Configure the Customer Assist Agent

Completed

The Customer Assist Agent is the frontline of your AI-powered contact center. It handles customer interactions by combining autonomous self-service with real-time assistance to service representatives, resolving routine issues instantly while ensuring complex cases get the context they need for a smooth handoff.

The agent works across voice, live chat, SMS, and social messaging channels, providing consistent service regardless of how customers reach you.

Core capabilities

The Customer Assist Agent operates across three key areas:

  • Understand customer intents and resolve issues: The agent uses generative AI to understand what customers need, conduct dynamic interviews to clarify intent, and resolve issues autonomously using your knowledge base and configured tools.
  • Assist representatives during live interactions: When escalation is needed, the agent hands off the conversation with full context—intent captured, sentiment analyzed, conversation summary assembled—so representatives can pick up seamlessly.
  • Proactively contact customers: The agent initiates outbound voice and SMS interactions to reach customers at the right moment, handling routine outbound communications autonomously while connecting representatives when needed.

Voice architecture tiers

The Customer Assist Agent supports three tiers of voice interaction, each representing a different level of AI sophistication:

Tier Architecture How it works
Classic voice Speech-to-text/text-to-speech with classic orchestration Transcribes speech to text and responds with text converted to speech. Limited to predefined conversation topic flows and expected answers.
GenAI voice Speech-to-text/text-to-speech with generative orchestration Uses the same speech pipeline but introduces generative orchestration, enabling conversations that go beyond predefined flows.
Premium GenAI voice Speech-to-speech with generative orchestration Uses real-time speech language models (LLMs) that actively listen, perform inference, and handle interruptions. Responses have subsecond latency, similar to speaking with another person.

Classic voice works well for structured, compliance-heavy workflows where deterministic outputs are required, such as regulatory disclosures. GenAI and Premium GenAI voice excel for open-ended interactions where customers describe problems in their own words.

Tip

The most effective approach often combines both: deterministic control for structured, auditable outcomes and generative intelligence for flexible, adaptive reasoning. You don't have to choose one tier for your entire contact center.

Configure proactive engagement

Proactive engagement lets your contact center initiate outbound communications through voice and SMS channels, reaching customers at the right moment rather than waiting for them to call in.

Set up an outbound workstream

  1. Create a workstream and select the Outbound option.
  2. In the Setup Behaviors section, select Setup.
  3. On the Phone number page, select a number from the Shared numbers list.
  4. Configure the language and outbound behaviors.
  5. Configure work distribution. Optionally, add an AI agent to the workstream and configure representative notifications.

Configure proactive engagement settings

  1. In the Copilot Service admin center, go to Productivity under Support experience. Select Manage for Proactive engagements, then select New.
  2. The Create new proactive engagement wizard guides you through these steps:
    • Audience: Choose how to source customer contact information—upload a file, use the CCaaS API, or use MCP (preview).
    • Engagement type: Select Led by representatives (always connect to a human) or Led by AI (let the agent interact first, then add representatives when needed).
    • Dialing modes: Select Copilot (AI-led, high-volume), Preview (representative reviews before calling), Progressive (calls match available representatives), or Predictive (model-optimized call volume).
    • Frequency limits: Set maximum engagements per day and per week, and configure quiet hours by phone number type.
  3. Review settings on the Summary page and select Create.

Note

A single workstream can run both AI-led and representative-led proactive engagements at the same time.

With the Customer Assist Agent configured, your contact center can handle customer interactions autonomously across channels, assist representatives with full context during escalations, and proactively reach out to customers when the moment is right.