Oracle AI Database Vector Store
Use the n8n Oracle AI Database Vector Store node to store documents in Oracle AI Database and retrieve relevant documents for chains, retrievers, tools, and AI agents.
Contents
- Requirements
- Connect the Nodes
- Use the Node
- Configure Node Parameters
- Configure Node Options
- Review Reference Links
Requirements
Your Oracle AI Database instance must support Oracle AI Vector Search. Create or select an Oracle credential in n8n. Connect an Oracle AI Database Embeddings sub-node when the vector store must generate embeddings.
Connect the Nodes
Use the Oracle AI Database Vector Store node in these workflow patterns:
- Insert and retrieve documents: Use the node as a regular node in the workflow connection flow.
- Connect to an AI Agent: Connect the vector store directly to the agent’s tools connector.
- Use a retriever: Connect the Vector Store Retriever to the vector store, then connect the retriever to a Question and Answer Chain or another chain.
- Use a question-answer tool: Connect the Vector Store Question Answer Tool to the vector store, then connect the tool to an AI Agent.
Use the Node
The node supports these modes:
- Insert Documents: Store documents and their embeddings in an Oracle vector table. The node creates the table when it does not exist.
- Get Many: Search the vector table and return the configured number of matching documents.
- Retrieve Documents (As Vector Store for Chain/Tool): Expose the vector store to a retriever or chain.
- Retrieve Documents (As Tool for AI Agent): Expose the vector store as a named tool that an AI agent can call.
Configure Node Parameters
Configure the following fields in the selected mode:
| Mode | Parameters |
|---|---|
| Get Many | Table Name, Prompt, and Limit. |
| Insert Documents | Table Name. |
| Retrieve Documents as a vector store | Table Name. |
| Retrieve Documents as an AI Agent tool | Name, Description, Table Name, and Limit. |
Use a specific tool name and description in AI Agent mode. A precise description helps the language model decide when to call the tool.
Configure Node Options
Distance Strategy
In Get Many and Retrieve Documents modes, select the distance strategy used to compare vectors:
- Cosine
- Inner Product
- Euclidean
- Manhattan
- Euclidean Squared
- Hamming
Metadata Filter
Use metadata filters in Get Many and Retrieve Documents modes. Multiple fields entered in the n8n interface are combined with AND.
Advanced filters pass through to Oracle AI Vector Search and can include arrays, nested filters, comparison operators such as $gte, exclusion operators such as $nin, and logical operators such as $and.
Set document metadata with the document loader before inserting documents.