Starburst is adding agentic AI capabilities to its platform, including a pre-built agent for insight exploration as well as tools and tech for building custom agents. It also added a new data catalog, moved a number of previously unveiled features into general availability, and announced an investment by one of its customers, Citibank.
Starburst built its data platform around Trino, the fast SQL query engine that forked from Presto. Starburst still does 90% of the Trino development work, and makes Trino available to its customers. But the company’s hybrid lakehouse platform has lately emerged as the main reason that customers choose Starburst.
Lately, those Starburst customers–including nine of the top 15 banks in the country–have been playing around with building and deploying agents on the Starburst Enterprise (which deploys anywhere, including on-prem) and Starburst Galaxy (the company’s managed lakehouse, available on all three public clouds).
That customer work directly led to today’s agentic AI announcements, said Justin Borgman, the company’s CEO and co-founder.
“Our customers brought us into this by building their own agents on top of Starburst and building their own applications that leverage AI functionality and saying, hey, we want more capability native in the platform,” he told BigDATAwire.
Those new agentic AI capabilities include Starburst AI Workflows, which includes a collection of capabilities, including vector-native AI search, AI SQL functions, and AI model access governance functions.
The AI search functions include a built-in vector store that allows users to convert data into vector embeddings and then to search against them. Starburst is storing the vector embeddings in Apache Iceberg, which it has built its lakehouse around. Starburst may be the first vendor to store vector embeddings in Iceberg.
Putting the embeddings into Iceberg just makes sense, Borgman says. “We’re really doubling down on Oceberg as the open format of choice for customers, and that means that full end-to-end RAG workflow can now be done within Starburst, within your lakehouse, if you will.”
Starburst is positioning its platform for big data, advanced analytics, and AI
Trino is a SQL engine, and SQL has been central to Starburst’s lakehouse. So it makes sense that Starburst would mix SQL functionality with its new AI push. Specifically, the new AI SQL help customers by enabling them to run prompts and built-in LLM tasks directly from SQL commands.
Starburst also announced a new pre-built AI agent that customers can use out of the box. This built-in agent gives customers a conversational interface they can use to ask natural language questions of data stored in the Starburst platform, or to document the data in preparation for building a data product out of it.
Early testing shows that customers are going to use the built-in Starburst agent in a variety of ways, Borgman says.
“Some of them are using it as really like the copilot to their business analyst or data scientist, where they’re now able to ask questions of the data, much the way that you would with GPT, but leveraging your own enterprise data,” Borgman says. “And then [some are using it with] our data product functionality, which allows you to create the business context around the data itself, that sort of business metadata that becomes super valuable in terms of minimizing hallucinations and giving you the most accurate response”
Finally, Starburst’s new governance functions play a big role in ensuring that nothing untoward occurs with customers’ new AI applications and AI workflows. These governance functions give customers fine-grained control over which particular pieces of data go to which particular models, Starburst says. These controls are important not only from a regulatory compliance perspective, but also for controlling costs, Starburst says.
Starburst CEO Justin Borgman
“This is what our enterprises need,” says Nathan Vega, Starburst product marketing manager. “Being able to know that my agent in the US is going to [be connected] to the right data, the right models and be in compliance–and same for EU or Singapore or wherever else they’re working in the world–I think that’s a that’s a really important piece to start really making not only the analytics across the business real, but also to really improve and make AI something that they can adopt at scale.”
Starburst also announced a new data catalog. While the company’s offerings work with enterprise data catalogs from companies like Alation, Atlan, and Collibra, the company found that customers could benefit from having a built-in data catalog too.
The new AI capabilities discussed above are in private preview. Starburst also announced the general availability of a slew of previously announced features. This includes Starburst Galaxy auto-tagging, which allows LLMs to detect sensitive data at the column-level; a new streaming ingest for real-time updates from Kafka; and live table maintenance on Iceberg tables; and deployment set routing for routing queries across a defined set of clusters.
Starburst announced several features now in public preview, including nanosecond timestamps for precision time-sensitive analytics; native ODBC support for Trino (in June); scheduled materialized view refreshes with Iceberg MV automatic refresh, and full support for data products on Iceberg. Starburst announced that some functions are now generally available, including live table maintenance on Iceberg tables; automated table maintenance for Iceberg tables; an streaming data ingest via Kafka; and AI-powered auto-tagging for better data governance.
Finally, Starburst announced that Citibank is not only a customer, but an investor. The investment wasn’t materal, Borgman says, but was symbolic of its use and support of Starburst.
“Citibank we’ve been working with for a long time,” he says. “They just decided to go enterprise wide with us in a very major, significant way. And as a result of that, decided to actually also make a strategic investment in Starburst. We weren’t looking for capital, but they asked us if they could because we’ve now become that strategic, that critical to the operations of the bank that they wanted to have have a piece of piece of Starburst in the process.”
Related Items:
Starburst CEO Justin Borgman Talks Trino, Iceberg, and the Future of Big Data
Spark-to-Starburst Engine Swap Speeds Big Driving Data for Arity
Starburst Bolsters Trino Platform as Datanova Begins
The post Starburst Brings AI Agents Into Data Platform appeared first on BigDATAwire.
Leave a Reply