Performance Insights

Performance Insights automatically generates a concise summary report at the end of each call, providing an overview of the agent’s performance. The report highlights the following key areas:

• Summary
• Improvement Areas
• Follow-up Actions
• Positive Feedbacks
• Negative Feedbacks
• Compliance with Notifications

Configurations on CA
Performance insights report can be managed through the RTG settings on CA.

The report can be customized with the sections a customer wants to see at the end of the call. By default, when the performance insights is enabled, the only section included is "Summary". Other sections can be selected from the dropdown list:

Configurations on Dataflow
Performance insights need to be added using the Conversation Insights node on the dataflow project being used with the copilot application.

The input outputs for this node will be according to the insight modules that are required. For example, the "Compliance with Notifications" module needs to receive the RTG Notification event, which is provided by the CA Realtime node. Therefore, the output of CA Realtime node will need to be connected to the Conversation Insights node in order to obtain the "Compliance with Notifications" output, otherwise the field will be empty on the report. More details on this node can be found in the Conversation Insights.

ML service
A request is sent to the POST /v1/conversation-insights API from data flow in order to generate the performance insights from the ML service.

Sample request:

curl -X 'POST' \
  'http://192.168.20.59:55059/v1/conversation-insights' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
  "inputs": [
    {
      "dialogue": {
        "id": "string",
        "participants": [
          {
            "id": "agent",
            "role": "agent"
          },
          {
            "id": "customer",
            "role": "customer"
          }
        ],
        "version": "1.0.0",
        "metadata": {},
        "transcript": [
          {
            "id": "0",
            "participantId": "agent",
            "content": "Thank you for calling Power Telecom. It’s Carlos speaking. How can I help you today?",
            "metadata": {}
          },
          {
            "id": "1",
            "participantId": "customer",
            "content": "Hi Carlos. I looked at the plans on the website and I think I want to change my plan. I would like to upgrade to Starter Seven. I keep running out of minutes and gigabytes with my current plan.",
            "metadata": {"moment": {
                                    "type": "frustration_detected",
                               "level": "medium",
            "action_recommended": "Empathize and offer assistance quickly"
          }}
          },
          {
            "id": "2",
            "participantId": "agent",
            "content": "I totally understand that, sir. Starter Seven is a great plan. Please give me a second to pull up your account.",
            "metadata": {}
          },
          {
            "id": "3",
            "participantId": "customer",
            "content": "OK",
            "metadata": {"moment": {
                                    "type": "plan_switch_request",
                               "level": "medium",
            "action_recommended": "Change the customer'\''s plan."
          }}
          },
          {
            "id": "4",
            "participantId": "agent",
            "content": "Perfect! It’s done. Anything else I can help you with today?",
            "metadata": {}
          },
          {
            "id": "5",
            "participantId": "customer",
            "content": "Thank you! That was fast and easy! Please forward me the details of this plan change via email. Thank you for your assistance.",
            "metadata": {}
          }
        ]
      }
    }
  ],
  "model": "",
  "language": "",
  "max_tokens": 2048,
  "temperature": 0,
  "reasoning_effort": "",
  "insights": {},
  "provider_settings": {
    "provider": "openai",
    "OPENAI_API_KEY": "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",
    "AZURE_API_KEY": "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",
    "AZURE_API_BASE": "",
    "AZURE_API_VERSION": "",
    "AWS_ACCESS_KEY_ID": "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",
    "AWS_SECRET_ACCESS_KEY": "string",
    "AWS_REGION_NAME": "string",
    "GEMINI_API_KEY": "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",
    "TOGETHERAI_API_KEY": "string",
    "GROQ_API_KEY": "string",
    "VERTEXAI_PROJECT_ID": "string",
    "VERTEXAI_LOCATION": "string",
    "VERTEXAI_PRIVATE_KEY": "string",
    "VERTEXAI_CLIENT_EMAIL": "string",
    "VERTEXAI_TOKEN_URI": "string"
  }
}'

Sample response for this request:

{
  "results": [
    {
      "id": "string",
      "summary": "A customer called to upgrade their plan due to running out of minutes and data. The agent quickly assisted by changing the plan to Starter Seven. The customer expressed satisfaction with the fast service and requested an email confirmation of the plan change.",
      "topic": "Plan upgrade",
      "conversation_result": "Customer satisfied",
      "non_resolution_risk": "Low",
      "actions": [
        "Changed customer's plan",
        "Pulled up customer's account",
        "Agreed to send email confirmation"
      ],
      "positive_feedbacks": [
        "The agent quickly understood the customer's needs and provided a solution.",
        "The agent was efficient in changing the plan.",
        "The agent maintained a positive and helpful tone throughout the conversation."
      ],
      "negative_feedbacks": [],
      "improvement_areas": [],
      "follow_up_actions": [
        "Send email confirmation of the plan change to the customer"
      ],
      "moment_compliance_summary": [
        {
          "action_recommended": "Empathize and offer assistance quickly",
          "complied": "yes"
        },
        {
          "action_recommended": "Change the customer's plan.",
          "complied": "yes"
        }
      ],
      "churn_risk": "Low"
    }
  ]
}