Pinecone and ZoomInfo use Dedicated Read Nodes for real-time recommendations

Pinecone said its next-generation serverless slab architecture and Dedicated Read Nodes became generally available, and it cited a deployment with ZoomInfo that serves real-time contact recommendations. The company framed the update around performance requirements for Artificial Intelligence (AI) applications that vary by concurrency, latency, traffic, and cost.

The release ties the architectural approach to operational limits it attributed to databases that add vector features after the fact and to open-source vector systems that require tuning, cluster management, and ongoing operations. It said on-demand indexes scale elastically with payment tied to queries made, while Dedicated Read Nodes are used for sustained high queries per second and low-latency workloads.

Dedicated Read Nodes, generally available, provide resource isolation and guaranteed warm data to avoid delays linked to cold fetches. Pinecone also described its slab architecture as storing vectors in large contiguous units to reduce fragmentation and support consistent performance.

ZoomInfo selected Pinecone to power a real-time recommendation engine serving personalized contact suggestions across more than 390 million and growing high-dimensional embeddings and over 100,000 namespaces. The release said the Applied AI team moved from prototype to production in weeks, met strict sub-second latency requirements, and reported an order-of-magnitude improvement in throughput during scale testing with Dedicated Read Nodes.

“Pinecone's slab architecture and Dedicated Read Nodes gave us the speed, consistency, and isolation we needed to run real-time recommendations at scale. Instead of managing infrastructure, we spend our time improving our recommendation model and the product itself. That has reduced the time our customers spend researching, filtering, and evaluating contacts—from hours to minutes—by giving them the right people to reach out to with a single click.” said Carlos Nunez, Vice President of Engineering and Applied AI at ZoomInfo. “Every company today has multiple AI applications, and each one has different performance and cost profiles. Pinecone is the only vector database that lets customers run all of these workloads in one place, with exceptional speed, accuracy, and cost efficiency at scale. Our serverless slab architecture and Dedicated Read Nodes deliver trusted knowledge to our customers and make it possible to deploy production-grade RAG, search, recommendation systems, and agents without compromise.” said Ash Ashutosh, CEO of Pinecone.