Faster search. Fewer instances.
Overview

Introduction

Hyperspace is a cloud based hybrid search engine that allows fast and scalable queries for advanced applications at billion scale. Hyperspace uses cloud FPGA hardware to provide unprecedented latency, typically measuring x10-x100 faster than industry benchmarks, and at reduced costs. The Hyper Space engine query syntax is native Python with supported functionality for candidate generation and scoring for similarity and vector searches. 

Hybrid Search

Vector Search
Vector databases are optimized for storing and retrieving high-dimensional data such as vector embeddings. Vector databases allow efficient store of complex data, such as time series and embedded information. They have many common usages, from NLP and word embedding to information retrieval and recommendation systems.  
Document and Keyword Search
Classic keyword search databases are designed to handle large volumes of unstructured and semi-structured data, such as documents, key-value pairs, graphs, and time-series data. These databases schema-less by nature, allowing for greater flexibility in accommodating evolving data and to better handle dynamic and rapidly changing data requirements.
Hyperspace Hybrid Search
Hyper space offers a low latency  hybrid (keyword and vector search) that combines the best of both worlds. The hyperspace index formulates data as key-value pairs, where dense and sparse vectors are stored under a designated key.
Hyperspace Queries are Lightning Fast at Billion Scale
Hyperspace engine utilizes cloud FPGA hardware, allowing unprecedented latency without performance reduction in high throughput and at billion scale.
  • Hybrid Search: Hyperspace allows both vector and metadata searches, providing the best of both worlds. 
  • Billion Scale: Hyperspace adopts native Python syntax, facilitating a seamless transition and natural migration of existing codebases.
  • 10x performance: Hyperspace offers x100-x10 lower latency when compared to industry benchmarks, allowing more complex logic in lower latency frames.
  • Cost Efficiency: Hyperspace can significantly reduce machine time requirements and associated costs.
This is some text inside of a div block.

Hyperspace Index Combines Vector data and Metadata

Hyperspace documents are formulated as key-value pairs, where vectors are stored under designated keys. Each record can contain vectors, key-value pairs, or any combination of the two.
‍

Manage Data Using Native Python Syntax

Hyperspace engine allows CRUD operations in native python syntax

Easily Run Hybrid Search