Integrate Vector Search with AI Technologies

Key Concepts

Large Language Models (LLMs)

You can integrate Atlas Vector Search with LLMs and LLM frameworks to build AI-powered applications. When developing with LLMs, you might encounter the following limitations:

• Stale data: LLMs are trained on a static dataset up to a certain point in time.

• No access to local data: LLMs don't have access to local or personal data.

• Hallucinations: LLMs sometimes generate inaccurate information.

Retrieval-Augmented Generation (RAG)

Retrieval-Augmented Generation (RAG) is an architecture for LLM applications that's designed to address these limitations. In RAG, you perform the following actions:

1. Store your custom data as vector embeddings in a vector database.

2. Use vector search to retrieve semantically similar documents from the vector database. These documents augment the existing training data that LLMs have access to.

3. Prompt the LLM. The LLM uses these documents as context to generate a more informed and accurate response.

To learn more, see What is retrieval-augmented generation (RAG)?.

Frameworks

You can integrate Atlas Vector Search with the following open-source frameworks to store custom data in Atlas and implement RAG with Atlas Vector Search.

Services

You can also integrate Atlas Vector Search with the following AI services.

API Resources

Refer to the following API resources as you develop with AI integrations for Atlas Vector Search:

• LangChain Python API Reference

• LangChain JS/TS API Reference

• LlamaIndex API Reference

• Semantic Kernel Python API Reference

• Semantic Kernel C# API Reference