This document describes how a custom widget in the WxCC agent desktop, or a third-party custom desktop, can interact with the Serving API service. This interaction allows the retrieval of Virtual Agent transcripts and the Virtual Agent transfer summary when a call is transferred from a virtual agent to a human agent.

anchorPrerequisites

1. An access token with the cjp:user scope is required. For details about OAuth token generation, please refer to this link.
2. The interaction ID of the current call is needed. For a custom widget in the WxCC Agent desktop, please refer here. For third-party desktops, please refer here.

Note: The Interaction ID or Task ID is the same as the Conversation ID mentioned in this document.

anchorServer host details

To retrieve the server details:

1. Construct the following REST request:

   • URL: https://u2c.wbx2.com/u2c/api/v1/user/catalog

   • Method: GET

   • Query Params:

     Parameter	Value
     types	WCC
     services	wccai-serving-api

2. Add an Authorization header.

   Example: Authorization: Bearer <access_token>

3. Execute the REST request.

Here's an example response:

{
  "services": [
     {
     "serviceName": "wccai-serving-api",
     "logicalNames": [
       "https://serving-api-streaming.wxcc-us1.cisco.com:443"
      ],
     "serviceUrls": [
        {
          "baseUrl": "https://serving-api-streaming.wxcc-us1.cisco.com:443",
          "priority": 5
        }
     ],
     "internalServiceUrls": [],
      "ttl": -1,
     "id": "urn:WCC:US1:wccai-serving-api"
   }
  ],
 "format": "servicelist"
}

anchorVirtual Agent transcripts API

This is a server-side gRPC streaming API that takes flags and a conversation ID as input. It returns the transcripts of the conversation between the caller and the virtual agent (Bot) in a historical manner. You can find the proto definitions here.

1. Use the RPC, StreamingInsightServing.

   • Input- StreamingInsightServingRequest
   • Output - stream StreamingInsightServingResponse

   Here's an example StreamingInsightServingRequest request payload for retrieving the VA and caller transcripts:

       {
           "insightServingRequest": {
           "agentDetails": {
             "agentId": "1234"
           },
           "conversationId": "4801c6ce-7a58-4704-b674-1748d53b7085",
           "orgId": "61194836-4001-42d3-bea8-56df1460e522",
           "historicalVirtualAgent": true,
           "historicalTranscripts": true
       }
   

2. The response will be an array of transcripts, including both the speech transcripts of the virtual agent and the caller.

Here's a sample response object for the Virtual Agent transcript:

{
  "orgId": "orgid_01",
  "conversationId": "conversation_01",
  "roleId": "roleid_02",
  "utteranceId": "utterance_04",
  "role": "IVR",
  "insightType": "virtual_agent",
  "serviceProvider": "nuance",
  "startTimestamp": 1612868857983,
  "endTimestamp": 1612868857983,
  "publishTimestamp": 1612868857983,
  "isFinal": true,
  "messageId": "messageId_01",
  "configId": "Test123",
  "languageCode": "en-US",
  "responseContent": {
     "replyText": [
      "Your flight from Bangalore to Delhi on August 31st for 85 people is booked. Thank you!"
    ],
     "agentTransfer": {
     "metadata": {
        "userData": {
           "timezone": "Asia/Kolkata"
         }
       }
     }
   }
}

And here's a sample response object for the Caller transcript

{
  "orgId": "org_01",
  "conversationId": "conversation_01",
  "roleId": "roleid_01",
  "utteranceId": "utterance_01",
  "role": "IVR",
  "insightType": "transcript",
  "serviceProvider": "google",
  "startTimestamp": 1612862799876,
  "endTimestamp": 1612862799876,
  "publishTimestamp": 1612862799877,
  "isFinal": true,
  "messageId": "mesage_id_01",
  "configId": "configid_01",
  "languageCode": "en-US",
  "responseContent": {
    "alternatives": [
      {
        "transcript": "I would like to book a flight to New york city."
      }
    ],
    "result_end_time": {
      "nanos": 360000000
    },
    "messageType": "TRANSCRIPT"
  }
}

anchorVirtual Agent Call Summary API

This is a unary gRPC API that returns the call summary of the conversation between the caller and the virtual agent (bot), based on the conversation ID. You can find the proto definition here.

1. Use the RPC, InsightServing.
   • Input - InsightsServingRequest
   • Output - InsightsServingResponse
2. Set InsightType to CALL_INSIGHTS (Enum value - 5).

Here's an example request payload for retrieving the agent handoff summary:

{
  "conversationId": "f4908103-82ef-405c-8f51-49f4de28181b",
  "insightType": 5,
  "messageId": "virtual-agent-call-summary",
  "orgId": "61194836-4001-42d3-bea8-56df1460e523"
}

And here's a sample response for the call summary:

{
  "orgId": "org_01",
  "conversationId": "conversation_01",
  "roleId": "roleid_01",
  "utteranceId": "utterance_01",
  "role": "IVR",
  "insightType": "call_insights",
  "serviceProvider": "google",
  "startTimestamp": 1612862799876,
  "endTimestamp": 1612862799876,
  "publishTimestamp": 1612862799877,
  "isFinal": true,
  "messageId": "virtual-agent-call-summary",
  "configId": "configid_01",
  "languageCode": "en-US",
  "responseContent": {
    "content": "{\"hand_off_reason\":\"talk to an agent\",\"details\":\"The caller's reason for the call is to block their credit card. The call was handed off to a live human agent because the caller requested to talk to an agent.\",\"call_reason\":\"block credit card\"}",
    "callInsightType": "VA_CALL_SUMMARY"
  }
}