Firebase Vector Store

Firestore supports native vector similarity search, making it a fully managed vector store for RAG applications. The Firebase plugin automatically handles embedding generation, document indexing, and vector queries.

Installation

<dependency>
    <groupId>com.google.genkit</groupId>
    <artifactId>genkit-plugin-firebase</artifactId>
    <version>1.0.0-SNAPSHOT</version>
</dependency>

Setting up a retriever

import com.google.genkit.plugins.firebase.FirebasePlugin;
import com.google.genkit.plugins.firebase.retriever.FirestoreRetrieverConfig;

Genkit genkit = Genkit.builder()
    .plugin(GoogleGenAIPlugin.create(apiKey))
    .plugin(FirebasePlugin.builder()
        .projectId("my-project")
        .addRetriever(FirestoreRetrieverConfig.builder()
            .name("my-docs")
            .collection("documents")
            .embedderName("googleai/gemini-embedding-001")
            .vectorField("embedding")
            .contentField("content")
            .distanceMeasure(FirestoreRetrieverConfig.DistanceMeasure.COSINE)
            .defaultLimit(5)
            .build())
        .build())
    .build();

Indexing documents

Documents are embedded automatically when indexed. The plugin generates embeddings using the configured embedder and stores them alongside the content in Firestore.

List<Document> docs = List.of(
    Document.fromText("Genkit is a framework for building AI apps"),
    Document.fromText("Firebase provides cloud services for apps")
);
genkit.index("firebase/my-docs", docs);

Retrieving with vector similarity

List<Document> results = genkit.retrieve("firebase/my-docs", "What is Genkit?");

Using results in RAG

genkit.defineFlow("ragQuery", String.class, String.class, (ctx, question) -> {
    List<Document> context = genkit.retrieve("firebase/my-docs", question);
    return genkit.generate(GenerateOptions.builder()
        .model("googleai/gemini-2.5-flash")
        .prompt(question)
        .docs(context)
        .build()).getText();
});

Auto-creating database and vector index

The plugin can automatically create the Firestore database and the required composite vector index:

FirestoreRetrieverConfig.builder()
    .name("my-docs")
    .collection("documents")
    .embedderName("googleai/gemini-embedding-001")
    .vectorField("embedding")
    .contentField("content")
    .createDatabaseIfNotExists(true)
    .createVectorIndexIfNotExists(true)
    .build()

You can also create the vector index manually via the Firebase CLI:

gcloud firestore indexes composite create \
  --collection-group=documents \
  --field-config=field-path=embedding,vector-config='{"dimension":"768","flat":{}}' \
  --database="(default)"

Configuration options

Option	Default	Description
name	(required)	Retriever name — used as firebase/{name}
collection	(required)	Firestore collection name
embedderName	(required)	Embedder to use (e.g., googleai/gemini-embedding-001)
vectorField	"embedding"	Firestore field storing the vector
contentField	"content"	Firestore field storing the text content
distanceMeasure	COSINE	COSINE, EUCLIDEAN, or DOT_PRODUCT
defaultLimit	10	Number of results to return
distanceThreshold	none	Maximum distance threshold
metadataFields	none	Additional fields to include in document metadata
embedderDimension	768	Embedding vector dimension
databaseId	"(default)"	Firestore database ID
createDatabaseIfNotExists	false	Auto-create Firestore database
createVectorIndexIfNotExists	false	Auto-create composite vector index

Requirements

• Firebase project on the Blaze (pay-as-you-go) plan
• Application Default Credentials or a service account JSON
• GCLOUD_PROJECT or GOOGLE_CLOUD_PROJECT environment variable (auto-detected)

Sample

See the firebase sample for a complete RAG pipeline with Firestore vector search.