Kong AI Gateway 3.10 introduces significant enhancements aimed at improving AI governance and developer productivity, primarily through automated Retrieval-Augmented Generation (RAG) to reduce large language model (LLM) hallucinations and comprehensive PII sanitization to ensure data privacy. The new AI RAG Injector plugin automates data retrieval processes, minimizing developer workload while increasing response accuracy by providing LLMs with vetted information. The update also includes built-in support for sanitizing personally identifiable information across multiple languages, offering a reliable and scalable solution to maintain compliance and protect user data. Additionally, the release enhances model consumption with native SDK support, allowing seamless integration with existing client libraries and facilitating easier migration to Kong's platform. The introduction of cost-based load balancing optimizes AI model usage based on token consumption and cost, while the expanded support for vector databases with pgvector offers flexibility in storing embeddings. These updates collectively enable organizations to streamline AI operations, enhance security, and more efficiently manage AI workloads.