次の方法で共有


エージェントを非構造化データに接続する

AI エージェントは、多くの場合、ドキュメント コレクション、ナレッジ ベース、テキスト コーパスなどの非構造化データに対してクエリを実行して質問に回答し、コンテキストに対応した応答を提供する必要があります。

Databricks には、Vector Search インデックスと外部ベクター ストア内の非構造化データにエージェントを接続するための複数のアプローチが用意されています。 事前構成済みの MCP サーバーを使用して、Databricks Vector Search インデックスにすぐにアクセスしたり、AI Bridge パッケージを使用してレトリバー ツールをローカルで開発したり、特殊なワークフロー用のカスタム取得機能を構築したりできます。

このページでは、次の方法を示します。

Databricks Vector Search インデックスのクエリを実行する

エージェントで Databricks Vector Search インデックスのクエリを実行する必要がある場合は、Databricks で管理されている MCP サーバーを使用します。

  1. Databricks マネージド埋め込みを使用してベクター検索インデックスを作成します。 ベクター検索エンドポイントとインデックスの作成を参照してください。
  2. エージェントを作成し、Vector Search Index: https://<workspace-hostname>/api/2.0/mcp/vector-search/{catalog}/{schema}/{index_name} の事前構成済みのマネージド MCP URL に接続します。

管理対象 MCP サーバーに接続するエージェントを作成する方法については、「 Databricks マネージド MCP サーバーの使用」を参照してください。

AI Bridge を使用してベクター検索取得ツールをローカルで開発する

Databricks Vector Search 取得ツールをローカルで構築するには、databricks-langchainなどの databricks-openaiを使用します。 これらのパッケージには、既存の Databricks リソースからレトリバーを作成するための from_vector_searchfrom_uc_function などのヘルパー関数が含まれています。

LangChain/LangGraph

Databricks AI Bridge を含む最新バージョンの databricks-langchain をインストールします。

%pip install --upgrade databricks-langchain

次のコードは、仮想ベクトル検索インデックスに対してクエリを実行し、それを LLM にローカルでバインドするレトリバー ツールをプロトタイプ化して、ツール呼び出し動作をテストできるようにします。

エージェントがツールを理解し、ツールを呼び出すタイミングを判断するのに役立つわかりやすい tool_description を提供します。

from databricks_langchain import VectorSearchRetrieverTool, ChatDatabricks

# Initialize the retriever tool.
vs_tool = VectorSearchRetrieverTool(
  index_name="catalog.schema.my_databricks_docs_index",
  tool_name="databricks_docs_retriever",
  tool_description="Retrieves information about Databricks products from official Databricks documentation."
)

# Run a query against the vector search index locally for testing
vs_tool.invoke("Databricks Agent Framework?")

# Bind the retriever tool to your Langchain LLM of choice
llm = ChatDatabricks(endpoint="databricks-claude-sonnet-4-5")
llm_with_tools = llm.bind_tools([vs_tool])

# Chat with your LLM to test the tool calling functionality
llm_with_tools.invoke("Based on the Databricks documentation, what is Databricks Agent Framework?")

セルフマネージド埋め込みを使用して直接アクセス インデックスまたは差分同期インデックスを使用するシナリオでは、 VectorSearchRetrieverTool を構成し、カスタム埋め込みモデルとテキスト列を指定する必要があります。 埋め込みを提供するためのオプションを参照してください。

次の例では、VectorSearchRetrieverToolキーとcolumns キーを使用してembeddingを構成する方法を示します。

from databricks_langchain import VectorSearchRetrieverTool
from databricks_langchain import DatabricksEmbeddings

embedding_model = DatabricksEmbeddings(
    endpoint="databricks-bge-large-en",
)

vs_tool = VectorSearchRetrieverTool(
  index_name="catalog.schema.index_name", # Index name in the format 'catalog.schema.index'
  num_results=5, # Max number of documents to return
  columns=["primary_key", "text_column"], # List of columns to include in the search
  filters={"text_column LIKE": "Databricks"}, # Filters to apply to the query
  query_type="ANN", # Query type ("ANN" or "HYBRID").
  tool_name="name of the tool", # Used by the LLM to understand the purpose of the tool
  tool_description="Purpose of the tool", # Used by the LLM to understand the purpose of the tool
  text_column="text_column", # Specify text column for embeddings. Required for direct-access index or delta-sync index with self-managed embeddings.
  embedding=embedding_model # The embedding model. Required for direct-access index or delta-sync index with self-managed embeddings.
)

詳細については、VectorSearchRetrieverToolを参照してください。

OpenAI

Databricks AI Bridge を含む最新バージョンの databricks-openai をインストールします。

%pip install --upgrade databricks-openai

次のコードは、架空のベクター検索インデックスに対してクエリを実行し、OpenAI の GPT モデルと統合するレトリバーをプロトタイプ化します。

エージェントがツールを理解し、ツールを呼び出すタイミングを判断するのに役立つわかりやすい tool_description を提供します。

ツールに関する OpenAI の推奨事項の詳細については、 OpenAI 関数呼び出しに関するドキュメントを参照してください

from databricks_openai import VectorSearchRetrieverTool
from openai import OpenAI
import json

# Initialize OpenAI client
client = OpenAI(api_key=<your_API_key>)

# Initialize the retriever tool
dbvs_tool = VectorSearchRetrieverTool(
  index_name="catalog.schema.my_databricks_docs_index",
  tool_name="databricks_docs_retriever",
  tool_description="Retrieves information about Databricks products from official Databricks documentation"
)

messages = [
  {"role": "system", "content": "You are a helpful assistant."},
  {
    "role": "user",
    "content": "Using the Databricks documentation, answer what is Spark?"
  }
]
first_response = client.chat.completions.create(
  model="gpt-4o",
  messages=messages,
  tools=[dbvs_tool.tool]
)

# Execute function code and parse the model's response and handle function calls.
tool_call = first_response.choices[0].message.tool_calls[0]
args = json.loads(tool_call.function.arguments)
result = dbvs_tool.execute(query=args["query"])  # For self-managed embeddings, optionally pass in openai_client=client

# Supply model with results – so it can incorporate them into its final response.
messages.append(first_response.choices[0].message)
messages.append({
  "role": "tool",
  "tool_call_id": tool_call.id,
  "content": json.dumps(result)
})
second_response = client.chat.completions.create(
  model="gpt-4o",
  messages=messages,
  tools=[dbvs_tool.tool]
)

セルフマネージド埋め込みを使用して直接アクセス インデックスまたは差分同期インデックスを使用するシナリオでは、 VectorSearchRetrieverTool を構成し、カスタム埋め込みモデルとテキスト列を指定する必要があります。 埋め込みを提供するためのオプションを参照してください。

次の例では、VectorSearchRetrieverToolキーとcolumns キーを使用してembeddingを構成する方法を示します。

from databricks_openai import VectorSearchRetrieverTool

vs_tool = VectorSearchRetrieverTool(
    index_name="catalog.schema.index_name", # Index name in the format 'catalog.schema.index'
    num_results=5, # Max number of documents to return
    columns=["primary_key", "text_column"], # List of columns to include in the search
    filters={"text_column LIKE": "Databricks"}, # Filters to apply to the query
    query_type="ANN", # Query type ("ANN" or "HYBRID").
    tool_name="name of the tool", # Used by the LLM to understand the purpose of the tool
    tool_description="Purpose of the tool", # Used by the LLM to understand the purpose of the tool
    text_column="text_column", # Specify text column for embeddings. Required for direct-access index or delta-sync index with self-managed embeddings.
    embedding_model_name="databricks-bge-large-en" # The embedding model. Required for direct-access index or delta-sync index with self-managed embeddings.
)

詳細については、VectorSearchRetrieverToolを参照してください。

ローカル ツールの準備ができたら、エージェント コードの一部として直接運用するか、Unity カタログ関数に移行できます。これによって、検出可能性とガバナンスが向上しますが、一定の制限があります。

次のセクションでは、取得元を Unity カタログ関数に移行する方法について説明します。

Databricks の外部でホストされているベクター インデックスのクエリを実行する

ベクター インデックスが Azure Databricks の外部でホストされている場合は、Unity カタログ接続を作成して外部サービスに接続し、エージェント コードでその接続を使用できます。 「AI エージェント ツールを外部サービスに接続する」を参照してください。

次の例では、PyFunc フレーバー エージェントに対して Databricks の外部でホストされているベクター インデックスを呼び出すレトリバーを作成します。

  1. 外部サービス (この場合は Azure) への Unity カタログ接続を作成します。

    CREATE CONNECTION ${connection_name}
    TYPE HTTP
    OPTIONS (
      host 'https://example.search.windows.net',
      base_path '/',
      bearer_token secret ('<secret-scope>','<secret-key>')
    );
    
  2. Unity カタログ接続を使用して、エージェント コードで取得ツールを定義します。 この例では、MLflow デコレーターを使用してエージェント トレースを有効にします。

    MLflow レトリバー スキーマに準拠するには、retriever 関数は List[Document] オブジェクトを返し、Document クラスの metadata フィールドを使用して、 doc_urisimilarity_scoreなど、返されるドキュメントに属性を追加する必要があります。 MLflow ドキュメントを参照してください

    import mlflow
    import json
    
    from mlflow.entities import Document
    from typing import List, Dict, Any
    from dataclasses import asdict
    
    class VectorSearchRetriever:
      """
      Class using Databricks Vector Search to retrieve relevant documents.
      """
    
      def __init__(self):
        self.azure_search_index = "hotels_vector_index"
    
      @mlflow.trace(span_type="RETRIEVER", name="vector_search")
      def __call__(self, query_vector: List[Any], score_threshold=None) -> List[Document]:
        """
        Performs vector search to retrieve relevant chunks.
        Args:
          query: Search query.
          score_threshold: Score threshold to use for the query.
    
        Returns:
          List of retrieved Documents.
        """
        from databricks.sdk import WorkspaceClient
        from databricks.sdk.service.serving import ExternalFunctionRequestHttpMethod
    
        json = {
          "count": true,
          "select": "HotelId, HotelName, Description, Category",
          "vectorQueries": [
            {
              "vector": query_vector,
              "k": 7,
              "fields": "DescriptionVector",
              "kind": "vector",
              "exhaustive": true,
            }
          ],
        }
    
        response = (
          WorkspaceClient()
          .serving_endpoints.http_request(
            conn=connection_name,
            method=ExternalFunctionRequestHttpMethod.POST,
            path=f"indexes/{self.azure_search_index}/docs/search?api-version=2023-07-01-Preview",
            json=json,
          )
          .text
        )
    
        documents = self.convert_vector_search_to_documents(response, score_threshold)
        return [asdict(doc) for doc in documents]
    
      @mlflow.trace(span_type="PARSER")
      def convert_vector_search_to_documents(
        self, vs_results, score_threshold
      ) -> List[Document]:
        docs = []
    
        for item in vs_results.get("value", []):
          score = item.get("@search.score", 0)
    
          if score >= score_threshold:
            metadata = {
              "score": score,
              "HotelName": item.get("HotelName"),
              "Category": item.get("Category"),
            }
    
            doc = Document(
              page_content=item.get("Description", ""),
              metadata=metadata,
              id=item.get("HotelId"),
            )
            docs.append(doc)
    
        return docs
    
  3. レトリバーを実行するには、次の Python コードを実行します。 必要に応じて、 要求にベクター検索フィルター を含めて結果をフィルター処理できます。

    retriever = VectorSearchRetriever()
    query = [0.01944167, 0.0040178085 . . .  TRIMMED FOR BREVITY 010858015, -0.017496133]
    results = retriever(query, score_threshold=0.1)
    

リトリーバーにトレースを追加する

MLflow トレースを追加して、レトリバーを監視およびデバッグします。 トレースを使用すると、実行の各ステップの入力、出力、およびメタデータを表示できます。

前の例では、 メソッドと解析メソッドの両方に __call__を追加しています。 デコレーターは、関数が呼び出されたときに開始し、戻ったときに終了する スパン を作成します。 MLflow は、関数の入力と出力、および発生した例外を自動的に記録します。

LangChain、LlamaIndex、OpenAI ライブラリのユーザーは、デコレーターを使用してトレースを手動で定義するだけでなく、MLflow 自動ログを使用できます。 「 アプリケーションへのトレースの追加: 自動トレースと手動トレース」を参照してください。

import mlflow
from mlflow.entities import Document

## This code snippet has been truncated for brevity, see the full retriever example above
class VectorSearchRetriever:
  ...

  # Create a RETRIEVER span. The span name must match the retriever schema name.
  @mlflow.trace(span_type="RETRIEVER", name="vector_search")
  def __call__(...) -> List[Document]:
    ...

  # Create a PARSER span.
  @mlflow.trace(span_type="PARSER")
  def parse_results(...) -> List[Document]:
    ...

Agent Evaluation や AI Playground などのダウンストリーム アプリケーションでレトリバー トレースが正しくレンダリングされるようにするには、デコレーターが次の要件を満たしていることを確認します。

MLflow の互換性を確保するようにレトリバー スキーマを設定する

取得元または span_type="RETRIEVER" から返されたトレースが MLflow の標準取得スキーマに準拠していない場合は、返されたスキーマを MLflow の予期されるフィールドに手動でマップする必要があります。 これにより、MLflow がリトリーバーを適切にトレースし、ダウンストリーム アプリケーションでトレースをレンダリングできるようになります。

取得元スキーマを手動で設定するには:

  1. エージェント 定義するときにmlflow.models.set_retriever_schemaを呼び出します。 set_retriever_schemaを使用して、返されたテーブル内の列名を、primary_keytext_columndoc_uriなどの MLflow の予期されるフィールドにマップします。

    # Define the retriever's schema by providing your column names
    mlflow.models.set_retriever_schema(
      name="vector_search",
      primary_key="chunk_id",
      text_column="text_column",
      doc_uri="doc_uri"
      # other_columns=["column1", "column2"],
    )
    
  2. other_columns フィールドを含む列名の一覧を指定して、取得元のスキーマに追加の列を指定します。

  3. 複数のレトリバーがある場合は、各レトリバー スキーマに一意の名前を使用して、複数のスキーマを定義できます。

エージェントの作成時に設定された取得スキーマは、レビュー アプリや評価セットなどのダウンストリーム アプリケーションとワークフローに影響します。 具体的には、doc_uri 列は、取得元によって返されるドキュメントのプライマリ識別子として機能します。

  • レビュー アプリには、レビュー担当者が応答を評価し、ドキュメントの配信元を追跡するのに役立つdoc_uriが表示されます。 アプリ UI の確認を参照してください。
  • 評価セット では、 doc_uri を使用して、取得元の結果を定義済みの評価データセットと比較して、取得元の再現率と精度を判断します。 評価セット (MLflow 2) を参照してください。

Unity カタログ関数を用いたベクター検索の取得ツール

Databricks では、ほとんどのエージェント ツールに MCP サーバーを推奨していますが、Unity カタログ関数を使用したツールの定義は、プロトタイプ作成に引き続き使用できます。

モザイク AI ベクター検索インデックス クエリをラップする Unity Catalog 関数を作成できます。 この方法の特徴は次のとおりです。

  • ガバナンスと検出可能性を備えた運用環境のユース ケースをサポートします
  • 内部で vector_search() SQL 関数を使用します
  • 自動 MLflow トレースをサポート
    • page_contentエイリアスを使用して、関数の出力を metadataに合わせる必要があります。
    • 最上位レベルの出力キーとしてではなく、metadataを使用して、追加のメタデータ列を 列に追加する必要があります。

ノートブックまたは SQL エディターで次のコードを実行して、関数を作成します。

CREATE OR REPLACE FUNCTION main.default.databricks_docs_vector_search (
  -- The agent uses this comment to determine how to generate the query string parameter.
  query STRING
  COMMENT 'The query string for searching Databricks documentation.'
) RETURNS TABLE
-- The agent uses this comment to determine when to call this tool. It describes the types of documents and information contained within the index.
COMMENT 'Executes a search on Databricks documentation to retrieve text documents most relevant to the input query.' RETURN
SELECT
  chunked_text as page_content,
  map('doc_uri', url, 'chunk_id', chunk_id) as metadata
FROM
  vector_search(
    -- Specify your Vector Search index name here
    index => 'catalog.schema.databricks_docs_index',
    query => query,
    num_results => 5
  )

このレトリバー ツールを AI エージェントで使用するには、UCFunctionToolkitでラップします。 これにより、MLflow ログで RETRIEVER スパンの種類が自動的に生成され、MLflow を介した自動トレースが可能になります。

from unitycatalog.ai.langchain.toolkit import UCFunctionToolkit

toolkit = UCFunctionToolkit(
    function_names=[
        "main.default.databricks_docs_vector_search"
    ]
)
tools = toolkit.tools

Unity カタログ取得ツールには、次の注意事項があります。

  • SQL クライアントでは、返される行またはバイトの最大数が制限される場合があります。 データの切り捨てを防ぐには、UDF によって返される列値を切り捨てる必要があります。 たとえば、substring(chunked_text, 0, 8192) を使用して、大きなコンテンツ列のサイズを小さくし、実行中に行の切り捨てを回避できます。
  • このツールは vector_search() 関数のラッパーであるため、vector_search() 関数と同じ制限が適用されます。 制限事項を参照してください。

UCFunctionToolkitの詳細については、Unity カタログのドキュメントを参照してください。