How Data complements AI

How Data complements the AI layer.

Agentic Friendly

Data does not replace AI. It complements it by adding the data movement, discovery, analytical storage, distributed processing, streaming, and BI services that some AI solutions need around the AI layer.

Core relationship

LayerMain role
FoundationRuns and secures the platform
AIServes and manages GenAI capabilities
DataAdds enterprise data and analytics services

Practical extension points

  • Data ingestion in Studio brings data from external systems through managed source and destination connectors
  • The data catalog shows the governed tables each user can discover
  • Compute in Studio and the Spark helper run distributed processing
  • Kafka supports event-driven data movement and streaming use cases
  • Kubeflow provides ML pipeline orchestration and notebook-based development
  • MLflow, reached through the BSQAI API, records experiments and registers models for serving
  • Lakekeeper and Spark Operator provide catalog-backed Iceberg processing
  • Tuples authorizes Data resources per tenant and publishes optional audit events
  • Superset exposes business-facing dashboards and BI views over analytical data

What this enables

  • AI systems that depend on synchronized enterprise data
  • Models trained in Developer Workspace and served by the AI layer from the MLflow registry
  • Analytics and reporting experiences around platform data
  • Event-driven architectures that connect AI and data services
  • A cleaner separation between AI application logic and enterprise data operations

On this page