Agent Network Documentation

This document provides an overview of the agents used in the customer feedback categorization network.

Agent Network Overview

The agent network consists of two agents:

1. **Coding Agent:** This agent is responsible for the initial categorization of the customer feedback. 2. **QA Agent:** This agent reviews the categorization of the Coding Agent and provides feedback if necessary.

Agent Configurations

Coding Agent

• **Purpose:** To analyze the customer feedback and assign a primary and secondary coding based on the provided categories.
• **LLM Configuration:** Uses the Azure OpenAI `gpt-4o` model by default.
• **Input:** Customer feedback snippet (text).
• **Output:** A JSON object with the following keys: `primary_coding`, `secondary_coding`, `reason`.

QA Agent

• **Purpose:** To ensure the quality and accuracy of the categorization performed by the Coding Agent.
• **LLM Configuration:** Uses the Azure OpenAI `gpt-4o` model by default.
• **Input:** The JSON output from the Coding Agent.
• **Output:**
• `APPROVED` if the categorization is correct.
• Feedback to the Coding Agent if the categorization is incorrect.