Managing Your Data Estate – Putting AI To Work Instead of Play

Town Hall Insights
Minneapolis CDAO Community

Erik Zwiefel

Chief Data and Analytics Officer, Americas Data & AI

Microsoft Corporation


Kapil Narang

Chief Data and Analytics Officer

Ameriprise Financial


Tim Deutsch

Chief Data Officer

Sentara Healthcare


Luke Salzwedel

Senior Director, Analytic Innovation & Strategy

Blue Cross Blue Shield of Minnesota


Angie Schulke

Vice President - Applications, Data & Business Intelligence

Life Time

MARCH 2024

When you transform an organization, your goal is to end up with connected, informed, intelligent, and responsive data. Yet data silos, disconnected systems, and incongruent data can keep you from this organizational nirvana. Combine those challenges with how much focus there is now on GenAI, and organizations have an even tougher hill to climb. 

The question becomes this: Regardless of what business unit, system, or cloud your data resides within, how do you use the hype and capabilities around AI to build a broader data estate? At a recent Town Hall discussion in Minneapolis, CDAOs came together to discuss how to create the most effective ecosystem, adopt new GenAI capabilities to maximize data, analytics, and AI investments, and simplify leveraging data fabric and AI.

Erik Zwiefel, Chief Data and Analytics Officer of Americas Data & AI at Microsoft, started the discussion by talking about how significantly AI is changing the world. He spoke about emerging use cases and capabilities we’ve never seen before, such as the ability to reason, and referred to Microsoft research about GPT-4 showing beginning signs of intelligence.

He said organizations are scrambling to figure out where AI fits in their data estate and how to leverage it. He talked about how AI is poised to drive many changes but is only as good as the data you give it. He pointed out that organizations are chipping away at “low-hanging fruit”—or where to start with AI—but also need to consider where to go from there and how else to use it. That, he explained, is where we run into how immensely complex our modern data and analytics stack are. The result of us having manually integrated different systems over the years is that our data and applications are segmented, fragmented, and in silos. It’s all optimized for personal use, not for broader organizational use and not for AI.

He said that leveraging AI now can be a huge risk if we haven’t done a good job with data governance. He referred to an IDC survey stating a majority of employees can access data sets they shouldn’t. As AI starts crawling, he said, that access will be discovered. 

To create a modern data and analytics layer that’s consistent, he said, Microsoft settled on bringing Microsoft data to a centralized data lake. 

Existing and emerging AI capabilities can help refine that data for optimization, and master data management/large language models can be a huge help. He gave an example of using AI to create the first draft of a data catalog, which a person could then edit.

He called data “the fuel that powers AI,” adding that AI can then help manage that data. Zwiefel said his dream is to build a cycle that brings data to a new level and increases our ability to drive value for business.

Key Takeaways from the Discussion

  1. How data infrastructure and governance policies support AI

Organizations are all in the same boat, said one CDAO, with the same struggles regarding bringing AI on board. One executive agreed, saying data is the foundation, and likely not ready yet. “Garbage in, garbage out. There’s a huge spotlight on data, which is showing where the problems are, but ‘sunlight is the best disinfectant.’” 

Organizations need to invest in AI’s supporting foundational needs, such as security and data security, said one CDAO. Another executive brought up governance as a big issue, saying existing models and policies need updating for AI capabilities. One CDAO agreed, saying the lack of investment and maturity in data governance impacts AI and especially those asking for AI, who will become disconnected from where the underlying, existing issues are regarding data and data quality.

  1. Maximizing the ROI of data, analytics, and AI investment

CDAO executives agreed that it’s still early days regarding best practices. “There are lots of research trials and proof of concept work going on in the area of AI before enabling it in the enterprise,” said one. Another participant agreed, saying they are still thinking more about pilot projects and less about ROI from using AI. They added that the focus is more on internal productivity and experience, customer experience, and not yet on revenue. One CDAO said it’s important to focus on data governance, modernizations, and data platforms before strategizing about value. 

Focusing on what’s essential to the organization is how to get ROI, said one CDAO, and another noted it’s important to know what big ideas you’re going after and focus on those use cases for AI. Executives agreed they need to know what solutions in the market are truly mature and which are best for what use cases. “What’s real, and how can we use it in the most cost-effective manner?” asked one. 

This type of prioritization is important, agreed another, saying there are hundreds of ideas about how to use AI and GenAI to help the business, and they’re clearly focused on optimizing cost—but there are also revenue generation opportunities, and that perhaps that’s the next step. Some agreed that the answer is to continue doing proof of concepts (POCs) to make the problems visible so they see the total cost of success.

  1. How do you determine the best ways to incorporate AI?

Leadership doesn't yet understand the needs of AI, said one CDAO—they just say, ‘Go do it and come back to me when it’s done.’ One executive asked, “How do we prioritize the hundreds of ideas coming in from the business? What technology is going to be the best to enable and most cost-effective? Also, what’s best in governing front-end access to data in dashboards vs. the raw data under the hood?” The discussion turned to healthcare as an example, with the question of how AI impacts patients, doctors, and nurses and how they can interact while still maintaining a patient’s data and privacy.

Overall, CDAOs agreed that organizations are still figuring out how to incorporate AI and GenAI and how their foundational infrastructure and policies need to be updated to support the new technologies. There’s currently more emphasis on pilot projects than full AI usage and ROI. 

CDAOs can continue the discussion on data estates and AI, and other topics, at an upcoming gathering, which you can find on our calendar. If you are already an Evanta community member, you can register in MyEvanta. Or, you can apply to join a community of your CDAO peers to stay fully up-to-date on key topics for data and analytics leaders.

by CDAOs, for CDAOs

Join the conversation with peers in your local CDAO community.