Summary

Use Workers AI LLM to generate concise summaries of podcast descriptions.

Endpoint Design

GET /podcast/123456?summary=true

Response includes:

{
  "podcast": { ... },
  "summary": "A concise 2-3 sentence summary of what this podcast is about."
}

Implementation Notes

• Use Workers AI LLM (e.g., @cf/meta/llama-2-7b-chat-int8) for summarization
• Cache summaries in KV (they don't change once generated)
• Consider batch processing for popular podcasts
• Limit summary length to 150-200 characters

Approach

1. Fetch podcast description from iTunes API
2. If summary not in KV cache, generate via LLM
3. Store generated summary in KV with long TTL (30 days)
4. Return cached summary on subsequent requests

Acceptance Criteria

• /podcast/123?summary=true returns AI-generated summary
• Summaries are cached in KV
• Graceful fallback if AI unavailable
• Feature-flagged for gradual rollout
• Tests added