Introduction to Conversational Agents

Introduction to Conversational Agents

Introduction

This hack will be an introduction to the world of Customer Engagement modernization. We will teach you how to build a robust conversational agent using Google Cloud’s Customer Engagement Suite.

You will be building a conversational agent for the HR department of a startup named Piped Piper. Your agent will be the central place for employees to find authoritative answers on company policies as well as place to get data from other internal systems (eg. Workday for PTO balance, etc). This agent will demonstrate how a company can increase employee productivity by centralizing access to knowledge; from data stored in internal data sets to API driven systems, all of this can be accessed using natural language from a single chat.

Learning Objectives

In this hack you will be standing up a conversational chat agent and making it useful in many instances to help Piped Piper’s employees. More specifically we will be exploring the tool for building agents called Conversational Agents (formerly DialogFlow CX) which is part of Google Cloud’s Customer Engagement Suite (CES).

You will:

1. Create your first agent
2. Set up a basic Playbook to detect an escalation scenario
3. Set up an authoritative knowledge base using internal HR policy documents to help answer questions
4. Create an external system API call for data we can use in responses

Challenges

• Challenge 1: First Agent On The Scene
  • Create and initialize a Conversational Agent
• Challenge 2: Following the Playbook
  • Set up a basic Playbook to detect an escalation scenario
• Challenge 3: Getting Smarter
  • Give the agent access to PDF documents to use in chat answers
• Challenge 4: Agent Phone Home
  • Call an external system’s API for even more knowledge

Prerequisites

• Basic knowledge of GCP
• Access to a GCP environment

Contributors

• Roman Mezhibovski
• Gino Filicetti
• Lexi Flynn