The path to AI success isn’t about having vast data alone — it’s about how well enterprises manage, understand, and innovate with that data. The recent Evanta Chief Data and Analytics Officers (CDAO) Summit in London gathered enterprise leaders across industries to confront a paradox: while AI demands boundless access to data, it equally insists on governance, security, and compliance.

My time at the summit underscored a sharp reality: Building effective AI capabilities isn’t just about technical prowess. It requires an evolution in how data is approached — from handling unstructured data, which defines modern integration challenges, to managing regulatory burdens and tensions between accessibility and control. Here’s a closer look at where global enterprises are succeeding and struggling, and what’s next in the journey toward scalable, ethical AI.

Data maturity isn’t linear, and that’s the point

At Fivetran, we advocate a model known as the data maturity curve, helping organizations assess their progress along stages from reactive to proactive and, eventually, to innovative. But what became clear at the summit is that most organizations today exist at multiple stages simultaneously. While basic analytics and structured data reporting are now table stakes, unstructured data presents an ongoing challenge to reach higher maturity. This is because unstructured data governance — essential for compliance and innovation — is still in its infancy.

One CTO of a leading industrial enterprise illustrated this perfectly: While his team achieved real-time insights from structured data, their unstructured data — including 7.1PB of legal and customer files — remained siloed, largely untouched. Another leader from a large financial institution expressed frustration over the simultaneous abundance and inaccessibility of over 40 years of detailed historical market data. Across industries, this pattern emerges – the untapped potential of unstructured data to unlock predictive insights and fuel AI-driven innovation.

Regulatory realities and the global divide

While organizations of all kinds struggle with unstructured data, regulation does not impose the same obligations and potential barriers to innovation everywhere. With the advent of the EU AI Act, executives in Europe reported spending between 50% and 80% of their time on governance. In comparison, one Asia-based company reported aggressively pursuing a “move fast” approach, prioritizing customer feedback and market relevance over rigorous regulatory compliance. The divergent approaches underscore a fundamental truth: AI’s future will be defined by not only what is technically feasible but in what is legally and ethically permissible.

For many global enterprises, this contrast drives a renewed need for frameworks that balance compliance with agile innovation. Fivetran's role in this equation is twofold: centralizing data to provide a single source of truth, while ensuring that data access, usage, and governance align across diverse regulatory landscapes. This is where automation and advanced metadata management come into play — automating repetitive governance tasks so organizations can focus on strategic, impactful initiatives.

Data access and ownership: The new battleground for AI 

As organizations build out AI capabilities, tensions over data access and ownership are growing. On one side, product teams are racing to leverage AI to gain a competitive edge, pushing for unfettered access to data. On the other, data governance leaders and security teams are tasked with safeguarding sensitive information, duties that are mission-critical to organizational security but often perceived as an impediment to innovation.

One proposed solution discussed at the summit was enhanced metadata tagging at the document level — essentially treating metadata as a kind of “permissioning passport.” By tagging data with information about its type, origin and permissible use cases, organizations can streamline access for product teams while maintaining a strong governance stance. The future, many attendees agreed, lies in developing sophisticated data catalogs where metadata isn’t just an afterthought but a central organizing principle.

For Fivetran, this means offering solutions that provide centralized metadata and visibility across data sources, making it easier to enforce policies, ensure compliance and speed time-to-insight. When metadata is thoughtfully managed, it eliminates redundancy and cuts down the cost and risk associated with sprawling, disjointed data ecosystems. 

AI needs data literacy, not just data scientists

While discussions around talent gaps in AI often focus on the shortage of skilled data scientists, the truth is that data literacy across functions is equally critical. Without a baseline understanding of data’s potential — and limitations — many organizations face the risk of unrealistic expectations or, worse, misguided AI initiatives that fail to deliver meaningful business outcomes.

At Fivetran, we believe that every team member working with data needs a level of literacy to enable sound decision-making. This includes not just technical skills but also ethical awareness, critical for maintaining trust in AI models. The executives at the summit echoed this need, highlighting that educating teams in data stewardship is as critical as investing in data scientists.

Successful AI isn’t just about access to more data or regulatory compliance; it’s about fostering a culture of responsible data stewardship. For global enterprises, the imperative is to balance innovation with accountability. This means advancing not just the technical sophistication of data ecosystems but also the maturity of data governance and literacy throughout the organization.

Fivetran’s mission aligns closely with this vision: to empower organizations to scale data responsibly, with tools that centralize data, automate governance and facilitate collaboration between product and IT teams. By enabling enterprises to effectively govern data and optimize unstructured data sources, Fivetran is helping organizations worldwide to harness the potential of AI sustainably, securely, and with confidence.

[CTA_MODULE]