Design a desktop web app "AI Support Insights" dashboard for an AI-powered customer service team. Viewport: 1440px width desktop. Goal: Give support leaders a clear view of how the AI Agent and human agents are performing, which topics are working well, and where to improve content or workflows. Layout: - Top filter bar: - Date range picker - Channel filter (Chat, Email, Phone, Messaging) - Segment filter (Free, Pro, Enterprise) - First row: KPI cards (4 cards) - AI Resolution Rate - Deflection Rate (tickets resolved by AI without human handoff) - Avg. Time to Resolution - AI-predicted CSAT - Second row: Two main charts - Left: "Conversations handled by AI vs Human" - Stacked bar chart by day/week. - Right: "Top Intents & Success Rate" - Bar chart showing intents (Refund, Shipping, Login, Billing) with accuracy / resolution metrics. - Third row: Quality & Escalation - Left: "Escalation Reasons" list - Table showing: Intent, Escalation rate, Common failure reason. - Right: "Conversation Quality Samples" - Table with sampled conversations: - Intent - AI Quality score (auto QA) - Manual QA status (Pass / Needs review) - Link/button "Open conversation" Visual style: - Clean analytics UI with a white background and subtle card shadows. - Use a consistent color palette across charts (AI vs Human, Positive vs Negative). - Typography: crisp, data-friendly sans-serif. - Minimal decoration; focus is on clarity and interpretability. Add: - A small callout at the bottom right: "Suggested improvements", listing: - "Add article: Refunds above $500" - "Improve workflow: Damaged package claims" - Each with a button "View details".