Since its inception, Starburst Data has enabled users to connect disparate data sources to fuel data-driven decisions. On Wednesday, the vendor unveiled Context Studio to enable customers to integrate context spread across different systems to better inform AI agents.
Starburst's Context Studio, which the vendor introduced during its Launch Point virtual user conference and is now in private preview, is built to engender trust in AI outputs. Context Studio automatically collects the relevant information that agents require to deliver accurate outputs -- without forcing users to build pipelines that move data and business logic into a central location -- and applies it alongside data when agents call on data to inform their actions.
In addition to bringing context together through a federated architecture, Context Studio ensures that any agent that connects to Starburst over a Model Context Protocol server receives the same definitions, enables users to approve and trace every definition before agents use them and features a context loop so feedback gets turned into proposed improvements that can be approved or declined before being changed and influencing other answers.
"Context Studio is a significant new capability that expands on Starburst's federated data processing and data product capabilities to enable customers to identify, reconcile and certify context from across their data estate and provide knowledge that enables trusted agentic decision-making," Matt Aslett, an analyst at ISG, told TechTarget.
In addition, he noted that because Context Studio builds on Starburst's federated architecture, it is somewhat different than the context layers being built by competing vendors that only enable agents to discover context within their platform environment.
"All providers across the analytics and data stack are refactoring their products to provide context to AI agents, [but] … while other platform-specific context environments may be limited to the information available within that platform, Context Studio builds on Starburst's established position as an enabler of federated data processing across a heterogeneous data landscape," Aslett said.
Fueled by feedback
Trouble connecting agents with relevant context remains one of the main reasons the majority of AI initiatives still fail.
Hyperscalers AWS, Google Cloud and Microsoft have already introduced capabilities aimed at enabling AI tools to access contextually appropriate information in real time so they can deliver trusted outputs, as have data platform providers including Databricks and Snowflake. In addition, more specialized vendors such as Alteryx, Informatica and Tableau have all unveiled tools designed to connect agents with context.
"Context Studio is a significant new capability that expands on Starburst's federated data processing and data product capabilities to enable customers to identify, reconcile and certify context from across their data estate."

Matt Aslett
Analyst, ISG
Now, Starburst is similarly addressing the context gap for AI agents after both observing customers trying to build agents and working to build Aida, its own AI-powered assistant, according to Justin Borgman, CEO of Starburst.
"As customers started building on top of [Aida], and as we built more of our own agents internally, we all hit the same wall: context," he told TechTarget. "Definitions and business logic lived in different tools and none of it worked together, so agents couldn't reach the level of trust needed to move from pilot to production. That's why we built Context Studio to harvest and serve the context you already have."
Starburst expects to make Context Studio generally available during the first quarter of 2027 after a public preview period beginning in January, according to Borgman. In the interim, the vendor is monitoring how well it surfaces and addresses conflicting definitions, whether it gives different agents consistent answers and whether it preserves the access controls customers already have in place.
While Context Studio is in private preview for customers, Starburst is already running Aida on Context Studio. Introduced in April and made generally available in May, Aida draws on governed definitions in Context Studio to provide responses based on context that is federated across an enterprise's various sources.
Devin Pratt, an analyst at IDC, noted that the significance of Context Studio is that it adds meaning to the data that Starburst users can already query wherever it lives. That meaning is what enables agents to find and operationalize context to make their responses more accurate and trustworthy.
"Agents answer from definitions, [and] Context Studio helps make sure they're the right ones," Pratt told TechTarget.
Meanwhile, like Aslett, Pratt added that while competing vendors are also developing context layers, Starburst stands out by applying definitions to data where it's created and stored rather than forcing customers to build costly and complex pipelines to import data into a centralized environment.
"Most approaches ask enterprises to move their definitions in," Pratt said. "Starburst brings context to where the data already lives."
Designed to deliver context, but more needed
Aslett noted that Context Studio includes many of the core components it takes to connect agents with context, including business terms, rules and metrics, and both business domains and query histories. However, like many vendors in the process of trying to better enable customers to build trustworthy agents, Starburst would be wise to also include information related to business decisions such as reasoning and planning, and information related to the outcomes of previous decisions and actions.
"This is an aspect of context that remains a work in progress for many providers and would be a potential addition to Context Studio as its development progresses," Aslett said.
Pratt likewise noted that capabilities such as source-linked definitions, approvals and per-user policies provide a solid foundation for Context Studio to deliver on its promise of providing agents with governed, consistent context. Adding the same versioning capabilities that Starburst applies to data products would be a logical improvement, he suggested.
"Trust is traceable -- every definition should show its source, its approver and its history," Pratt said.
Eric Avidon is a senior reporter for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.