Analyze the Genesis KB ingestion pipeline's capability to scale to 50+ platforms:

Current pipeline: Sitemap → Fetch → Extract → Chunk → Embed (Gemini 3072d) → Qdrant + PostgreSQL

For each step, identify:
1. What works at current scale (1,398 Telnyx pages)?
2. What breaks at 50 platforms × 5,000 pages = 250,000 pages?
3. Cost at scale (Gemini embedding costs, Qdrant storage, PG storage)
4. Time to ingest a new platform (hours? minutes?)
5. Freshness maintenance (re-crawling strategy)

Optimization opportunities:
- Cloudflare Workers for parallel fetch (edge-distributed)
- Incremental updates (only re-embed changed pages)
- Embedding model selection (cheaper alternatives to Gemini 3072d?)
- Compression and deduplication

Calculate the TOTAL COST to ingest the top 20 platforms.