<\/span><\/h3>\nAmazon Q in Connect is a generative AI-powered customer service assistant integrated within Amazon Connect. Unlike Amazon Connect Wisdom, which surfaces documents and information from knowledge bases, Amazon Q uses large language models (LLMs) to deliver real-time, conversational, and contextual recommendations<\/strong>. It not only detects customer intent using conversational analytics but also generates proactive solutions and suggests actions that agents can take instantly.<\/p>\n
\n<\/span><\/span>2. What are the key components you can customize in Amazon Q in Connect\u2019s generative AI system?<\/strong><\/span><\/h3>\nYou can customize AI prompts, AI guardrails, and AI agents<\/strong>:<\/p>\n\n- \n
AI prompts<\/strong> define what the model should do, e.g., answering questions or summarizing a conversation.<\/p>\n<\/li>\n- \n
AI guardrails<\/strong> enforce responsible AI usage by filtering harmful content, redacting sensitive data, and limiting hallucinations.<\/p>\n<\/li>\n- \n
AI agents<\/strong> orchestrate these prompts and guardrails into structured workflows for search, self-service, or agent assist scenarios.<\/p>\n<\/li>\n<\/ul>\n
\n<\/span><\/span>3. What are the supported channels for Amazon Q in Connect?<\/strong><\/span><\/h3>\nAmazon Q in Connect works with Voice, Chat, and Email<\/strong> channels. However, it does not support the Task channel<\/strong>. To prevent issues, it\u2019s recommended to use a Check contact attributes block<\/strong> to exclude tasks from reaching the Q in Connect block.<\/p>\n
\n<\/span><\/span>4. How does Amazon Q in Connect assist agents in real-time?<\/strong><\/span><\/h3>\nAmazon Q in Connect listens to calls or reads chat messages and automatically detects customer intent<\/strong>. It then uses natural language understanding (NLU) to recommend articles, actions, or responses<\/strong>. Agents can also ask questions directly in natural language to receive instant answers.<\/p>\n
\n<\/span><\/span>5. What is the role of Contact Lens in enabling Amazon Q in Connect?<\/strong><\/span><\/h3>\nContact Lens real-time analytics<\/strong> is essential for enabling Q in Connect with voice calls. It processes call transcripts in real-time to detect intent and provides relevant recommendations. For chat interactions, Contact Lens is not required.<\/p>\n
\n<\/span><\/span>6. Can Amazon Q in Connect be used for customer self-service? Explain how.<\/strong><\/span><\/h3>\nYes. It can be integrated with Amazon Connect bots and flows<\/strong> to provide AI-driven self-service. It can answer customer questions, complete tasks (like rescheduling appointments), and, if needed, escalate to a live agent while preserving the interaction context.<\/p>\n
\n<\/span><\/span>7. What are the default tools in Amazon Q in Connect for self-service?<\/strong><\/span><\/h3>\nThere are four out-of-the-box tools:<\/p>\n
\n- \n
QUESTION<\/strong> \u2013 Answers queries.<\/p>\n<\/li>\n- \n
ESCALATION<\/strong> \u2013 Escalates to an agent.<\/p>\n<\/li>\n- \n
CONVERSATION<\/strong> \u2013 Engages in small talk.<\/p>\n<\/li>\n- \n
COMPLETE<\/strong> \u2013 Ends the session when the issue is resolved.<\/p>\n<\/li>\n<\/ul>\n
\n<\/span><\/span>8. How do you route conversations in flows based on Amazon Q in Connect\u2019s decisions?<\/strong><\/span><\/h3>\nYou use the Check contact attributes flow block<\/strong> to check the selected tool (saved as a Lex session attribute) and create conditional branching<\/strong>. This enables routing decisions like escalation or call completion.<\/p>\n
\n<\/span><\/span>9. What content types does Amazon Q in Connect support for ingestion?<\/strong><\/span><\/h3>\nIt supports HTML, DOCX, PDF, and plain text (UTF-8)<\/strong> files. Files must be under 1 MB<\/strong> and not password protected. PDFs must not contain embedded scripts or encryption.<\/p>\n