The Refounding of Alation: From Data Catalogs to AI Agents

Almost thirteen years into building one of the most successful data catalog companies in the world, Satyen Sangani made an unexpected admission at revAlation Chicago: "I feel like I'm refounding the company."

Speaking to customers and partners at Alation's annual user conference, the CEO and co-founder didn't make this statement lightly. The data management landscape has fundamentally shifted, driven by two seismic forces: the demand for measurable business value and the emergence of agentic AI. These changes aren't just reshaping Alation's product roadmap—they're redefining what it means to manage data in the enterprise.

From Wikipedia to ChatGPT: The evolution of enterprise knowledge

When Alation launched in 2012, the opportunity was clear: we could build Wikipedia for enterprise data. Create a collaborative platform where data producers could document their knowledge and data consumers could discover what they needed. The data catalog was born, and with it, an entire category that Alation pioneered.

For a decade, this model worked. Budgets were effectively unlimited. Data was considered valuable in its own right. Success was measured by adoption—how many people used the catalog, how many searches were performed, and how much metadata was curated.

But then the world changed.

"In a world where zero interest rates went to much higher interest rates and where IT budgets became compressed, people would basically look around and say, I'm not sure that adoption necessarily always equals value," Sangani explained. Data teams suddenly found themselves under pressure to demonstrate clear ROI, to draw straight lines from their work to business outcomes.

At the same time, AI has transformed how people interact with the world. ChatGPT and other large language models don’t just answer questions—they solve problems. The "agentic web" has emerged, where users can go from asking a question to making a restaurant reservation. And the gap between what consumers expect and what enterprise data tools can deliver has become impossible to ignore.

For data management, this signals a profound change. Traditional categories—catalogs, governance, BI tools, databases—remain critical under the hood, but their boundaries are eroding. Data management is becoming declarative: less about the tools themselves and more about the business impact they enable. The question for data leaders is no longer “what framework do we deploy?” but rather: “what problem are we solving?” and “what value does it deliver?”

This convergence of economic pressure and technological possibility is the catalyst for Alation’s transformation. It’s also the strategic imperative for every data team: to reorient from managing data as an end in itself to building data products that deliver measurable business outcomes.

The age of data products: Why purpose-built data wins over volume

The shift from cataloging to value creation requires a fundamental reordering of how data teams work. Instead of starting with data and hoping it leads to outcomes, successful organizations now start with outcomes and work backwards.

This is where data products come in.

"Data products give you a container for value," Sangani explained. "They give you a purpose—a reason to do the work of data management. And that purpose is everything because that purpose defines how deep you go, how far you go, how wide you go."

The contrast with traditional cataloging is stark. In the old model, teams would connect dozens or hundreds of data sources, curate metadata, measure completion percentages, and hope that widespread adoption would eventually lead to business value. The problem? You could catalog forever without ever reaching a meaningful outcome.

Data products flip this script entirely. They force a simple but powerful question: What business problem are we solving? For a media empire and a bottling behemoth, data products offered a path to solve real problems with data that had plagued their businesses for decades. 

BBC delivers a single source of truth with data products

The British Broadcasting Corporation (BBC) faced a challenge familiar to many large enterprises: dozens of different reports measuring the same metrics, creating confusion and distrust. Rather than attempting to catalog everything, they built data products around their most critical business metrics.

The result? They consolidated 500 users onto a unified platform—not by measuring catalog searches or adoption rates, but by delivering a product that solved a specific business problem.

Swire Coca-Cola’s single data product signals success

For bottling and distribution company Swire Coca-Cola, business success hinges on a single measure: On Time In Full (OTIF). If an order arrives late or incomplete, they've failed. It's that binary.

The company had 60 different reports sourcing from 60 different data sources, all attempting to measure OTIF—and predictably, none of them agreed. By building a data product specifically for this metric, they created a single source of truth that everyone could trust.

"They weren't trying to boil the ocean with all their supply chain data," Sangani noted. "They basically said if we got this one metric right, this one critical metric that impacts our business, we can work back to the thing that impacts our business the most."

The data product approach allowed them to ask the right questions: What governance do we need? Where should we measure quality? How much lineage is required for trust? What access controls protect our IP while ensuring availability? By focusing on one high-value outcome, they could design the entire data management process to serve that purpose.

JLL: 4x Faster proposal generation with AI agents

Perhaps the most striking example of where this is all heading comes from Jones Lang LaSalle (JLL), the global commercial real estate giant headquartered in Chicago. Their use case demonstrates how data products can become the foundation for truly transformative AI agents.

JLL's commercial real estate agents regularly create proposals for large corporate clients like Meta, Procter & Gamble, and Kroger. Each proposal requires synthesizing vast amounts of information: property location data, comparable market prices, energy characteristics, climate improvement costs, lease histories, and client preferences. Previously, this process could take 20 hours of manual work or more.

JLL built AI agents that automatically generate these proposals by drawing on comprehensive data products containing market intelligence, climate data, client behavior patterns, and historical lease information. The results speak for themselves:

• 4x faster proposal creation

• 18% fewer vacancies due to more competitive, data-driven proposals

• 70% reduction in manual work, freeing agents to focus on relationship building rather than document creation

"This is the art of what is entirely possible," Sangani emphasized. "With AI, humans can be so much more effective, so much more thoughtful, so much more impactful with the work that they otherwise would have done."

But here's the critical insight: JLL's AI agents only work because they're built on trusted, contextualized data products. Without that foundation, the agents would be unreliable at best, dangerous at worst.

How data teams can evolve: From catalogers to agent builders

The implications for data teams are profound. As Sangani put it bluntly: "Data teams will become agent builders."

This isn't happening overnight, but the trajectory is clear:

Data stewards → Curation agent managers

Tomorrow's data stewards will spend less time curating metadata and more time managing agents—monitoring their performance, tuning their behavior, adjusting their context, and determining what they should and shouldn't do.

Data analysts → Data product and agent builders

Analysts will think less about one-off reports and more about reusable data products that solve specific business problems. They'll also build agents, because they're closest to the business processes that agents need to understand.

Operations leaders → Agent managers and architects

Operations leaders will design agentic workflows, mapping business processes and identifying where AI can augment or replace manual work.

"While I don't think this transformation is going to happen overnight, I do think this change is inevitable," Sangani told the audience. "You might go faster in marketing, you may go much slower in safety operations, but you will go in this direction."

The critical requirement for this evolution? Understanding business processes at a granular level. Every data team attempting to implement AI agents quickly realizes that the only thing that matters is mapping the current state: What happens in step one? Step two? Where do people source data? What confirmation is required? What's the documented process versus the actual process?

Only after understanding the current state can teams redesign workflows with AI, identify data requirements, build the necessary data products, and implement appropriate governance.

This represents a fundamental reordering of work: starting with outcomes, then business processes, then data requirements, then data products, and finally governance policies. It's the opposite of the traditional approach that began with policies, procedures, and frameworks.

Ultimately, precision is the linchpin. As Sangani emphasized, for AI to truly work in the enterprise, it must be able to interact directly with structured databases—reading from them with accuracy, writing back with integrity, and doing so in a way that is trustworthy at scale. If AI is processing transactions, every entry must be correct. If it’s onboarding new customers, every field must be accurate. This isn’t just an efficiency question; it’s a precision question. And it’s why a trusted knowledge layer—anchored in governance, data products, and context via metadata—is essential for the next era of data management.

The knowledge layer for data and AI: Alation's vision

Alation's evolution reflects this broader industry shift. By providing an open knowledge layer for both data and AI, Alation delivers a platform that enables organizations to:

1. Build trusted, contextualized data through enhanced cataloging and governance capabilities

2. Create high-value data products using new marketplace and product builder tools

3. Develop precision AI agents that can reliably interact with structured enterprise data

4. Maintain sovereignty across any platform, avoiding lock-in to specific compute, model, or storage vendors

"We are effectively becoming an application development platform," Sangani explained. "Almost all software is going to become agentic. As you build software, as you build capability, you are going to be leveraging, in five to ten years, agents in almost every business process you build."

This is why Alation is now selling not just to data teams, but to AI and engineering teams—a shift that wouldn't have made sense 18 months ago but is essential today.

The company's investment strategy reflects this vision:

• Core cataloging and governance remain the foundation, with continued enhancements to curation, lineage, and search capabilities

• Data products enable AI to work atop structured enterprise data, with new tools for discovery, marketplace functionality, and product building in the roadmap

• Precision agents represent the cutting edge—enabling customers to move from providing insights to fundamentally transforming their businesses

• Openness and portability ensure enterprises avoid vendor lock-in. By owning their data and their AI, organizations can adapt and evolve with the market and, ultimately, control their own destinies.

Importantly, Alation recognizes that customers are at different stages of this journey. Some need to solidify their governance foundation. Others are ready to build data products. A few are prepared to pioneer agentic workflows. The platform is designed to meet organizations wherever they are—and help them move forward at their own pace.

The future is here—but not evenly distributed

Sangani closed with a quote he loves: "The future is here, it's just not evenly distributed."

revAlation Chicago offered a sneak peak of that future—from customers still building their catalog foundation to those deploying sophisticated AI agents. The message is clear: the transformation from data cataloging to agentic AI is underway, but every organization will navigate it at their own pace.

For data leaders, the path forward requires:

• Starting with outcomes rather than data sources

• Building data products that contain measurable value

• Mapping business processes before implementing AI

• Investing in the knowledge layer that makes agents trustworthy

"We want to bring you along for the journey," Sangani said. "For those of you that are super brave and super adventurous, we're here for you. And for those looking for really much more conservative capability, we want to be able to give you that too."

Thirteen years after founding Alation to empower a curious and rational world, Sangani is refounding the company for the age of AI. The vision remains the same—enabling people to answer questions and take action with data. But the tools, the processes, and the possibilities have fundamentally changed.

The data catalog was just the beginning. The real work—building the knowledge layer for agentic AI—is happening now.

Curious to see for yourself? Book a demo with us today.