Timescale Vector Launches to Enable Developers to Build Production AI Applications at Scale With PostgreSQL

Print Friendly, PDF & Email

Developers can now leverage the robustness and reliability of PostgreSQL for AI while minimizing operational complexity, increasing efficiency and achieving faster search speeds

Timescale, the cloud database company, announced the launch of Timescale Vector, enabling developers to build production AI applications at scale with PostgreSQL. With Timescale Vector, which sits atop Timescale’s production-grade cloud PostgreSQL platform, developers can now leverage a single platform for managing relational data, vector embeddings, time-series data, analytics and event data that powers their next-generation AI applications. Developers can now bring AI products to market faster, more reliably and efficiently than with traditional vector databases.

The news comes as the rise of large language models (LLMs) like GPT-4, Llama 2, and Claude 2 continues to drive explosive growth in AI applications. At the heart of this growth is vector data and specifically, vector embeddings. To power next-gen AI systems, developers need to efficiently store and query vectors. With a myriad of vector databases in the market, developers face a paradox of choice: adopt new, niche databases built specifically for vector data or use familiar, general-purpose databases, like PostgreSQL, with extensions for vector support. Niche databases for vector data have proven to be operationally complex, requiring developers to maintain a separate database, where teams are required to duplicate, synchronize and keep track of data across multiple systems. Engineering teams also face the steep learning curve of learning a new query language, system internals, APIs and optimization techniques. Additionally, most niche vector databases are unproven, nascent technology, with unproven long term stability and reliability. 

To combat these challenges, Timescale is further extending PostgreSQL with capabilities to store, query and manage vector data at scale. Timescale Vector benefits developers and their teams by:

  • Simplifying the AI application stack, giving developers a single place for the relational data, vector embeddings, and time-series, analytics, and event data that powers their next-generation AI applications. This removes the need for developers to manage another piece of infrastructure and minimizes the operational complexity of data duplication, synchronization, and keeping track of updates across multiple systems. Because Timescale Vector is still PostgreSQL, it inherits the 30+ years of battle testing, robustness, and reliability of PostgreSQL, giving developers more peace of mind about their database choice for data that’s critical to a great user experience.
  • Speeding up ANN search on millions of vectors, enhancing pgvector with a state-of-the-art Approximate Nearest Neighbor (ANN) index inspired by the DiskANN algorithm, in addition to offering pgvector’s HNSW and ivfflat indexing algorithms. Timescale Vector achieves 243% faster search speed at 99% recall than Weaviate, a specialized vector database, and between 39.39% and 363.48% faster search speed than previously best-in-class PostgreSQL search indexes (pgvector HNSW and pg_embedding respectively) on a dataset of one million OpenAI embeddings.
  • Optimizing time-based vector search, leveraging automatic time-based partitioning and indexing to efficiently find recent embeddings, constrain vector search by a time range or document age, and store and retrieve LLM response and chat history with ease.
  • Simplifying the handling of metadata and multi-attribute filtering, as developers can leverage all PostgreSQL data types to store and filter metadata, JOIN vector search results with relational data for more contextually relevant responses, and write full SQL relational queries incorporating vector embeddings. 

“We launched Timescale over six years ago with the idea that we’re more than just a PostgreSQL extension — we’re making PostgreSQL easier, faster, and more cost effective for developers building data-intensive applications,” said Ajay Kulkarni, CEO and Co-founder, Timescale. “The launch of Timescale Vector signifies our commitment to continuing to solve the biggest developer pain points so they can focus on building new AI applications more efficiently on a database foundation that’s fast, reliable and battle-tested.” 

Availability and Pricing

Timescale Vector is available today in early access on Timescale, the PostgreSQL cloud platform, for new and existing customers. During the Early Access period, Timescale Vector will be free to use for all Timescale new and existing customers. To learn more, visit https://www.timescale.com/ai.

Sign up for the free insideBIGDATA newsletter.

Join us on Twitter: https://twitter.com/InsideBigData1

Join us on LinkedIn: https://www.linkedin.com/company/insidebigdata/

Join us on Facebook: https://www.facebook.com/insideBIGDATANOW

Speak Your Mind