Design the complete technical architecture for a Platform Mastery AI Agent product at scale:

Frontend:
- Customer dashboard (Next.js + Vercel AI SDK?)
- Chat interface (streaming, code highlighting, action buttons)
- Platform selection and onboarding wizard

Backend:
- KB ingestion pipeline (already built in Genesis: sitemap → fetch → chunk → embed → Qdrant)
- RAG query engine (Gemini embeddings + Qdrant vector search)
- Agent runtime (Claude API / Gemini API for reasoning)
- Action execution layer (MCP tools per platform)
- User management and billing (Stripe)

Infrastructure:
- Cloudflare Workers (edge caching, rate limiting)
- Elestio PostgreSQL (source of truth)
- Qdrant (vector search)
- Redis (sessions)

Scale considerations:
- Cost per query at 1K, 10K, 100K daily queries
- Multi-tenant architecture
- Platform freshness (re-crawling schedule)

Output a Mermaid architecture diagram + component descriptions.