Skip to main content

Key Features

Shaped is a relevance database that excels in retrieving documents based on user-behaviour or semantic meaning.

Wide model library​

Shaped provides 15+ embedding and scoring models out of the box.

Huggingface integration​

The Shaped Engine API integrates with any encoder from Huggingface's library of text encoders or multi-modal encoders based on CHIP.

Multi-modal encoders​

The Shaped Engine API supports multi-modal encoders based on the CHIP architecture.

Simple data ingestion​

Shaped stores data in Tables. Tables can be populated with one of our 20+ connectors or through custom data upload.

Real-time data availability​

The Shaped data layer automatically indexes new data when it is available.

Flexible data pipeline support​

Tables can be generated from complex data pipelines. This includes:

  • Ingestion from 20+ data sources via built-in connector or custom data upload
  • Creating unified table views from distinct datasets

Serverless architecture​

Shaped provides a serverless platform. Your usage can scale as needed without explicitly provisioning more resources.