LangChain4j Embedding Store

Since Camel 4.14

Only producer is supported

The LangChain4j Embedding Store component provides integration with LangChain4j embedding stores for vector database operations. This component enables storing, retrieving, and searching embeddings across multiple vector database implementations through LangChain4j’s unified interface.

Features

The LangChain4j Embedding Store component offers the following key features:

  • Vector Operations: Add, remove, and search embeddings in vector databases

  • Similarity Search: Perform semantic search with configurable scoring thresholds

  • Metadata Filtering: Search with metadata-based constraints using filters

  • Multi-Database Support: Support for various vector databases including Qdrant, Milvus, Weaviate, Neo4j, and others

  • Flexible Configuration: Configure embedding stores via direct instances or factory patterns

Supported Operations

The component supports three main operations controlled by the CamelLangchain4jEmbeddingStoreAction header:

  • ADD - Store embeddings with optional text segments and metadata. Supports single and batch operations, as well as caller-supplied IDs.

  • REMOVE - Delete embeddings by single ID, collection of IDs, or metadata filter.

  • SEARCH - Perform similarity search with configurable filters and scoring

URI format

langchain4j-embeddingstore:embeddingStoreId[?options]

Where embeddingStoreId is a unique identifier for the embedding store instance.

Configuring Options

Camel components are configured on two separate levels:

  • component level

  • endpoint level

Configuring Component Options

At the component level, you set general and shared configurations that are, then, inherited by the endpoints. It is the highest configuration level.

For example, a component may have security settings, credentials for authentication, urls for network connection and so forth.

Some components only have a few options, and others may have many. Because components typically have pre-configured defaults that are commonly used, then you may often only need to configure a few options on a component; or none at all.

You can configure components using:

  • the Component DSL.

  • in a configuration file (application.properties, *.yaml files, etc).

  • directly in the Java code.

Configuring Endpoint Options

You usually spend more time setting up endpoints because they have many options. These options help you customize what you want the endpoint to do. The options are also categorized into whether the endpoint is used as a consumer (from), as a producer (to), or both.

Configuring endpoints is most often done directly in the endpoint URI as path and query parameters. You can also use the Endpoint DSL and DataFormat DSL as a type safe way of configuring endpoints and data formats in Java.

A good practice when configuring options is to use Property Placeholders.

Property placeholders provide a few benefits:

  • They help prevent using hardcoded urls, port numbers, sensitive information, and other settings.

  • They allow externalizing the configuration from the code.

  • They help the code to become more flexible and reusable.

The following two sections list all the options, firstly for the component followed by the endpoint.

Component Options

The LangChain4j Embedding Store component supports the following options which are listed below.

Name Description Default Type

action (producer)

The operation to perform: ADD, REMOVE, or SEARCH.

Enum values:

  • ADD

  • REMOVE

  • SEARCH

LangChain4jEmbeddingStoreAction

configuration (producer)

The configuration;.

LangChain4jEmbeddingStoreConfiguration

embeddingModel (producer)

Autowired Embedding model for auto-computing embeddings from message body text. When set, ADD and SEARCH operations can accept plain text body instead of requiring a pre-computed embedding in the CamelLangChain4jEmbeddingsEmbedding header. The header always takes precedence when present.

EmbeddingModel

embeddingStore (producer)

Autowired Direct embedding store instance for vector operations.

EmbeddingStore

embeddingStoreFactory (producer)

Autowired The embedding store factory to use for creating embedding stores if no embeddingstore is provided.

EmbeddingStoreFactory

lazyStartProducer (producer)

Whether the producer should be started lazy (on the first message). By starting lazy you can use this to allow CamelContext and routes to startup in situations where a producer may otherwise fail during starting and cause the route to fail being started. By deferring this startup to be lazy then the startup failure can be handled during routing messages via Camel’s routing error handlers. Beware that when the first message is processed then creating and starting the producer may take a little time and prolong the total processing time of the processing.

false

boolean

maxResults (producer)

Maximum number of results to return for SEARCH operation.

5

Integer

minScore (producer)

Minimum similarity score threshold for SEARCH operation (0.0 to 1.0).

Double

returnTextContent (producer)

When true, SEARCH returns List with text content instead of List.

false

boolean

autowiredEnabled (advanced)

Whether autowiring is enabled. This is used for automatic autowiring options (the option must be marked as autowired) by looking up in the registry to find if there is a single instance of matching type, which then gets configured on the component. This can be used for automatic configuring JDBC data sources, JMS connection factories, AWS Clients, etc.

true

boolean

Endpoint Options

The LangChain4j Embedding Store endpoint is configured using URI syntax:

langchain4j-embeddingstore:embeddingStoreId

With the following path and query parameters:

Path Parameters

Name Description Default Type

embeddingStoreId (producer)

Required The id of the embedding store.

String

Query Parameters

Name Description Default Type

action (producer)

The operation to perform: ADD, REMOVE, or SEARCH.

Enum values:

  • ADD

  • REMOVE

  • SEARCH

LangChain4jEmbeddingStoreAction

embeddingModel (producer)

Autowired Embedding model for auto-computing embeddings from message body text. When set, ADD and SEARCH operations can accept plain text body instead of requiring a pre-computed embedding in the CamelLangChain4jEmbeddingsEmbedding header. The header always takes precedence when present.

EmbeddingModel

embeddingStore (producer)

Autowired Direct embedding store instance for vector operations.

EmbeddingStore

embeddingStoreFactory (producer)

Autowired The embedding store factory to use for creating embedding stores if no embeddingstore is provided.

EmbeddingStoreFactory

maxResults (producer)

Maximum number of results to return for SEARCH operation.

5

Integer

minScore (producer)

Minimum similarity score threshold for SEARCH operation (0.0 to 1.0).

Double

returnTextContent (producer)

When true, SEARCH returns List with text content instead of List.

false

boolean

lazyStartProducer (producer (advanced))

Whether the producer should be started lazy (on the first message). By starting lazy you can use this to allow CamelContext and routes to startup in situations where a producer may otherwise fail during starting and cause the route to fail being started. By deferring this startup to be lazy then the startup failure can be handled during routing messages via Camel’s routing error handlers. Beware that when the first message is processed then creating and starting the producer may take a little time and prolong the total processing time of the processing.

false

boolean

Message Headers

The LangChain4j Embedding Store component supports the following message header(s), which is/are listed below:

Name Description Default Type

CamelLangchain4jEmbeddingStoreAction (producer)

Constant: ACTION

The action to be performed.

Enum values:

  • ADD

  • REMOVE

  • SEARCH

String

CamelLangchain4jEmbeddingStoreMaxResults (producer)

Constant: MAX_RESULTS

Maximum number of search results to return.

5

Integer

CamelLangchain4jEmbeddingStoreMinScore (producer)

Constant: MIN_SCORE

Minimum similarity score for search results.

Double

CamelLangchain4jEmbeddingStoreFilter (producer)

Constant: FILTER

Filter for metadata-based constraints (used in SEARCH and REMOVE operations).

Filter

CamelLangchain4jEmbeddingStoreEmbeddingId (producer)

Constant: EMBEDDING_ID

Caller-supplied embedding ID for single ADD operations.

String

CamelLangchain4jEmbeddingStoreEmbeddingIds (producer)

Constant: EMBEDDING_IDS

Caller-supplied embedding IDs for batch ADD operations.

List

Usage

Configuring an Embedding Store

The component requires an EmbeddingStore instance. Register it in the Camel registry:

Java-only: programmatic EmbeddingStore configuration and registry binding
EmbeddingStore<TextSegment> embeddingStore = PgVectorEmbeddingStore.builder()
    .host("localhost")
    .port(5432)
    .database("vectordb")
    .user("postgres")
    .password("postgres")
    .table("embeddings")
    .dimension(384)
    .build();

context.getRegistry().bind("myEmbeddingStore", embeddingStore);

Auto-Embedding with embeddingModel

When an EmbeddingModel is configured (either explicitly or via autowiring from the registry), the ADD and SEARCH operations can accept plain text in the message body instead of requiring a pre-computed embedding in the CamelLangChain4jEmbeddingsEmbedding header. The component will automatically compute the embedding from the body text.

If the CamelLangChain4jEmbeddingsEmbedding header is present, it always takes precedence over auto-embedding.

  • Java

  • YAML

from("direct:store")
    .to("langchain4j-embeddingstore:myStore?action=ADD");

from("direct:search")
    .to("langchain4j-embeddingstore:myStore?action=SEARCH&returnTextContent=true");
- route:
    from:
      uri: direct:store
    steps:
      - to:
          uri: langchain4j-embeddingstore:myStore
          parameters:
            action: ADD
- route:
    from:
      uri: direct:search
    steps:
      - to:
          uri: langchain4j-embeddingstore:myStore
          parameters:
            action: SEARCH
            returnTextContent: true

This eliminates the need for a separate langchain4j-embeddings:embed step in the route.

Storing Embeddings (ADD Operation)

Store embeddings with optional text segments. Without embeddingModel, a pre-computed embedding must be provided via the langchain4j-embeddings component:

  • Java

  • YAML

from("direct:store")
    .to("langchain4j-embeddings:embed")
    .to("langchain4j-embeddingstore:myStore?action=ADD");
- route:
    from:
      uri: direct:store
    steps:
      - to:
          uri: langchain4j-embeddings:embed
      - to:
          uri: langchain4j-embeddingstore:myStore
          parameters:
            action: ADD

The response body contains the generated embedding ID.

ADD with Caller-Supplied ID

You can provide your own embedding ID using the CamelLangchain4jEmbeddingStoreEmbeddingId header:

from("direct:store-with-id")
    .to("langchain4j-embeddings:embed")
    .setHeader("CamelLangchain4jEmbeddingStoreEmbeddingId", constant("my-custom-id"))
    .to("langchain4j-embeddingstore:myStore?action=ADD");

When both a caller-supplied ID and a text segment are present, the component preserves both using addAll with singleton lists.

Batch ADD (addAll)

For RAG ingestion pipelines, batch adding is significantly more efficient. Send a List<String> or List<TextSegment> as the body to the langchain4j-embeddings component to produce batch embeddings:

from("direct:batch-store")
    // Body is a List<String> or List<TextSegment>
    .to("langchain4j-embeddings:embed")
    // After batch embed, the CamelLangChain4jEmbeddingsEmbeddings header contains List<Embedding>
    // and the CamelLangChain4jEmbeddingsTextSegments header preserves the original text segments
    .to("langchain4j-embeddingstore:myStore?action=ADD");
// Body: List<String> of generated IDs

You can also supply caller-supplied IDs for the batch via the CamelLangchain4jEmbeddingStoreEmbeddingIds header.

The sizes of EMBEDDING_IDS, EMBEDDINGS, and text segments must all match. A size mismatch throws IllegalArgumentException.

Searching Embeddings (SEARCH Operation)

Perform similarity search to find relevant content. Without embeddingModel, a pre-computed query embedding must be provided:

  • Java

  • YAML

from("direct:search")
    .to("langchain4j-embeddings:embed")
    .to("langchain4j-embeddingstore:myStore?action=SEARCH&maxResults=5&minScore=0.7");
- route:
    from:
      uri: direct:search
    steps:
      - to:
          uri: langchain4j-embeddings:embed
      - to:
          uri: langchain4j-embeddingstore:myStore
          parameters:
            action: SEARCH
            maxResults: 5
            minScore: 0.7

The response contains a list of EmbeddingMatch objects with the matching text segments and scores.

Returning Text Content Directly

Use the returnTextContent option to get a list of strings instead of EmbeddingMatch objects:

  • Java

  • YAML

from("direct:search")
    .to("langchain4j-embeddings:embed")
    .to("langchain4j-embeddingstore:myStore?action=SEARCH&maxResults=5&returnTextContent=true")
    .log("Found texts: ${body}");
- route:
    from:
      uri: direct:search
    steps:
      - to:
          uri: langchain4j-embeddings:embed
      - to:
          uri: langchain4j-embeddingstore:myStore
          parameters:
            action: SEARCH
            maxResults: 5
            returnTextContent: true
      - log:
          message: "Found texts: ${body}"

Removing Embeddings (REMOVE Operation)

The REMOVE operation supports multiple strategies:

Remove by Single ID

Delete a single embedding by its ID (body is a String):

  • Java

  • YAML

from("direct:remove")
    .setBody(simple("${header.embeddingId}"))
    .to("langchain4j-embeddingstore:myStore?action=REMOVE");
- route:
    from:
      uri: direct:remove
    steps:
      - setBody:
          expression:
            simple:
              expression: "${header.embeddingId}"
      - to:
          uri: langchain4j-embeddingstore:myStore
          parameters:
            action: REMOVE

Remove by Collection of IDs

Delete multiple embeddings by passing a Collection<String> as the body:

from("direct:remove-batch")
    .setBody(constant(List.of("id-1", "id-2", "id-3")))
    .to("langchain4j-embeddingstore:myStore?action=REMOVE");

Remove by Metadata Filter

Delete all embeddings matching a metadata filter using the CamelLangchain4jEmbeddingStoreFilter header:

from("direct:remove-by-filter")
    .setHeader("CamelLangchain4jEmbeddingStoreFilter", constant(metadataFilter))
    .to("langchain4j-embeddingstore:myStore?action=REMOVE");
A REMOVE with no body and no filter throws IllegalArgumentException. This is intentional — destructive "clear all" operations require explicit intent.

Complete RAG Pipeline Example

A complete example showing document ingestion and retrieval:

Java-only: multi-route RAG pipeline with split and tokenize
// Ingestion route: chunk, embed, and store documents
from("file:documents?include=.*\\.txt")
    .split().tokenize("\n\n") // Split by paragraphs
    .to("langchain4j-embeddings:embed")
    .to("langchain4j-embeddingstore:ragStore?action=ADD")
    .log("Stored embedding with ID: ${body}");

// Query route: embed query, search, and return text results
from("direct:query")
    .to("langchain4j-embeddings:embed")
    .to("langchain4j-embeddingstore:ragStore?action=SEARCH&maxResults=3&returnTextContent=true");

Supported Vector Databases

The component supports any vector database that LangChain4j provides an EmbeddingStore implementation for:

  • PGVector - PostgreSQL with pgvector extension

  • Qdrant - High-performance vector database

  • Milvus - Cloud-native vector database

  • Weaviate - Vector search engine

  • Chroma - Open-source embedding database

  • Pinecone - Managed vector database

  • Neo4j - Graph database with vector capabilities

  • InMemoryEmbeddingStore - For testing purposes

Refer to the LangChain4j Embedding Stores documentation for the complete list and configuration options.

Back to top