Components
New Embedding Writers for Qdrant, PGVector and Pinecone
We're introducing three new experimental components for writing embeddings to popular vector databases: Qdrant, PGVector and Pinecone.
To support vector-native workflows and retrieval-augmented generation (RAG) scenarios, we’re introducing three new embedding writer components in Keboola:
These components allow you to push embedding vectors directly into the vector database of your choice, enabling seamless integration with AI and semantic search use cases.
Each writer is implemented as a standard data destination component, which means you can plug it directly into your existing flows after generating embeddings.
Key Benefits
- Vendor flexibility: Choose the backend that fits your architecture or cost model.
- Simple configuration: Accepts standard Keboola tables with pre-computed embeddings.
- Ready for experimentation: All components are currently marked experimental, so your feedback is highly welcome!
Component Previews
Qdrant
PGVector
Pinecone
If you're building AI-powered apps on top of Keboola, give these writers a try and let us know how they work for you!