TileDB announced controlled availability of TileDB Carrara, an omnimodal data intelligence platform addressing the inability to discover, access, and analyze all data types, from structured tables to non-tabular data such as genomics, imaging, documents, models, agents, in a unified, governed environment.

TileDB Carrara treats all data as “modalities”—reachable, shareable data or code assets with inherent structure and domain-specific semantics. Whether population variants, bioimaging, PDF documents, Jupyter notebooks, user-defined functions, or external tables from existing data warehouses, assets become discoverable, and queryable.

This release of TileDB Carrara supports:

  1. Organization: A unified catalog for registering and searching across all modalities, supporting rich metadata filters and LLM-powered discovery
  2. Structuring: Leveraging TileDB’s shape-shifting array technology to efficiently capture any complex data type, optimized for cloud storage backends
  3. Collaboration: Teamspaces that function as secure data products, enabling users to create, share, and audit access to diverse assets within data clean rooms
  4. Analyze: An infrastructure for massively distributed computations, workflows, notebooks, and dashboards that manages all the data. 

While Carrara’s omnimodal architecture serves enterprises across sectors, its impact is particularly useful in life sciences and healthcare, where the integration of multimodal data (genomics, proteomics, clinical imaging, electronic health records, and sensor data) is essential.

https://www.tiledb.com/blog/introducing-carrara