New 5.5
Vector Search
Perform Similarity Searches on Text, Images and other Unstructured Data
High-performance querying for finding related information to uncover insights
Vector search is now part of the Hazelcast unified platform, enabling new use cases like semantic search, fraud detection and RAG. In addition, support for ingesting unstructured data has been added to the Pipeline API, making it simple to construct scalable, robust embedding pipelines and run them on the same runtime.
- Use Pipeline API to create job
- Collection is searchable while additional embeddings are being processed
- Update to data gets processed immediately
- VectorCollection structure = efficient updates
- Use same key to replace VectorDocument
- New vectors = index update
Vector Search with Hazelcast
Learn how to use Hazelcast to build and search a vector data collection
Features
Industry Leading Vector Search Engine Performance
Hazelcast Platform harnesses the pioneering JVector 2.0, one of the most advanced vector search engines.
High Performance, Low Latency Embedding and Ingest
Pipelines receive immediate updates from any source and take advantage of cluster compute to generate embeddings and have a continuously up-to-date vector index for similarity search.
Java and Python Clients
Search the collection using Java or Python clients. Scan index for Approximate Nearest Neighbors (ANN) and collect matching keys for data retrieval.
Scalable Architecture
Scale vector embedding throughput by simply adding more nodes.
Take the next step
See how strong consistency, performance, resilience, and scale drive an AI-centric future.