RAG (Retrieval Augmented Generation)

RAG enhances AI model output by injecting relevant information from your data into prompts before sending them to the model.

Defining a retriever

Retriever myRetriever = genkit.defineRetriever("myStore/docs", (ctx, request) -> {
    List<Document> docs = findSimilarDocs(request.getQuery());
    return new RetrieverResponse(docs);
});

Defining an indexer

Indexer myIndexer = genkit.defineIndexer("myStore/docs", (ctx, request) -> {
    indexDocuments(request.getDocuments());
    return new IndexerResponse();
});

Indexing documents

List<Document> docs = List.of(
    Document.fromText("Paris is the capital of France."),
    Document.fromText("Berlin is the capital of Germany.")
);
genkit.index("myStore/docs", docs);

Retrieve and generate

// Retrieve relevant documents
List<Document> relevantDocs = genkit.retrieve(
    "myStore/docs", "What is the capital of France?"
);

// Generate with context
ModelResponse response = genkit.generate(
    GenerateOptions.builder()
        .model("openai/gpt-4o-mini")
        .prompt("Answer based on context: What is the capital of France?")
        .docs(relevantDocs)
        .build());

Embeddings

Generate vector embeddings for semantic search:

List<Document> documents = List.of(
    Document.fromText("Hello world"),
    Document.fromText("Goodbye world")
);
EmbedResponse response = genkit.embed("openai/text-embedding-3-small", documents);

Vector store plugins

Genkit Java supports multiple vector databases out of the box:

Local Vector Store File-based store for development

Weaviate Open-source vector database

PostgreSQL (pgvector) Vector search in PostgreSQL

Pinecone Managed vector database

Firebase Firestore Firestore vector search