• MATCHES
  • SCHEMA
  • HUMAN
  • DATA

How to perform vector similarity searches with your AI

Connect your AI to Oracle 23ai once. From there, your agent can query vector data and find related records using native distance metrics.

Ask AI about this page

Short answer

How do I run vector similarity searches with my AI?

Your AI uses the Oracle Vector DB MCP to execute similarity queries against your 23ai tables. It identifies the closest matches based on the vector distance metrics stored in your database. You'll see the specific results and table structures below.

Search outcomes

Where your data lands

What happens when you run a similarity query.

MATCHES

Nearest neighbor results

The AI identifies the specific rows that most closely match your input vector. You get the raw data from those matching records.

SCHEMA

Table structure details

The AI looks up the columns and vector types in your Oracle tables. This ensures the search query matches your actual database schema.

HUMAN

Manual query refinement

If the search returns too many or too few results, you decide how to adjust the distance thresholds or the search parameters.

DATA

Vector metadata access

The AI pulls the associated text or metadata tied to the found vectors. You get the context behind the mathematical matches.

The workflow

What your AI does when the search arrives.

The AI moves from inspecting your database to executing the actual similarity math.

  1. Inspect the database

    The AI checks which tables are available in your Oracle instance to find where the vector data lives.

    list_tables

  2. Verify table structure

    It examines the specific columns and vector dimensions to ensure the query is valid for your schema.

    describe_table

  3. Execute similarity search

    The AI runs the vector search using the native distance metrics defined in your Oracle 23ai environment.

    vector_search

  4. Format results

    The AI parses the returned neighbors and presents the most relevant data points to you.

    vector_search

Try it

Copy these to start.

Use these prompts to begin interacting with your Oracle data.

Starting points

Replace the bracketed text with your actual table names or specific search terms.

Oracle Database Connector

You're all set. Choose your MCP client and follow the setup instructions.

Connector linkhttps://edge.vinkius.com/vk_preview_Bs3fBgLmPqQPV1iPDfqOKsvG2g10fcqpkDZdhwp8/mcp

Claude Desktop

Follow the steps below to connect in seconds.

  1. 1In Claude Desktop, open Settings → Connectors.
  2. 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
  3. 3Click Add and start a new chat — Oracle Database capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "oracle-vector-db-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_Bs3fBgLmPqQPV1iPDfqOKsvG2g10fcqpkDZdhwp8/mcp"
    }
  }
}
Copy into chat04

  • List all the tables in my Oracle database so I can see where the vectors are stored.

  • Show me the structure of the 'product_embeddings' table.

  • Find the 5 nearest neighbors for this vector: [0.12, 0.45, -0.08] in the 'image_vectors' table.

  • Search the 'customer_profiles' table for vectors similar to this user description.

  • Claude
  • ChatGPT
  • Cursor
  • VS Code
  • Windsurf
  • Claude Code
  • JetBrains
  • Cline

Start here

Connect Oracle Vector DB once, then ask.

You only need to link the MCP once. Your credentials stay encrypted, and you can immediately start querying your Oracle 23ai instance through your preferred AI client.

Connect Oracle Vector DB to your AI

FAQ

How this task behaves

  • 01

    Can the AI create new vector columns in my database?

    No. The AI can only read existing table structures and perform searches on existing data.

  • 02

    Does the AI delete any of my vector data?

    No. This MCP only provides tools for reading and searching data, not for deleting or modifying it.

  • 03

    Can I use different distance metrics like Cosine or Euclidean?

    Yes, the AI uses the native distance metrics you have configured within your Oracle 23ai tables.

  • 04

    How does the AI know which table to search?

    You tell the AI which table to use in your prompt, or it can use the list of tables to find the right one.

  • 05

    Can the AI write complex SQL joins for me?

    The AI uses the specific vector search tool to find neighbors, but it does not perform arbitrary SQL writes.

  • More questions about Oracle Vector DB? The Connector page answers them. See everything the Oracle Vector DB Connector can do

Connect Oracle Database to Claude, Cursor, ChatGPT & more