Designed for cost-effective, sub-second response times at all data volumes, the new offering can solve multi-dimensional, complex problems by combining structured and unstructured data
Teradata has unveiled the Teradata Enterprise Vector Store, an in-database solution that enhances vector data management—a critical component for Trusted AI. This offering is set to integrate NVIDIA NeMo Retriever microservices, part of the NVIDIA AI Enterprise software platform. It is capable of processing billions of vectors and seamlessly integrating them into existing enterprise systems, delivering response times as swift as tens of milliseconds. Designed for cost-effective scalability, the Enterprise Vector Store addresses complex, multifaceted business challenges by combining structured and unstructured data.
This solution establishes a unified, trusted repository for all data, building upon Teradata’s robust support for retrieval-augmented generation (RAG) and paving the way for dynamic agentic AI applications, such as augmented call centers.
Vector stores are foundational for organizations aiming to leverage agentic AI. However, many existing vector stores involve trade-offs that can make them challenging or costly to apply to the most demanding business problems. Some offer speed but are limited to small datasets, while others can handle large vector volumes but lack the speed required for agentic AI use cases. The true potential is realized when organizations can combine rapid processing with massive computational capabilities, especially when integrating unstructured datasets with mission-critical structured data.
Louis Landry, Teradata’s CTO, emphasized the importance of vector stores in data management practices, stating that their impact is limited when they are slow or siloed. He highlighted Teradata’s expertise in high concurrency and linear scaling, as well as its ability to harmonize data and support RAG, positioning the Enterprise Vector Store as a dynamic, trusted foundation for large organizations adopting agentic AI.
The Teradata Enterprise Vector Store is engineered to efficiently support use cases requiring vector capabilities and RAG applications. With built-in, cost-effective scaling and near-seamless integration, it enables enterprises to maximize value and insights from unstructured data while optimizing expenditures. Leveraging Teradata’s hybrid capabilities, the Enterprise Vector Store is an ideal choice for organizations seeking flexible scaling across cloud and on-premises environments, facilitating a transition towards an agentic AI future while optimizing existing infrastructure.
Also Read: Precisely Introduces AI Enhancements in Data Integrity Suite
By managing unstructured data in multi-modal formats—including text, video, images, and PDFs—the Enterprise Vector Store unifies structured and unstructured data for comprehensive analysis. It engages with the full lifecycle of vector data management, from embedding generation and indexing to metadata management and intelligent search. This processing occurs within the existing Teradata system, which excels in flexible deployment options, including cloud, on-premises, or hybrid environments. The solution supports industry-leading frameworks like LangChain and RAG, alongside comprehensive data management and governance practices essential for Trusted AI. Additionally, planned temporal vector embedding capabilities are designed to enhance trust and explainability by tracking data changes over time, thereby improving accuracy and decision-making.
The Enterprise Vector Store is expected to integrate NVIDIA NeMo Retriever to provide a leading information retrieval solution with high accuracy and data privacy, enabling enterprises to generate business insights in real-time. Developers can fine-tune NeMo Retriever microservices in combination with community or custom models to build scalable document ingestion and RAG applications connected to proprietary data across various locations. NVIDIA NeMo Retriever extraction is designed to enable customers to utilize information and insights from unstructured data sources such as PDFs, facilitating the development of RAG-based applications that leverage real-time knowledge appended with information from across the corporate IT estate.
Pat Lee, Vice President of Strategic Enterprise Partnerships at NVIDIA, highlighted the significance of data in accurate AI inference, noting that the integration of Teradata Enterprise Vector Store with NVIDIA AI Enterprise and NVIDIA NeMo Retriever can unlock institutional knowledge stored in unstructured documents to power intelligent AI agents.
A practical application of this technology is demonstrated in the augmented call center use case. Here, the Teradata Enterprise Vector Store utilizes agentic AI and RAG to transform customer service, making it faster, more efficient, and tailored to individual customer needs. AI agents also enable upsell and cross-sell opportunities during customer interactions.
For instance, an insurance company stores contracts for its millions of customers in PDF format within an object store. It also employs a hybrid data platform for mission-critical customer 360 data. When a customer inquires about coverage, a multi-agent system leverages rapid access (as low as tens of milliseconds) to harmonized data to provide precise, context-aware responses.
The Teradata Enterprise Vector Store is currently available in private preview, with general availability anticipated in July.
