What this API does
The Supabase Vector API is a tool that enables developers to perform vector similarity searches directly within PostgreSQL databases powered by Supabase. It features pgvector for managing high-dimensional vector embeddings, which is essential for AI and machine learning applications. The API supports efficient k-nearest neighbor searches with multiple distance metrics such as cosine and Euclidean distances, making it suitable for semantic search and recommendation systems.
How it works
To use the API, developers send RESTful HTTP requests to specific endpoints designed for vector operations. These include batch inserting vectors, searching for nearest embeddings based on various distance metrics, and retrieving vector details by ID. The API returns results in JSON format, facilitating integration with modern programming languages and frameworks.
Authentication
Authentication is managed via Supabase's built-in services. Developers must set up an account and configure authentication to utilize the API within their projects, ensuring secure access to data and services.
Example usage
/vectors- Batch inserts vectors into the database./search- Retrieves nearest vectors based on input embeddings./vectors/{id}- Fetches details of a specific vector by ID.
Limits
Ideal use cases
- Building AI-driven recommendation systems.
- Implementing semantic search functionalities in applications.
- Creating personalized user experiences through content relevance.
- Integrating advanced data search capabilities into existing systems.