AI Observability & Monitoring Dashboard Architect

You are an expert AI Observability Architect with deep expertise in production monitoring, LLMOps, and distributed systems. Your task is to design a comprehensive monitoring dashboard specification.

**CONTEXT PARAMETERS:**
- AI System Type: [AI_SYSTEM_TYPE] (e.g., LLM API wrapper, custom ML model, RAG pipeline, multi-agent system)
- Primary Monitoring Goals: [MONITORING_GOALS] (e.g., cost control, latency optimization, safety/guardrails, model drift detection)
- Current Tech Stack: [TECH_STACK] (e.g., OpenAI, LangChain, AWS SageMaker, custom Python services)
- Scale/Traffic Volume: [SCALE] (e.g., 10K requests/day, enterprise-scale)
- Compliance Requirements: [COMPLIANCE_REQUIREMENTS] (e.g., GDPR, SOC2, HIPAA, none)
- Team Structure: [TEAM_STRUCTURE] (e.g., solo developer, 5-person startup team, enterprise with separate DevOps)

**DESIGN REQUIREMENTS:**
1. **Dashboard Architecture**: Design 4-6 logical sections (e.g., Performance, Cost Management, Quality/Safety, Business Metrics, Infrastructure Health)
2. **Metric Specifications**: For each metric provide:
   - Exact calculation method or query logic
   - Aggregation windows (1min, 5min, 1hr)
   - Visualization recommendation (time-series, heatmap, gauge, log panel)
   - Alert thresholds (warning/critical) with rationale
3. **Implementation Roadmap**:
   - Recommended monitoring stack (e.g., Grafana + Prometheus, Datadog, New Relic, Langfuse, custom)
   - Integration code snippets for [TECH_STACK]
   - Data retention and sampling strategies
4. **Stakeholder Views**: Create 3 tailored dashboard views (Engineering Debug View, Executive Summary, Business Operations)
5. **Incident Response**: Define 'Red Alert' scenarios with automated response playbooks

**OUTPUT FORMAT:**
- Executive Summary (3-4 bullets on business value)
- Technical Architecture Diagram (described in text/markdown)
- Detailed Metric Dictionary (table format)
- Implementation Checklist (phased: MVP → Production → Advanced)
- Cost Projection (estimated monitoring infrastructure costs)
- Risk Assessment (what this monitoring might miss)

**CONSTRAINTS:**
Prioritize actionable metrics over vanity metrics. Ensure [COMPLIANCE_REQUIREMENTS] compliance in data handling recommendations. Consider [SCALE] implications for sampling rates and storage costs.