Overview

Agent Assist in SmartReach is a real-time support tool integrated into the Agent Desktop. Agent Assist uses speech-to-text and machine learning to guide agents during customer interactions. It enhances efficiency by providing live transcriptions, automated summaries, and contextual assistance, enabling agents to deliver more accurate and effective responses. 

The system provides user-friendly Agent Assist rule creation with predefined templates to accelerate setup. It provides simple mapping of triggers (keywords, intents, events) to actions (suggestions, knowledge, workflows).

Agent Assist has several components:

• Real-time interaction analysis to continuously monitor conversations (voice or chat) to detect intent, sentiment, or key phrases.
• Context and meaning for not just keywords but also relevant suggestions.
• Connected sources, such as knowledge bases, scripts, CRM data, or workflows.
  Information sources based on predefined conditions.

These components ensure that the right guidance appears at the right moment, improving agent efficiency. Agent Assist is highly customizable, allowing you to tailor it to your organization’s workflows.

You can customize Agent Assist according to your requirements:

1. Provide Knowledge Base articles, scripts, and responses.
2. Create suggestions with your brand voice and policies.
3. Create rules based on specific intents, phrases, or scenarios.
4. Provide different suggestions for sales, support, or retention teams.
5. Adjust complexity based on an agent’s experience level.
6. Connect to CRM systems, ticketing tools, or internal knowledge bases.

You can improve performance of your agents with the following best practices:

1. Refine your Knowledge Base by keeping content accurate, concise, and up-to-date, and removing outdated or redundant articles.
2. Create precise rules by using intent-based or contextual triggers where possible and avoiding overly broad triggers that create noise.
3. Leverage analytics and feedback to monitor which suggestions are used vs which suggestions are ignored.
4. Train and align agents to ensure agents understand how to use suggestions effectively and adopt best practices through training.

The main features of Agent assist are listed below:

Action Rules

To enhance agent performance, managers can configure Action Rules, which define specific conditions—such as keywords, intent, or sentiment—that trigger real-time alerts, suggestions, or automated actions. These rules ensure that agents receive timely guidance during interactions while also allowing managers to monitor and influence outcomes through predefined logic. See Action Rules for more information.

Categories

Action Rules are organized into Categories to simplify management and improve usability. Categories allow administrators to create groups of related rules (for example, Compliance, Sales Support, or Customer Experience), making it easier to maintain and apply them effectively across different interaction scenarios.  See Categories for more information.

Action Rules Report

The Action Rules Report provides visibility into how configured rules perform across interactions. It includes key metrics, such as total interactions, number of times rules were applied, and the rule application percentage, enabling managers to analyze effectiveness and optimize rule configurations. See Action Rules Report for more information.

Example Scenarios

1. Use case: Billing inquiries 
   Customer says: “refund,” “charged twice,” “wrong bill”
   Agent Action: 
   1.  Agent Assist shows “Billing Refund Process” knowledge article
   2.  Pre-written response template: “I understand your concern. Let me check your billing details and help resolve this quickly.”
2. Use case: Cancellation request:
   Detected negative sentiment OR cancellation intent.
   Agent Action:
   1. Agent Assist displays a retention script, such as Special offer (for example, discount).
   2. Advice to agent: “Before proceeding, check if the customer qualifies for a retention offer.”
3. Use case: Troubleshooting a technical issue:
   Customer mentions: “app not working,” “login issue”
   Agent Action: Agent Assist shows a guided checklist: 
   1. Confirm the app version.
   2. Ask the customer to clear the cache.
   3. Reinstall if needed.
   4. Escalate if unresolved.
4. Use case: High-value customer handling:
   CRM flag: VIP / high-value customer
   Agent Action: Agent Assist shows you priority handling instructions, such as personalized greeting: “Thank you for being a valued customer, I’ll prioritize your request.”
5. Use case: Regulatory disclosure:
   Customer asks about pricing, contracts, or legal terms
   Agent Action:
   1. Agent Assist Force-displays mandatory script:
      1. Legal disclaimer
      2. Required language

For more information about Agent Assist, see Accessing Agent Assist.