”Always On” Agentic AI: The Future of Data Governance?

by Nigel Turner, Principal Consultant, EMEA

I have advised, designed and helped many organisations to implement data governance for the last fifteen years. From its roots in the early 2000s this critical data discipline has increasingly become a “must have” capability for all organisations who aim to be data-driven and who are trying to navigate the exploding scope, scale and complexity of their data as the digital revolution picks up speed.  

Data governance has matured since its embryonic beginnings, and a proven body of best practices has evolved, encompassing topics including:

  • how to get the business to recognise data’s importance and play a leading role in its improvement
  • what data should be rigorously governed and why
  • what new roles and responsibilities are needed to implement data accountability
  • what processes and tools are required to support governance activities
  • and so on.

It at last feels as if data governance has finally come of age and that its future is assured.  

But a recent event I was part of has made me come to question that belief. A few weeks ago, I helped to organise and attended a Data Management Association (DAMA) event in Dublin, Ireland. This event, jointly curated by DAMA Ireland and DAMA UK, focused on the future of data management and the increasing influence of AI.  The keynote was delivered by Mike Ferguson, CEO of Intelligent Business Strategies. Mike’s session, and others on the evening, stimulated a lot of discussion and debate, both at the event itself, and afterwards in a very amenable Dublin pub that many attendees gravitated to. 

Most conversations of the evening focused on the future of data governance. Mike and other speakers contended that the advent of AI (and specifically agentic AI) is about to disrupt significantly the prevailing data governance paradigm described above. Given the fast pace of AI evolution, this cannot be dismissed as something to think about in the longer term. In contrast, agentic AI can and will pose an imminent challenge to data governance specialists and the roles they carry out today. If you think that’s an exaggeration, I recall that around five years ago the “hot” data job was that of a data scientist. With AI’s advent, I suspect that many data scientists are already nervously looking over their shoulders, worried about their current and future employment prospects. Some have probably already been supplanted by the technology.

So how will agentic AI transform and revolutionise data governance? Mainly by automating many of the roles and responsibilities currently undertaken by data owners and data stewards in most organisations today. This automation is radical in that it’s not just about applying agentic AI as a potential governance support tool, e.g. to help data owners and stewards identify implicit data quality rules or to track who is accessing critical data, but is a necessary and inevitable response to the changing demands of governing “always on”, fast-paced, complex and highly distributed data.  Managing an ever more fluid and dynamic data environment is already a major challenge for data governance today. AI can help solve this problem, but more than this can offer a superior, automated and more proactive means of governing data by largely removing human data stewards (in particular) out of the loop. Instead, agentic AI will become the primary data steward, operating 24/7 to ensure governance is perpetual and proactive.

Agentic AI governance can realise this in several ways. First, by enforcing the policies that an organisation has determined for the data it intends to govern. These policies would include data access, security, privacy, retention and quality. For example, a data quality rule can be fed into AI, and the tool will then apply that policy to all data instances as and when it encounters them. Contraventions can be flagged, or even corrected, by the tool depending on the levels of confidence placed in it. In any case AI will learn and so evolve to handle exceptions more proactively. To work, all this must be supported by an integrated technology infrastructure including data catalogues, data quality rules engines etc. – all essential components of emerging data fabric architectures and platforms.  

It must be stressed that this is not necessarily all bad news for data governance professionals. As you’ve probably already concluded, a human data owner or data steward will need to provide (at least initially) the policies and rules needed to train and operationalise AI. Moreover, oversight of the automated governance environment will continue to be essential to ensure AI is functioning as expected, particularly when event alerts are generated or remedial action taken. Most important, though, effective automated governance relies on the very thing it seeks to enhance – data. If current data and associated metadata is not fit for AI purposes, any attempt to automate will be hampered by data gaps, misunderstandings, inconsistencies and quality issues. A recent survey conducted by Harvard Business Review Analytic Services found that only 7% of organisations’ data is “completely ready” for AI adoption, with a further 27% stating their data is “not very, or at all ready.”[1] Many organisations do not as yet have the data management maturity to implement an “always on” governance environment.

But it does mean that data governance and AI professionals must work more closely together if this future state is to be achieved and ensure the path to eventual governance automation is understood and mapped out. The current wall between data and AI specialists in many companies was another topic highlighted many times during the event, with some expressing their frustration at the lack of mutual understanding of the importance of these disciplines working in synergy and not as separate endeavours.

I suggest several potential specific actions that can help to start to deliver an automated data governance infrastructure and help bridge the current data / AI gulf:

  • Within their organisations, data governance people should make the business case for governance to be a high priority when prioritising agentic AI use cases, e.g. a use case to monitor changes to critical data elements (CDEs) across key platforms and systems.
  • Governance and AI people need to liaise closely to ensure that the potential of agentic AI tools in data governance are jointly understood and requirements captured.
  • Larger data governance teams could consider recruiting an AI specialist role whose primary job is to identify and apply agentic AI to start to automate current governance activities.
  • Reconsider data governance data owner and data steward job descriptions and objectives to encourage role holders to embrace AI and additionally provide them with the relevant training and enhanced skills to explore potential agentic AI use in supporting their roles.

What is without question is that using agentic AI to enhance and improve data governance is already starting to happen and will accelerate over the next few years.  The logic behind “always on” data governance is inescapable and so it’s odds on that it is going to happen, like it or not. But how to thrive in this coming revolution? All current data governance professionals need to prepare themselves, upskilling to increase their familiarity with and potential adoption of agentic AI. By doing this they will help to determine their own future prospects, rather than be passive victims of it. As Dr. Seuss once said, “Only you can control your future”.

[1] Cited in ‘Only 7% of Enterprises Say Their Data is Completely Ready for AI…”, LinkedIn article, 9 July 2026.