Starburst Data Inc. today is expanding its push into enterprise artificial intelligence with the introduction of a platform aimed at allowing organizations to run AI workloads directly on distributed data without moving it into centralized repositories.

Announced at the company’s AI+Datanova event in Miami, the Enterprise Intelligence Platform combines Starburst’s federated query technology with a new business context layer, governance controls and agentic AI capabilities to address the inconsistency and fragmentation problems that confront organizations building scalable AI models.

The core of the announcement is the general availability of Aida, an AI assistant that lets business users query and analyze enterprise data using natural language. The company is also introducing AI-Ready Data Products that automate common tasks, new Apache Iceberg management capabilities and a bring-your-own-cloud deployment option aimed at customers that want greater infrastructure control.

The new platform is positioned as a “fully reimagined enterprise intelligence stack” that extends beyond the company’s original focus on federated analytics, said Jitender Aswani, senior vice president and global head of engineering and security at Starburst.

The company asserted that AI adoption is being slowed by the complexity of enterprise data environments, with data scattered across cloud platforms, data lakes, applications and operational systems.

Starburst’s solution is what it calls an enterprise context layer. The technology aggregates metadata, business rules and semantic relationships from catalogs, business intelligence tools and enterprise applications into a graph structure that AI systems can use to better understand enterprise data.

“Without shared context, different things had different meaning to different users,” Aswani said. “’Churn’ has different meanings in finance, product and sales.”

The platform is intended to reduce hallucinations and inconsistent answers from AI models by providing what Starburst calls “ground truth” across distributed data environments. The company said the context layer can ingest metadata from popular systems such as Collibra, Tableau and most enterprise catalogs without requiring data to be centralized.

Starburst is also expanding its support for Apache Iceberg, the open table format that anchors many data lakehouse architectures. Managed Icehouse includes Icehouse Ingest for batch and streaming ingestion and Icehouse LakeOps for automated table optimization and maintenance.

Aswani said managing Iceberg environments has become increasingly difficult as data volumes grow.

“Automated data maintenance, which we call LakeOps, allows our customers to just configure how they want their warehouse managed, and we will automatically analyze tables, see what needs compaction and remove snapshots,” he said.

The company is positioning its approach to governance and model flexibility as distinctive in an increasingly crowded enterprise AI market. The platform includes what Starburst calls an agent control plane with guardrails, evaluation capabilities and support for customer-selected models.

“We’re not tied to a single frontier model provider,” Aswani said. “Our customers can bring open-source models like Gemma, Qwen or any other model.”

The new bring-your-own-cloud deployment model, currently in preview, is aimed at enterprises concerned about sovereignty, compliance and infrastructure control. Customers retain ownership of computer, networking and data resources while still using Starburst’s managed services.

Aida is generally available immediately on Starburst Galaxy and Starburst Enterprise Platform, while several of the new platform capabilities are entering preview as part of the company’s May 2026 long-term support release.

Image: Starburst