Invia SMS: GPT Semantic Cache: Reducing LLM Costs and Latency via Semantic Embedding Caching