Starburst Data

Starburst offers an AI-powered Auto-Tagging feature in Starburst Galaxy that automatically scans data columns to detect sensitive information — such as names and email addresses — and tags them so that access and privacy policies can be applied. This feature reached general availability in 2025.

Galaxy's AI-powered Auto-Tagging uses large language models to scan data at the column level and identify personally identifiable information (PII). Once detected, columns are tagged, enabling attribute-based access control (ABAC) policies to be applied automatically. The feature supports human-in-the-loop review and custom user-defined classifiers, allowing data teams to confirm or adjust AI-generated tags before policies are enforced. This reduces the need for manual data reviews and helps organizations maintain compliance at scale without hand-labeling every data field.

Starburst deployed AIDA (AI Data Assistant), a conversational analytics tool that lets enterprise users ask questions about their data in plain English and receive answers drawn from distributed data sources without writing code. AIDA became generally available in May 2026 on both the Starburst Enterprise Platform and Starburst Galaxy.

AIDA accepts natural language queries, translates them into SQL using a large language model (LLM), executes those queries against configured data sources, and returns answers as summaries, charts, or data grids. It uses a ReAct (reason–act–observe) framework that iterates across live data samples and schema metadata to construct validated answers — going beyond simple text-to-SQL translation. Responses are tailored by user role (persona-specific): business leaders receive concise summaries while data engineers receive technical detail. Within the Starburst Enterprise Platform, AIDA supports multiple LLMs including models from Anthropic, OpenAI, and AWS Bedrock.

Starburst announced multi-agent AI infrastructure in October 2025, giving enterprise customers tools to build, manage, and orchestrate networks of AI agents that can query governed data, execute workflows, and interoperate via an open standard called Model Context Protocol (MCP). These capabilities were announced for general availability in Q4 2025.

The multi-agent infrastructure includes a new MCP (Model Context Protocol) server and an agent API that allow enterprises to create and coordinate multiple AI agents alongside Starburst's own built-in AI agent. The platform also provides unified access to vector stores — databases that store data in a format optimized for AI search — across Iceberg, PostgreSQL/PGVector, and Elasticsearch, enabling retrieval-augmented generation (RAG) workflows. Model usage monitoring dashboards let teams track LLM interactions, set usage limits, and audit AI activity for governance and cost control. These capabilities were originally announced at Starburst's AI & Datanova conference on October 9, 2025, with general availability targeted for Q4 2025.