mirror of
https://github.com/lancedb/lancedb.git
synced 2026-11-16 07:36:50 +00:00
Shortens the docs landing page and organizes its overview around
database features, with the multimodal data model, ML workflow, SDKs,
and PyTorch loader explained in prose. Keeps versioning visible,
branching in draft, and the existing `{#architecture}` anchor.
Validation: added documentation links and anchors verified; `git diff
--check` passes. `mint validate` reports the same four pre-existing
missing snippet-export warnings in the reranking and vector-search
pages, also reproduced on the unchanged revision.
Closes
https://linear.app/lancedb/issue/DOC-80/strengthen-what-is-lancedb
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brendan Clement <brendan.clement@lancedb.com>
65 lines
3.9 KiB
Plaintext
65 lines
3.9 KiB
Plaintext
---
|
|
title: LanceDB
|
|
sidebarTitle: "LanceDB"
|
|
description: "Multimodal lakehouse for AI."
|
|
icon: "/static/assets/logo/lancedb-icon-gray.svg"
|
|
keywords: ["multimodal lakehouse", "training", "feature engineering", "search", "open source", "oss"]
|
|
---
|
|
|
|
LanceDB Enterprise is a distributed multimodal lakehouse for AI. It gives you one data layer for search and
|
|
retrieval, curation, feature engineering, and model training. It is built on [Lance](/basics/lance), an open-source
|
|
columnar format for multimodal AI data. For the open-source retrieval library that runs inside your application,
|
|
see [LanceDB Embedded](/basics/lancedb-embedded).
|
|
|
|
LanceDB Enterprise depends only on object storage for its primary state, and NVMe/memory for caching in the serving
|
|
layer, giving it the latency profile of an in-memory database with the petabyte scalability and cost profile of S3.
|
|
It brings these workloads into one system, giving your team one system to become expert in and removing the need
|
|
for multiple database systems and synchronization pipelines between them.
|
|
|
|
Core features include:
|
|
|
|
- [**SQL**](/basics/sql) powered by Apache DataFusion
|
|
- [**Secondary indexes**](/indexing) for vector search, full-text search, and scalar filters
|
|
- [**Computed columns**](/feature-engineering/add-computed-column) with on-demand or scheduled refresh
|
|
- [**Materialized views**](/feature-engineering/materialized-views) with on-demand or scheduled refresh
|
|
- [**Time travel and versioning**](/management/table#versioning) to pin and revisit datasets
|
|
- **Search**: [vector](/search/vector-search), [full-text](/search/full-text-search), and
|
|
[hybrid](/search/hybrid-search), with [filters](/search/filtering) and [reranking](/reranking)
|
|
- [**Functions**](/curation/create-function) for custom transformations and model inference (early access)
|
|
- [**Row-level random access**](/basics/lance#capabilities) for retrieval and training
|
|
- [**Column additions**](/management/table#schema-add-new-columns) without rewriting existing columns
|
|
|
|
{/*
|
|
Draft until branching reaches early access. Add this item to the list above, after Time travel and versioning:
|
|
- [**Copy-on-write branching**](/tables/branching) that gives each experiment or agent a writable fork of a table to review before it reaches `main`
|
|
*/}
|
|
|
|
LanceDB Enterprise stores images, video, audio, text, embeddings, and metadata in
|
|
[schema-enforced tables](/management/table#multimodal). These features let you build
|
|
declarative ML data pipelines: ingest raw data and metadata, enrich it incrementally with computed columns and
|
|
Functions, and organize it into [curated datasets](/curation) with materialized views. Add features without
|
|
rewriting existing columns, then train from those datasets using row-level random access and sampling.
|
|
|
|
Connect through [SDKs](/basics/sdk) for Python, TypeScript, and Rust, and use the
|
|
[PyTorch data loader](/training/data-loading) for reproducible shuffling and streaming batches.
|
|
|
|
## Architecture {#architecture}
|
|
|
|
Queries and feature engineering run across a cluster, with background indexing and maintenance. Deploy
|
|
LanceDB Enterprise in your cloud or use a LanceDB-managed deployment.
|
|
|
|
Your application connects to the service through an SDK or [SQL client](/basics/sql). The cluster executes queries,
|
|
refreshes derived data, and maintains tables and indexes outside your application.
|
|
|
|
## Projects using LanceDB {#projects-using-lancedb}
|
|
|
|
Teams use LanceDB for world-model and robotics training, as in
|
|
[stable-worldmodel](https://github.com/galilai-group/stable-worldmodel) and
|
|
[lerobot-lancedb](https://github.com/lancedb/lerobot-lancedb), and for agent memory, as in
|
|
[CrewAI](https://www.lancedb.com/blog/crewai-rebuilt-agent-memory-on-lancedb),
|
|
[Cognee](https://github.com/topoteretes/cognee), and
|
|
[OpenClaw](https://docs.openclaw.ai/plugins/memory-lancedb).
|
|
|
|
Try out LanceDB using our [quickstart](/quickstart). For help planning an Enterprise deployment
|
|
that meets your scale and performance needs, [contact the LanceDB team](https://www.lancedb.com/contact).
|