Collate Inc., the AI for Data company, is introducing Collate 2.0, a major new release from the creators of OpenMetadata, adding a new AI-native experience powered by Collate’s semantic context graph.
According to the company, Collate 2.0 unlocks the potential of AI for anyone who needs to organize and understand data—including analysts, data engineers, and governance professionals.
Now users can tag, classify, document, curate, search, and analyze data by describing what they want to do conversationally. Guided by Collate’s semantic context graph, Collate translates prompts into reliable action, the company said.
For example: A data engineer could build a landing page that monitors pipeline health. A governance lead could view pending approvals and policy gaps. An executive could check the health of their data estate. Each view is LLM-powered, interactive, and automatically adapted to the user’s persona.
For recurring tasks, users can now create scheduled AI workflows called automations. Describe a task in natural language, and Collate will identify relevant agents and data sources, combine them into a workflow, and make sure it runs at the appropriate time. Five built-in templates cover common jobs such as regulatory compliance tagging, domain assignment, and data quality testing, Collate said.
Whether working with the AI or collaborating with colleagues, Collate saves conversations in memory as reusable knowledge, so teams can build on each other’s work.
Powering all this new functionality is Collate AI: a rich set of tools and automations for building and maintaining Collate’s AI capabilities, Collate noted.
Collate 2.0 provides agents for Metadata Ingestion, Documentation, Data Quality, Data Classification, Tiering, and more. It provides a no-code AI Studio for building custom agents, and AI SDK for teams that want to develop agents programmatically.
Collate’s new Context Center provides a central hub for managing semantics and memory. Every time a user corrects the AI in conversation, the change becomes a shared “memory nugget” inherited by future queries, capturing preferences and tribal knowledge and sharing them with every AI agent in the enterprise.
Collate 2.0 also includes a new Data Marketplace: an internal storefront for domains and data products with self-serve discovery, governed data product check-out, and end-to-end Data Access Request Workflows, so agents and teams easily find preferred sources of data.
"The data team is the last persona in the enterprise to get a real AI workflow," said Sriharsha Chintalapani, CTO of Collate. "Developers got Claude Code and Codex early because code already lives inside a designed system of context. Data did not—until now. Collate 2.0 closes that gap, combining agents, automations, developer tools, and a marketplace for governed data products on top of OpenMetadata, the open context layer for humans, AI assistants, and agents, accessible across every cloud and every LLM. "
Collate 2.0 gives AI agents semantic context. Context tells AI what data exists. Semantics tells AI what it means. Memory makes the knowledge persistent and shareable, the company said.
Collate uses more than 130 native connectors to map every enterprise data source to a single metadata graph to provide Context. It supplies agents with Semantics using a rich ontology to understand business concepts, reason over data, and reach accurate conclusions. And its shared Memory ensures feedback given to one agent benefits every agent, making every AI operation more cost-efficient and effective over time, the vendor said.
Collate 2.0 will be generally available in July 2026.
For more information about this news, visit www.getcollate.io.