Vectors Connector - Mule 4
home
MuleSoft Vectors Connector Overview
MuleSoft Vectors Connector allows MuleSoft customers to integrate with a variety of vector databases to improve the output of large language models (LLMs). It enables storing, retrieving, and searching vector embeddings of data which are crucial when implementing retrieval augmented generation (RAG)-based architectures.
Key Features
Vectors Connector provides MuleSoft customers with the ability to:
- Generate and query vector embeddings: Allows for semantic search across a corpus of enterprise data to augment user-provided prompts
- Store vectors in database of choice: Easily configure integration with various market-leading vector databases for future use of embeddings
- Process documents: Parse documents and chunk text into smaller, more manageable sections to generate more accurate embeddings
- Integrate with other relevant MuleSoft AI Connectors: Serve as memory for LLMs-integrated Mule apps via Anypoint Connectors (Inference, MuleSoft AI Chain, and Einstein AI)
Associated Use Cases
Deliver innovative and new user experiences by grounding LLMs in key enterprise knowledge. Possibilities include:
- Customer support AI agents - Enable customer-facing AI agents to analyze reported issues based on historical knowledge and documented best practices
- Conversational data analysis - Provide enterprise users with the option to query corporate data using natural language
- Employee chatbots - Empower AI agents to answer questions using information housed in a vector database
NOTE: This MuleSoft Vectors Connector is a Select Connector and is licensed under the terms of Customer's Main Services Agreement with Salesforce.
For the avoidance of doubt, this MuleSoft Vectors Connector is separate and distinct from the open source MuleSoft Vectors connector which may be available on Github. The terms governing the use, modification, and contributions to the MuleSoft Vectors connector which may be available on GitHub are exclusively subject to the applicable open-source license(s) associated therewith, and any such use, modification, or contribution is not supported by Salesforce.