LLM Integration

Claude / OpenAI / Gemini API integration for content generation, classification, summarisation. Prompt caching to control cost on repeated context. Streaming responses to the UI for first-token latency under 500ms. Retry + fallback model patterns for production resilience.