Semantic caching layer for AI SDK. Automatically generates embeddings for prompts, stores them in Upstash Vector and Redis, and reuses semantically similar responses to reduce token costs.
