriptparams.source="nice"
scriptparams.isCall=false
scriptparams.channel="chat"
scriptparams.baseUrl="{baseUrl}"
scriptparams.interactiveLanguage="{language}"
scriptparams.customdata="{customdata.asJSON()}"
ASSIGN scriptparamsjson="{scriptparams.asJSON()}"
scriptparamsjson = scriptparamsjson.urlencode()
```
7. Select **Apply** in the **Snippet Editor**.
8. Select **Save** in the upper-right corner of **Studio**.
### Create a Digital ACD Skill in NICE CXone Setup
An ACD skill is a routing category. It helps NICE CXone decide which agents should receive a chat.
1. Go to the **NICE CXone** home page.
2. Select the **Application Launcher** in the upper-left corner.
3. Select **ACD**.
4. Under **Contact Settings**, select **ACD Skills**.
5. Select **Create New** > **Single Skill**.
6. Set **Media Type** to **Digital** to handle text-based chats instead of phone calls.
7. Select the appropriate **Campaign**.
8. Select **Create**.
### Configure a Digital Chat Channel in NICE CXone Setup
The digital chat channel is the customer-facing entry point where chats begin. Connect the channel to both the Studio script and the ACD skill.
1. On the **NICE CXone** home page, select the **Application Launcher** > **ACD**.
2. Select **Digital** > **Points of Contact Digital**.
3. Select the **Chat** tile.
4. Select **Create Messaging Channel**.
5. Enter a name in the **Channel name** field.
6. On the **Points of Contact Digital** page, select **Edit** for the channel you created.
7. Select the **studio script** and **default skill** for the channel.
8. Select **Save**.
Assign at least one disposition to the digital ACD skill. A disposition is the outcome the agent selects when the conversation ends–for example, **Resolved** or **Closed**.The end-of-conversation summary requires a disposition to appear and remain available. Without a disposition, the chat screen may close automatically when the conversation ends.
## Validate the Digital Chat Integration
1. Go to the **NICE CXone** home page.
2. Select the **Application Launcher**.
3. Select **OMNICHANNEL ROUTING** > **Agent**.
4. Select **Integrated Softphone** > **Connect**.
5. Set the agent status to **Available**.
6. Return to the **NICE CXone** home page and select **Application Launcher** > **ACD**.
7. Select **Digital** > **Points of Contact Digital**.
8. Select the **Chat** tile.
9. Select **Initialization & Test** for your channel.
10. Select the **chat** icon in the lower-right corner of the initialization page.
11. Enter your name in the message initialization form and send a greeting such as **Hi** or **Hello** to start a chat. A chat request appears in **Agent Desktop**.
12. Verify that **Agent Desktop** assigns the chat to you.
13. From the status menu, select **Resolved** or **Closed** to end the conversation.
14. Select **Unassign & Dismiss** to disconnect the chat.
## Validation Checklist
* Verify the Agent AI widget loads in **Agent Desktop** when an agent receives a chat.
* Verify customer and agent messages appear in the conversation transcript.
* Verify the AI agent receives the custom data you configured in itsUserContext.
* Resolve or close the chat and verify the end-of-conversation summary appears.
* Verify **Unassign & Dismiss** disconnects the completed chat.
## Best Practices
* Use separate chat channels and skills for testing and live customer conversations.
* Verify the widget URL, **Bot ID**, **Client ID**, and **Client Secret** before saving the Studio script.
* Keep a record of each version of the Studio script and the changes you make to it.
* After changing the script, skill, or channel connection, start a new test chat. Existing chats may continue using the previous configuration.
* Assign a disposition to the skill so that the conversation summary remains available.
* Before turning on the integration for customers, test the complete process:
* Chat routing
* Agent AI panel opening
* Transcript updates
* Custom data transfer
* Summary generation
* Conversation closure
## Troubleshooting
| **Issue** | **Resolution** |
| ---------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Agent AI screen pop-up doesn’t open | Verify the Agent AI widget URL, **Bot ID**, **Client ID**, and **Client Secret** in the Studio script. Verify that the Studio script you configured associates with the digital channel. |
| Chat is not routed to the agent | - Verify that the digital channel uses the correct skill and studio script.
- Verify that you assigned the agent to the skill and the agent has an **Available** status. |
| Conversation transcript does not update | Verify the Studio script passes the `conversationId`, `threadIdOnExternalPlatform`, and `channelId` values. |
| Custom data is missing | Verify that the custom data object contains valid key-value pairs and that you include it in `scriptparams` as JSON. |
| Conversation summary does not appear | Verify that you assigned a disposition to the digital ACD skill and ended the chat with a **Resolved** or **Closed** status. |
| Studio script changes do not take effect | Save the Studio script, verify that the updated script associates with the channel, and start a new chat session. |
## View Interaction Details in Kore
During and after a digital chat, agents can review the interaction in the Agent AI widget. The widget displays the conversation transcript, conversation summary, and customer interaction history.
For additional conversation details, go to **Kore Platform** > **Automation AI** > **Analytics** > **Conversations**.
## Additional Notes
* The integration uses a screen pop-up to load the Agent AI widget in NICE CXone Agent Desktop; you don’t need Agent Assist Hub configuration.
* This guide supports digital chat only. Voice, SIP, Audio Relay Endpoint, and SIP URI settings do not apply.
* The Studio script passes NICE CXone conversation identifiers and custom data you configured to Agent AI.
* The summary pop-up requires you to assign a disposition to the digital ACD skill.
* After updating the Studio script or channel mapping, start a new chat session to validate the changes.