• SEMANTIC
  • CONTEXT
  • RELEVANCE
  • HUMAN

How to run kNN similarity searches with your AI

Connect your AI to Elasticsearch to find data based on meaning rather than just keywords. Once connected, your agent can query your vectors directly.

Ask AI about this page

Short answer

How can I perform kNN similarity searches using my AI?

Your AI uses the Elasticsearch Vector connector to execute multi-dimensional vector searches against your indices. This allows your agent to pull semantic matches that keyword searches often miss. You'll see how the AI interacts with your data below.

Search outcomes

What you get from these searches

The AI turns your natural language questions into vector queries.

SEMANTIC

Meaning-based matches

The AI finds documents that share the same intent or concept as your query. It looks at the underlying vectors instead of just matching words.

CONTEXT

Deep data retrieval

Your agent pulls relevant context from your Elasticsearch indices. This provides the necessary background for your AI to answer complex questions accurately.

RELEVANCE

High-precision results

By using kNN, the AI narrows down the most mathematically similar items. You get results that actually relate to the subject matter.

HUMAN

Manual review needed

The AI identifies the closest matches, but you decide if the semantic relationship is actually useful for your specific task.

The workflow

What your AI does when a search arrives.

The AI handles the heavy lifting of translating your intent into a vector query.

  1. Identify the target

    The AI looks at your request and determines which specific index contains the vectors needed for the search.

    get_index

  2. Execute the search

    The agent runs the similarity query against the specified index to find the nearest neighbors.

    search

  3. Process results

    The AI takes the raw vector matches and converts them into readable information.

    search

  4. Deliver answers

    Your agent presents the most relevant findings from the Elasticsearch index in a way that answers your question.

    search

Try it

Copy these to start.

Use these prompts to trigger your first vector searches.

Starting points

Swap out the bracketed text for your actual index names or search terms.

Elasticsearch Vector Connector

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

Connector linkhttps://edge.vinkius.com/vk_preview_kUWPWShL0BRuDE0k7qrS1ZYbK8Cj21vwQZnEsEiQ/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 — Elasticsearch Vector capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "elasticsearch-vector-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_kUWPWShL0BRuDE0k7qrS1ZYbK8Cj21vwQZnEsEiQ/mcp"
    }
  }
}
Copy into chat04

  • Find documents in the 'product_docs' index that are semantically similar to 'how to reset my password'.

  • Search the 'customer_feedback' index for entries that match the sentiment of 'the user is frustrated with latency'.

  • Using the 'knowledge_base' index, find any entries related to the concept of 'asynchronous processing'.

  • Run a similarity search in 'legal_contracts' to find clauses similar to 'force majeure provisions'.

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

Start here

Connect Elasticsearch Vector once, then ask.

Just link your account once. Your credentials stay encrypted, and you can immediately start querying your indices through your AI client.

Connect Elasticsearch Vector to your AI

FAQ

How this task behaves

  • 01

    Can the AI change my Elasticsearch index settings?

    No. The AI can only read data through the provided tools; it cannot modify your indices or configurations.

  • 02

    Does the AI delete any of my vector data?

    No. This connector is read-only. It has no capability to delete or overwrite your existing data.

  • 03

    Can I search across multiple indices at once?

    The AI can perform searches on specific indices you point it toward using the available tools.

  • 04

    Do I need to provide the vector embeddings myself?

    The AI uses the existing vectors already stored in your Elasticsearch indices to perform the similarity math.

  • 05

    Will the AI suggest new indices to create?

    No. The AI can only interact with indices that already exist in your Elasticsearch environment.

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

Connect Elasticsearch Vector to Claude, Cursor, ChatGPT & more