Customer Support Agent

Agent Description:

The Customer Support Agent automates the analysis of customer service tickets to help support teams identify common issues, track resolution metrics, and improve response strategies. It connects real-time APIs to fetch ticket data, identifies recurring problems and performance bottlenecks, and returns both analytical summaries and customer-facing resolutions.

Purpose and Components
  • Purpose: To process real-time support ticket data, analyze trends and performance, and assist with resolution insights and metrics.
  • Components:
    • A ticket fetcher that extracts support tickets from an external system.
    • An analysis engine that highlights patterns, calculates metrics, and suggests improvements.
    • A resolution generator for returning clear, structured insights and recommendations.
Supported Capabilities
  • Fetching real-time customer support ticket data from an external API.
  • Structuring and cleaning ticket data for analysis.
  • Identifying frequent issue types and customer pain points.
  • Calculating key performance metrics (for example, resolution rate).
  • Highlighting unresolved or long-open tickets for escalation.
  • Generating user-facing resolution suggestions.
LLM Used
  • Google Vertex
    Note: To learn more about the LLM and to modify its behavior, refer to the Configuring LLM settings section.

Sub-Agents

1. Ticket Summary Generator

  • Role:Data Extractor.
  • Scope:Connects to the customer support platform API to retrieve and format recent tickets.
  • Description: This sub-agent makes a GET request to an API endpoint and processes the response. It extracts and cleans the following fields:
    • Ticket ID

    • Customer Name

    • Issue Category (for example, Billing, Technical, Order)

    • Description

    • Timestamp

    • Status (Open/Resolved)

      The structured output is passed to the analysis agent.

  • LLM Used: Google Vertex (inherits from parent).
  • Tool Used: GET Request Tool

2. Customer Support Agent

  • Role:Insight Generator.
  • Scope:Analyzes structured ticket data and generates reports on trends and performance.
  • Description: This sub-agent receives structured ticket data and performs the following:
    • Identifies top issue categories and keywords.

    • Calculates the resolution rate: Resolved / Total Tickets.

    • Flags tickets that remain unresolved or have been open for too long.

    • Identifies repeat customers or recurring complaint patterns.

    • Generates a concise summary paragraph followed by:

      • A table of key metrics

      • A resolution guide with recommendations or responses to common issues

        It uses PDF input via MCP Client Tool to enhance resolution outputs based on reference documents (for example, knowledge base or prior ticket responses).

  • LLM Used: Google Vertex (inherits from parent).
  • Tool Used: MCP Client Tool
Tools Used:
  • GET Request Tool: Used by the Ticket Summary Generator to fetch real-time support tickets.
  • MCP Client Tool: Used by the Customer Support Agent to enhance ticket resolution using external reference documents (for example, support playbooks or PDFs).
Note: For details on modifying the Tools, refer Tools Library section.
Agent Workflow Behavior Summary
  1. The Ticket Summary Generator connects to an API to fetch recent customer support tickets.
  2. It extracts and cleans relevant fields, formatting the ticket data for further processing.
  3. The Customer Support Agent analyzes the structured data to:
    • Identify the most frequent issue categories and keywords.
    • Calculate performance metrics like resolution rate.
    • Flag unresolved or aging tickets for review.
    • Detect recurring customers or patterns.
  4. The agent then generates:
    • A concise summary of findings.
    • A structured metrics table showing trends and resolution status.
    • A resolution block offering customer-facing suggestions or instructions using PDF-based knowledge (via MCP Client Tool).
Sample Questions:
  • What are the most frequent types of customer queries in the last 30 days?
  • I updated my payment method, but my invoice still failed—can you help?
  • I keep receiving error 503 when I try to log in to my dashboard
  • I have been charged twice give me resolution for this