Pinecone, a leading platform for building AI applications, has recently announced the integration of industry-first inference capabilities into its vector database. This new feature includes fully-managed embedding and reranking models, as well as a unique approach to sparse embedding retrieval. By combining these capabilities with Pinecone’s dense retrieval technology, the platform sets a new standard for AI-powered solutions with cascading retrieval. This advancement aims to improve the development of AI applications, making them up to 48% more accurate and enabling faster and easier creation of AI-driven tools.
One of the key features of this new integration is the introduction of more granular role-based access controls (RBAC), which allows users to set API key roles for enhanced control over data plane operations. Additionally, customer-managed encryption keys (CMEK) provide users with greater control over their data encryption, improving tenant isolation. The platform also offers audit logs for control plane activities and the general availability of AWS PrivateLink for serverless indexes, further enhancing security and performance.
Pinecone’s composable platform now includes the pinecone-rerank-v0 proprietary reranking model, pinecone-sparse-english-v0 proprietary sparse embedding model, and a new sparse vector index type. These new security features and models make Pinecone’s platform even more powerful and versatile for building accurate and scalable AI applications.
Through its collaboration with Amazon Bedrock, Pinecone offers seamless integration that automates the ingestion, embedding, and querying of customer data as part of the large language model generation process. This integration allows customers to quickly generate more grounded and production-grade AI applications while running Retrieval-Augmented Generation (RAG) evaluations natively within Amazon Bedrock, eliminating the need for third-party tools.
Pinecone’s innovative approach combines inference, retrieval, and knowledge base management on a single platform, leading to significant performance improvements and new possibilities for AI application development. Customers can access Pinecone through the AWS Marketplace to accelerate deployment and optimize costs, further empowering developers to deliver better AI solutions.
Pinecone has already helped over 5,000 customers build faster, more accurate AI applications, and with the launch of this new vector database with inference capabilities, it is set to revolutionize the AI industry even further.
