Introduction
Power your AI applications without data pipelines or vendor lock-in.
๐ Embedding Sync
Embedding Sync is a managed service that keeps your application databases and vector databases in sync. It automatically synchronizes your application database (initially supporting Postgres) with your vector database (currently supporting Pinecone).
Embedding Sync allows you to integrate AI-powered features into your applications without the hassle of managing data pipelines. By using your own database and Pinecone vector store, youโll always have your data and wonโt have to worry about vendor lock-in.
There are three core components:
- ๐ Sync Engine: Our managed service that securely connects to your application database and vector store. It performs real-time synchronization and supports a variety of data types.
- ๐ Embedding Models: We offer automated embedding generation for your data columns. We currently support the SentenceTransformers all-MiniLM-L6-v2 model or OpenAIโs text-embedding-ada-002. More models are on the way.
- ๐ฎ Playground: An front-end UI that lets you quickly query your embeddings, helping you to explore and understand your data.
To get started, check out our quickstart tutorial.
๐พ Currently Supported Databases
Current supported databases:
- Postgres
- Pinecone (as a vector database)
If your preferred database isnโt listed, we would love to hear from you.
๐ Roadmap
- Support for additional databases like MySQL, MongoDB.
- Logical replication support for Postgres.
- Expanded range of embedding models including custom models.
- Support for multi-column embeddings and metadata sync.