• ACCURACY
  • RELIABILITY
  • EFFICIENCY
  • HUMAN

How to create contextualized embeddings with your AI

Connect Voyage AI once. Your agent then builds embeddings that understand the text around a specific chunk to fix broken retrieval.

Ask AI about this page

Short answer

How can I make my AI embeddings more contextual?

Use the create_contextualized_embeddings tool to generate vectors that account for the surrounding text of a document chunk. This prevents the AI from losing the meaning of a sentence when it's pulled out of its original paragraph. You'll get much more accurate search results in your RAG system.

The results of better embeddings.

What you get from better retrieval.

What happens when your agent uses Voyage AI.

ACCURACY

Better retrieval hits

The AI uses contextualized vectors to find the exact information you need. It stops pulling irrelevant snippets that lack context.

RELIABILITY

Reduced RAG errors

Your agent avoids the common trap of retrieving isolated sentences that don't make sense on their own. This leads to more coherent answers.

EFFICIENCY

Cleaner vector stores

The tool creates high-quality embeddings that represent the true intent of your documents. You spend less time tuning search parameters.

HUMAN

Manual review needed

You still need to check the final answers for logic. The AI handles the math of the embeddings, but you verify the reasoning.

The workflow

What your AI does when the data arrives.

The AI handles the heavy lifting of vector generation.

  1. Analyze text chunks

    Your agent looks at the specific text segments you want to embed.

    create_contextualized_embeddings

  2. Inject surrounding context

    The tool pulls in the neighboring text to ensure the embedding captures the full meaning.

    create_contextualized_embeddings

  3. Generate vectors

    The agent produces the final contextualized embeddings for your database.

    create_contextualized_embeddings

  4. Verify embedding quality

    You review the resulting vectors to ensure they align with your retrieval goals.

    create_contextualized_embeddings

Try it

Copy these to start.

Use these prompts to trigger the embedding process.

Starting points

Swap out the bracketed text for your actual file names or database IDs.

Voyage AI Connector

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

Connector linkhttps://edge.vinkius.com/vk_preview_x7lwFbbAdz5StvxoOu2Tkh9eDMu9Uh5i3GKObLMP/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 — Voyage AI capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "voyage-ai-ai-embeddings-api-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_x7lwFbbAdz5StvxoOu2Tkh9eDMu9Uh5i3GKObLMP/mcp"
    }
  }
}
Copy into chat04

  • Create contextualized embeddings for the text in my 'annual_report.pdf' file.

  • Use the Voyage AI tool to embed the contents of my 'customer_feedback.csv' sheet.

  • Generate contextualized embeddings for the documents in my 'project_docs' folder.

  • Run the embedding process on the snippets found in my 'knowledge_base' repository.

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

Start here

Connect Voyage AI once, then ask.

Just link your account once. Your credentials stay encrypted, and you can immediately start asking your agent to process your documents.

Connect Voyage AI to your AI

FAQ

How this task behaves.

  • 01

    Does this tool change my original documents?

    No. The tool only reads your text to create mathematical embeddings. Your original files remain untouched.

  • 02

    Can the AI delete my existing vector database?

    No. This tool only creates new embeddings. It does not have the capability to delete or modify your existing database entries.

  • 03

    How does this differ from standard embeddings?

    Standard embeddings look at a single chunk in isolation. This tool looks at the surrounding text to give that chunk more meaning.

  • 04

    Do I need to provide the context manually?

    No. The tool is designed to handle the contextualization as part of the embedding process.

  • 05

    Can the AI decide which parts of the document to embed?

    The AI follows your instructions, but you should specify which files or sections you want to process.

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

Connect Voyage AI to Claude, Cursor, ChatGPT & more