Reimagining Data Management: Tom Davenport on AI, Innovation, and Business Transformation

Tom Davenport is a distinguished thought leader in data, analytics, and artificial intelligence, shaping modern business thinking through more than 20 influential books and numerous articles. 

Davenport recently joined Alation's Satyen Sangani on the Data Radicals podcast, where he predicted that agentic AI would emerge as the defining AI trend of 2025. But what exactly is agentic AI, and how should data management professionals and Chief Data Officers (CDOs) prepare for this transformative shift? Read on to find out.

Agentic AI: Hype vs. reality

Agentic AI refers to AI systems capable of autonomously performing tasks beyond simple predictive functions—essentially, digital assistants capable of acting independently within defined parameters. Davenport acknowledges its limitations, noting the early-stage reality where current agentic AI tools are primarily suited for straightforward tasks. 

As Davenport mentioned during the interview, "Alexa is an agent created by Amazon. It could do things like turn on your lights or set your alarm. The real challenge is being good at both doing things and knowing things."

What is agentic AI, and why is it important?

At its core, agentic AI autonomously handles specific tasks or workflows, managing decisions with minimal human intervention. Davenport underscores its current practical limitations, pointing out simple applications like managing employee PTO requests. 

Yet, he stresses its transformative potential, particularly in sectors like healthcare, where AI could profoundly impact frontline delivery rather than remaining isolated in research labs. "Enormous potential AI in healthcare is found largely in the research lab and not at the bedside," he emphasizes, highlighting the need for deployment at the frontline.

More advanced examples illustrate how impactful agentic AI can be. Davenport describes a Toronto-based company he's worked with extensively: "They're very document-oriented, particularly financial documents and documents of record for various companies. They've been using agentic AI quite successfully to extract information from all types of documents, effectively turning documents into reusable data. It’s certainly useful and transformative."

Could AI replace software?

Despite bold predictions, Davenport asserts AI will complement rather than entirely replace traditional software: "AI may write a lot of the software, but it's a bit much to say it's going to somehow replace software. After all, AI itself is software." 

He adds historical context, saying: "We need to remember this [AI] has been around in some form for a while. Alexa from Amazon is an early example of an agent.” Yet the challenges of making these agents truly intelligent and context-aware remain significant.

From unstructured data to knowledge management

Generative AI has sparked renewed interest in knowledge management, especially in managing unstructured data. Davenport explains that generative AI simplifies turning unstructured sources into structured information. However, he stresses the ongoing necessity for human oversight and validation, given the inherent contextual judgments involved.

In the past, “It was very labor intensive to sort of figure out what was important about a piece of knowledge and capture it and make it accessible,” he points out. “Generative AI is quite good at it, but your data have to be in pretty good shape… Gen AI may help a little bit and give you an opinion, but humans certainly need to validate it.”

Challenges in building data-driven cultures

Though generative AI briefly revived enthusiasm for data-driven cultures, Davenport warns against temporary excitement. Citing recent surveys, he notes a surge and subsequent decline in interest, highlighting the persistent challenge of creating sustainable data-driven organizations. 

“It's not rocket science,” he opines. “It's having senior people who are interested in analytical, decision making, hiring people who can do the work: the day-to-day work of analytics, both the data management and the data analysis.”

The evolving role of the Chief Data Officer (CDO)

The rise of AI has reignited discussions about the role of the Chief Data Officer. Davenport notes that having too many specialized technology roles creates confusion. As he explains, "I was very surprised to learn that senior tech people think there are too many of these C-level tech roles, causing confusion in organizations. When you have a CIO, Chief Data Officer, Chief Digital Officer, we don't even have a unique acronym for digital and data officers." 

He argues organizations benefit from consolidating these roles under "a single, strong, business-oriented executive" overseeing data, analytics, digital, and technology functions. Davenport adds, "There are already examples of this happening, particularly in financial services, where executives manage not only tech functions but also customer service and operations, becoming integral parts of senior management teams."

Mike James, SVP and Head of Data & Analytics at the NBA echoed the need for business leaders who can “speak data” in his own interview when he opined, “Data and analytics works best when business leaders who are about to make very consequential business decisions have been highly trained in how to create hypotheses that are testable and then can give those hypotheses to people who are trained in leveraging information.” In other words, magic happens when leaders who have grown up in the business can translate intuition into data hypotheses that they can road-test with support from a data team. This level of cross-functional collaboration becomes even more important with the rise of AI. 

Defining and managing data products

The term "data product" often generates confusion. Davenport defines data products as reusable, structured, and well-documented data assets designed to deliver repeatable business value. Data product managers play a pivotal role in ensuring data becomes reusable, impactful, and aligned with business objectives, shifting from merely creating data to enabling active reuse and measurable impact.

Confusion about the term persists. “I did a survey [of senior IT people] and about half of the respondents…thought a data product was just the data, and the other half thought it was the data plus the analytics or the AI,” he explains. “So it's just confusing. I think or has been in the past.” For terms seeking to launch and leverage data products, aligning on an internal definition is critical. 

The AI hype cycle and the path ahead

According to Davenport, generative AI is moving beyond Gartner’s "peak of inflated expectations" toward market consolidation. 

“There's still a lot of debates about this,” he points out. “Within Goldman Sachs, you have some analysts saying, ‘Oh no, it's going to cause massive amounts of productivity gain and economic growth,’ and others are saying, ‘No way is that gonna happen.’ They can't even agree within one firm! And you have economists who can't agree at all… It seems to me that things are getting a bit inflated.

For these reasons, Davenport foresees a tempered economic impact as organizations grapple with the practical complexities of integrating AI. This perspective underscores the need for realistic expectations and strategic implementation.

Conclusion: The future of AI and enterprise readiness

Davenport’s insights offer a critical lesson: successful AI adoption demands practical implementation, thoughtful integration, consistent human oversight, and deliberate process redesign. Rather than rapidly replacing traditional systems, AI will integrate through gradual evolution and intentional organizational transformation.

Curious to learn how a data catalog can help you redesign processes for AI? Book a demo to learn more.