Redis, a data platform, announced a deeper integration with Amazon Bedrock aimed at enhancing the quality and reliability of generative AI applications. This builds on last year's successful integration of Redis Cloud as a knowledge base for building Retrieval-Augmented Generation (RAG) systems. Redis offers market-leading vector search performance and remains one of only three software vendors included in the Amazon Bedrock console.

Amazon Bedrock has introduced a RAG evaluation service that automates and costs efficiently evaluates LLM-generated responses. This service uses foundation models from Amazon and other leading AI providers, which helps developers assess responses accurately, reducing the occurrence of hallucinations. This new capability allows generative AI applications to be tailored more effectively to diverse use cases.

Retrieval-Augmented Generation combines specific data retrieval with the generative abilities of LLMs. Redis Cloud functions as a fast vector database for RAG, facilitating the storage and retrieval of vector embeddings that supply LLMs with timely and relevant information. This integration enables developers to link LLMs from the Bedrock console directly to their Redis vector database, simplifying workflows.

Despite these improvements, evaluating RAG systems remains a complex challenge. Developers struggle with assessing various aspects like text chunking strategies and model selections, often requiring extensive human evaluations to ensure quality. This makes the evaluation process both time-consuming and costly.

Manvinder Singh, VP of AI Product at Redis, stated, “When customers need fast and reliable vector search in production, they turn to us. However, LLMs are still prone to hallucinations. Our expanded partnership with Amazon Bedrock gives devs a powerful tool to create more accurate and trustworthy generative AI apps.”

Redis, founded in 2011, has established itself as a leading data platform with over 10,000 customers leveraging its solutions. The company is headquartered in San Francisco, with locations in Austin, London, and Tel Aviv.