Concept 3 of 5

Customer experience with gen AI

4 questions test this

Two offerings sit here, and the exam separates them cleanly. One is about finding things. The other is about handling conversations.

Agent Search

Google quality search applied to your own content rather than the public web. Formerly Vertex AI Search, now Agent Search on Gemini Enterprise Agent Platform, and worth knowing under both names because older material uses the previous one.

It indexes websites, structured records such as a product catalogue, and unstructured documents, and it returns results with the semantic understanding people expect from a search engine. The business cases are a site search that works, a product finder that understands what somebody described rather than what they typed, and internal search across company documents.

Its second job is grounding. Agent Search is what a Google Cloud retrieval setup usually retrieves from, which is why it appears again under grounding.

Google Search appears beside it for the other corpus. When a model needs current public information rather than your own, grounding with Google Search is the mechanism.

The Customer Engagement Suite

Four parts, and the exam asks which one a described situation needs.

Conversational Agents. The customer facing automation. Chat and voice agents that understand what a customer wants and complete it, answering from your content and calling your systems. This is the part that handles a conversation without a person.

Agent Assist. Support for the human agent during a live conversation. Suggested replies, the relevant article surfaced, and a summary written at the end so nobody types up the call. The customer never sees it, and the value is handling time and consistency.

Conversational Insights. Analysis after the fact, across volume. Why people are contacting you, which topics drive the longest calls, where conversations fail. This is the one that informs a roadmap rather than a conversation.

Contact Center as a Service. The contact centre platform itself, the telephony and routing the other three run on, delivered as a managed service.

Reading a question

The separating detail is where the help lands.

Automation the customer talks to is Conversational Agents. Help for the person taking the call, while the call is happening, is Agent Assist. Understanding what is happening across thousands of conversations is Conversational Insights. The underlying platform is Contact Center as a Service.

And if the question is about finding something rather than talking to somebody, it is Agent Search, on your content, or Google Search, on the public web.

Common misconceptions

Agent Assist is a chatbot that replaces the support agent.

It sits beside a human agent during a live conversation, suggesting answers and surfacing the relevant article. The customer facing automation is Conversational Agents. A question about helping the person on the phone is pointing at Agent Assist.

Agent Search is a rebranded Google Search.

It brings Google quality search to your own content, whether that is a website, a catalogue or internal documents. Google Search searches the public web. They appear together in the guide because both can ground a model, on different corpora.

Conversational Insights analyses conversations while they happen.

It analyses them afterwards, across volume, to find why people are calling and where conversations go wrong. The live assistance during a conversation is Agent Assist.

4 questions test this concept

A contact centre wants to reduce handling time by suggesting replies and surfacing the relevant knowledge article to the human agent while a call is in progress. Which part of the Customer Engagement Suite is this?

  • AConversational Agents.
  • BConversational Insights.
  • CAgent Assist.
  • DContact Center as a Service.
Check whether it stuck.

One per page, with a worked explanation.

Start the set
Related material