Find the right words
Lucene-powered full-text search with comparable scoring across segments.
Explore full-text searchIn active development
One lake.
One lake.
Every way
to search.
Full-text, vector, graph, and structured search.
One consistent dataset. Built on Apache Paimon.
◈One source of truth. Multiple index projections.
Many ways in. One consistent view.
Built on a proven foundation
◭ Apache Paimon✳ Apache Lucene♧ PostgreSQL▱ Object storage
A multimodal lakehouse database
Let your data do more.
Keep your architecture simple.
Search by meaning, match the exact words, follow a relationship, or apply a filter. Bring the results together on the same data foundation.
Search beyond keywords
Retrieve by vector similarity and combine semantic signals with exact matches.
Explore vector searchConnect the context
Traverse relationships and bring connected knowledge into your retrieval flow.
Explore graph traversalMake every result relevant
Use structured filters with a shared RowID execution layer and consistent visibility.
Explore hybrid queriesBuilt for fresh data
A durable object-store WAL and explicit eventual / wait_for consistency for the NRT path.
Keep a perspective on time
Snapshot, tag, and branch primitives keep data versions part of the architecture.
One query, multiple perspectives
From scattered signals
to a clearer answer.
Combine retrieval methods through a JSON DSL. ElasticLake brings the RowIDs together before fetching the rows.
✓
Explore the documentationBuilt around consistency
One source of truth.
A stronger foundation.
Paimon holds the data. Immutable indexes provide different ways to find it. A shared execution layer keeps retrieval tied to the same snapshot.
Unified query executionJSON DSL
Full-textVectorGraphFilter
Snapshot-aligned index projections↕
Apache PaimonSingle source of truth
Snapshots▱Object storage
One dataset, multiple projections
Indexes are derived from Paimon snapshots and designed to be rebuilt from the source.
A shared language for results
Every index returns RowIDs. One execution layer combines them and applies row visibility.
Time is part of the model
Snapshot-aligned indexes and versioning primitives make consistency explicit.
Designed for connected data
More context. More possibilities.
A foundation for applications that need more than one way to search.
AI & retrieval
Bring lexical, semantic, and relationship context together for RAG applications.
Hybrid retrievalSearch & discovery
Combine descriptive text, embedding similarity, and structured attributes.
Multimodal searchKnowledge & relationships
Explore connected entities and use graph expansion to enrich retrieval.
Graph explorationIn active developmentDeployment resources.
Deployment & docsDeployment resources.
One place to begin.
DockerKubernetes
Our GitHub repository is the home for Docker and Kubernetes deployment resources and product documentation. ElasticLake is under active development; full deployment and performance acceptance remain in progress.
A new perspective on your data.
Explore the architecture. Find deployment resources and documentation.