The companion agent works as an intelligent partner alongside human agents in real time. It can understand text and handle a variety of customer service conversations from address changes to billing investigations. It incorporates assistive features in one place to provide seamless collaboration.
Key benefits
- Inclusive interface: Manages conversations with both automatic suggestions and manual searches on a single screen.
- Updated integration: Handles all Agent Assist features. You can add new features or connections without building entirely from scratch.
- Extensive tool support: Supports next generation assistant tools like Grounding with Google Search (GWGS), OpenAPI, and datastore.
- Flexibility: Companion agent can route tasks and work through problems independently, so you can scale new integrations as your needs grow.
Companion agent features
The companion agent works with the resources and choices you provide to understand the end user's intent. It guides conversations to the intended outcome, in three phases:
- Configuration
- Decision
- Results
Configuration
The configuration phase is when you provide resources like knowledge documents, instructions, tools, and intended outcomes for the agent.
Decision
The companion agent uses an LLM to respond to your configuration. The LLM utilizes all provided resources, including preset rules, knowledge documents, and tools, to analyze human agent and end-user conversation, generate suggestions, and automate tasks to guide the conversation toward the intended outcome.
Results
The companion agent displays the results in a unified interface that seamlessly merges the automatic suggestions and conversational assistance. Human agents can see and interact with the interface.
Visual interface
The companion agent displays the following windows:
- Next best action
- Interactive panel
Next best action
When you start a customer service conversation, the next best action panel offers suggestions through a guidance agent or a knowledge agent. These suggestions are the actions the human agent can take during the conversation. Some of the next best actions also show related resources. These resources extract information from the knowledge base or Google Search.
Interactive panel
The interactive panel functions as a conversational assistant for the human agent. In this panel, a human agent can enter questions related to the conversation.
The interactive panel consists of the following tabs:
Your assistant
Human agents can broadly ask questions about the conversation, search a knowledge base or Google, and find a tool to resolve customer queries. The companion agent takes the answer directly from the knowledge base and gives related suggestions to the human agent in this panel. If that agent asks something unrelated to the conversation, the companion agent performs a web search.
Summary
The conversation summary is generated in this tab if you click generate summary. The summary can consist of sections situation, action, outcome, resolution, and customer satisfaction. A human agent can also access the previous versions of the summary through the history panel.
Settings
On this tab, a human agent can change to dark or light mode.
Grounding with Google Search (GWGS)
The companion agent's Google Search tool uses GWGS — a generative AI feature on Gemini Enterprise Agent Platform to generate suggestions grounded in real-time Google Search results. Grounded results are generated based on end-user prompts, customer-provided context, and Google Search engine data, alongside recommended search suggestions.
By utilizing real-time web information, GWGS enables the companion agent to provide accurate, data-driven responses for topics involving current news, global knowledge, or dynamic subjects, keeping conversations rooted in factual, up-to-date content.
The use of the companion agent Google Search tool is subject to Section 20(k) (GWGS) of the Google Cloud Platform Service Specific Terms.