We use cookies to improve site navigation, analyze site usage, and enhance your user experience. Click "Accept" to enable cookies or "Reject" to reject cookies. To learn more, read our Privacy Policy.

Reimagine 2025 Recap: Unlocking AI-Ready Data, One Agent at a Time

James Nguyen
James Nguyen
··

7 min read

Cover Image for Reimagine 2025 Recap: Unlocking AI-Ready Data, One Agent at a Time

We’re entering a new era where humans and AI collaborate through shared understanding, not just shared tools. This means data teams are facing a new mandate: stop preparing for AI, and start working with it.

At Reimagine 2025, Collate’s virtual summit hosted in partnership with the OpenMetadata community, attendees tuned in to explore a new vision where metadata, semantics, automation, and governance converge to create an AI-ready foundation for innovation.

Couldn’t make it? You can now watch the summit on demand. It’s just over an hour, so you can listen over lunch. It’s packed with actionable insights and demos that make this vision of rethinking how we work with data feel directly within reach. If you’re a data platform leader, governance leader, practitioner, or executive, you won’t want to miss this.

Here’s a sneak peek at what you’ll learn.

From Metadata to Meaning: Keynote with Suresh Srinivas

Suresh Srinivas, CEO and Co-founder of Collate, and Founder of OpenMetadata, opened the summit with a challenge to the audience:

How do we make data truly ready for AI—not just accessible, but interpretable, trustworthy, and actionable?

“As we transition from people-ready data to AI-ready data, context alone is not enough. Context describes what data is. But AI needs semantics to give data meaning. Semantics creates a common language for humans and AI agents to interpret the data consistently and accurately.”

— Suresh Srinivas, CEO and Co-founder, Collate

These principles for bridging the gap between raw data and intelligence set the tone for Reimagine 2025.

Suresh traced his history from founding Hadoop/Hortonworks and serving as Chief Data Architect at Uber, to how OpenMetadata evolved from a foundational graph for data discovery, observability, and governance, into a semantic layer grounded in open standards, such as RDF and DCAT. This progression is enabling new interfaces where AI doesn’t just assist, but truly understands. “Semantics enables conversational interfaces where people can collaborate with AI instead of navigating complex UIs,” he added.

It was an inspiring reminder that AI-ready data isn’t just the finish line. It’s a mindset shift. And Collate is building the foundation to make it real.

AskCollate in Action: Talk to Your Data, Not Your Tools

What if managing your data felt as natural as asking a teammate a question?

That’s the idea behind AskCollate, Collate’s new AI-native conversational interface. The keynote laid the conceptual foundations, and the AskCollate demo showed the practical payoff.

In a series of rapid-fire use cases, Collate’s engineering team demonstrated how data stewards, analysts, and engineers can manage metadata, identify issues, and explore business metrics—all in natural language. With AskCollate, collaborative data investigation becomes fast and intuitive.

We saw a data steward:

  • Auto-generate new glossary terms.

  • Tag Snowflake tables based on relevance.

  • Assign metadata without writing a line of code.

Then an analyst jumped in to:

  • Query performance for eco-friendly product lines.

  • Build visualizations through chat.

  • Discover that eco products contributed 43% of revenue—and quickly uncover a reporting anomaly.

Instead of jumping between tools, everything happened within a single, conversational interface. And when the team uncovered a product showing $0 in sales in the revenue breakdown, it triggered an immediate quality investigation in the next demo.

Real-Time Quality and Root Cause Analysis

What if issue resolution took minutes instead of hours (or days) to deliver cleaner data ultimately?

Data quality issues lead to productivity delays and lost time. Endless back-and-forth between teams only compounds inefficiencies in the process. But with AskCollate, the workflow is different.

Two negative unit price values flagged by AskCollate revealed a clear source of error, helping the team trace why one product was inaccurately reporting zero revenue.

We watched a user trace lineage and profile data to generate 16 automated test cases to catch future anomalies—all in under 10 minutes. It was a moment that felt genuinely transformative, as the impact went beyond efficiency gains. By building trust in the data itself, teams can have genuine confidence in the reasoning behind every decision they make.

Collaboration and Accessibility

What if your go-to place for collaboration could also be your system of insight?

In this session, we saw how AskCollate seamlessly integrates with Slack, making it easy for data leaders to track asset performance, validate trust, and assign accountability, all without switching tools.

By querying product performance inside Slack, the team instantly retrieved results, traced them back to underlying tables, confirmed test case coverage, and identified the data owners. A concise board-ready summary was then generated, combining revenue performance, asset health, and team accountability in a single message.

This demo brought one of the summit’s biggest takeaways to life: conversational interfaces aren’t just for analysts. They give data leaders the insights they need, exactly where they work.

Introducing AI Studio: Build Agents That Work for You

What if anyone could build an AI agent that understands your data, your policies, and your workflows?

That’s the promise of AI Studio, Collate’s UI-based, no-code experience for building and scheduling your own AI agents. Harsha Chintalapani, Collate’s CTO and Co-founder, introduced the platform as the next evolution of Collate’s AI stack. One that puts autonomy and scale directly into the hands of data teams. According to Harsha, “You can't just drop a generic LLM into your data stack and expect it to understand your lineage, governance, or business context. That's why we are building AI agents grounded in metadata—that understand your tables, dashboards, and models. It's how we move from static catalogs to to active data systems.”

In the demo, Harsha created a GDPR Policy Agent from scratch. It scanned a Snowflake environment, identified tables with PII, tagged them with rich metadata, traced lineage across pipelines, and compiled a report. No engineering required.

“To work in production, AI agents need three things: full visibility across your data landscape, context-aware accuracy, and trustworthy, governed, explainable answers. You only get that when your agents are built on a unified metadata graph—exactly what Collate and OpenMetadata provide.”

— Harsha Chintalapani, CTO and Co-Founder, Collate

Agents can be scheduled to run continuously, enabling proactive data governance and management. The flexibility means teams can build agents for data documentation, policy enforcement, data quality, or whatever the business demands.

What would it look like to see the whole data pipeline powered by AI? What used to take weeks of back-and-forth between data modeling, engineering, and BI teams now happens in a single session. Harsha started with a question—what are the top eco-friendly products by revenue?—and carried the task all the way through to a live dashboard, all with AI automation:

  • Discovery: AskCollate found the trusted Tier 1 assets.

  • Querying: SQL was generated, executed, and verified.

  • Modeling: A dbt model was created, with a GitHub pull request and CI/CD checks.

  • Visualization: A Power BI dashboard was generated based on the dbt view.

It was a fully traceable, auditable pipeline created without a single ticket or back-and-forth between data teams. The agents all share the same foundational unified metadata graph, enabling a common understanding of the data, with a shared security model at every step of the process.

End-to-End Pipeline Automation

The next session took us deeper into Collate’s built-in agents designed to reduce manual lift and improve governance coverage.

  • The Collate Documentation Agent auto-generates column and table descriptions with full context.

  • The Collate Tiering Agent uses lineage and usage metrics to prioritize data assets.

  • The Collate Auto-Classification Agent identifies sensitive data, like SSNs, using pattern recognition and structure—not just column names.

Model Context Protocol (MCP): Connecting AI and Metadata

The MCP is a major architectural milestone that enables AI agents to interact with metadata systems in a governed and auditable manner.

Available on OpenMetadata, MCP enables agents to read context and safely write changes across systems. An OpenMetadata Developer Advocate demonstrated how an agent using GPT-4 and the open-source Goose framework can propagate schema-level certifications across hundreds of downstream tables in seconds.

MCP ensures that every agent operates with the same standards, controls, and semantic understanding to scale effectively.

Looking Ahead: A Future Built on Shared Intelligence

To close out the summit, Suresh brought us back to the core principles that have guided OpenMetadata from its inception—and how they’re evolving to meet the demands of an AI-ready future.

1. Shared understanding through semantics. Controlled vocabulary has evolved into a foundation for ontologies that lend meaning to data. This shift creates a common language between humans and AI.

2. Collaboration without boundaries. What was once limited to people-to-people workflows now includes AI agents as active participants. These agents not only execute but also interpret and collaborate alongside teams.

3. Automation that’s intelligent by design. Traditional rule-based systems have given way to autonomous agents that extend beyond task execution to decision support and proactive governance.

Together, these principles form the foundation for data intelligence without boundaries. This intelligence is explainable and open by default.

Whether you’re a data steward, engineer, analyst, or leader, this new approach to working with data is within reach. Collate and OpenMetadata are building the infrastructure to enable AI not only to understand your data but also to manage it for you.

Watch the full Reimagine 2025 summit on demand to see for yourself.

;