Joel Acha has spent more than 25 years in the analytics world and knows the space inside out. These days he’s a Lead Oracle Data Architect at Capgemini, an Oracle ACE, a regular contributor to the Oracle Analytics and AI Community, and he frequently blogs about analytics and AI. As he puts it, “I enjoy sharing practical architectural guidance, challenging misconceptions, and helping teams avoid problems they should not have to learn the hard way.”
It was a real pleasure to chat with Joel. Here’s what he had to say.
Modernizing the Data Estate
Organizations are building governed, flexible data platforms that can support traditional reporting, self-service analysis, and a growing range of AI-driven use cases. The direction of travel is clear: simplify the architecture, cut down unnecessary data movement, and make trusted data easier for the business to get to.
Joel’s currently working on a transformation program for a large UK public sector organization. “I’m leading the design and build of a lake house-style platform. Modernization is not only about newer technology, but about creating a simpler, more flexible architecture for future needs. Our aim isn’t simply to stand up another data repository, but to create a foundation that can combine structured and unstructured data, support analytics at scale, and prepare the organization for AI capabilities that were not part of the original program design.”
That’s exactly where newer AI data platforms and lake house approaches come in. They’re designed to unify capabilities and reduce complexity. Oracle’s AI Data Platform, for example, is a governed environment that brings together data, analytics, and AI services, organized through a medallion architecture that refines data from raw to business-ready. “One of the most significant advantages of this model is that not all data has to be physically consolidated in one place. Oracle AI Data Platform and open lake house approaches increasingly allow organizations to analyze distributed data while keeping governance intact. It makes trusted data easier to access across the enterprise.”
Self-service analytics, but governed
Self-service analytics remains one of the biggest promises of modern platforms. “Traditionally, business teams relied on central reporting functions to build dashboards and answer questions. By the time those requests were fulfilled, priorities had often shifted. Self-service changes that model by allowing users to explore data directly, acting on their own initiative, and respond more quickly to emerging issues.”
In reality, adoption usually starts with the power users inside business teams—people already comfortable working with data extracts, spreadsheets, or ad hoc reporting tools looking for a better way to get accurate, up-to-date insights.
But Joel is careful not to oversell it. “True self-service should not mean unmanaged data sprawl. When users work from governed data inside the platform rather than exporting extracts, organizations retain control over access, lineage, and policy enforcement. That balance between empowerment and governance is what turns self-service into a scalable operating model rather than a workaround.”
Data quality, lineage, and stewardship
The common reality in enterprise transformation is that many organizations still don’t have a formal data quality function. Responsibility often sits with data owners and stewards in the business, who define the rules that central engineering teams then enforce by engineering ingestion and transformation pipelines.
“In medallion-style architectures, that usually means applying quality checks in the silver layer and shaping trusted, reusable assets in the gold layer. This progression from raw to validated to business-ready data is a widely used pattern for improving trust and reuse across analytics and AI workloads.”
That governance foundation matters even more as organizations rush toward AI. “If the underlying data lacks quality, lineage, semantics, and context, AI will only amplify the problems.” There may be huge excitement around generative and agentic AI right now, but long-term success still comes back to getting the fundamentals right.
From shared platforms to data products
In large organizations, different teams use different analytics tools, but they still need a common access model and shared governance policies. “A well-designed lake house can act as an analytics-agnostic foundation: whatever tool sits upstream, access controls and data policies remain consistent.”
This also makes it easier to support broader, cross-functional insight. Although many business teams still work in silos, demand is growing for analytics that combine data from multiple functional areas, such as HR, finance, and operations. “The concept of a data product is useful here: business-oriented structures in the gold layer, tailored to how particular users want to consume and interpret data. The same governed source data can be shaped in different ways for different groups, creating curated views that fit distinct use cases without duplicating the underlying platform.”
In practice, that means teams can build repeatable patterns and accelerators for common industry use cases instead of starting from scratch every time.
Driving adoption and measuring value
Oracle AI Data Platform solves a clear access problem for users: it makes data easier to find, easier to trust, and easier to use. But technology alone doesn’t guarantee adoption. “The focus is on incrementally delivering value, rather than waiting for a “big bang” rollout. This means users start realizing benefits early, even as the platform continues to evolve.”
Measuring value then becomes a matter of understanding how analytics and AI are used in the business. “Tracking which groups use the platform most, which data products are adopted, and where usage is weak can provide a practical feedback loop. That insight helps teams refine products, identify unmet needs, and focus enablement where it matters most.”
Augmented analytics and AI
Over the last decade, analytics has evolved from centrally governed dashboards to self-service exploration. “The next step is the use of AI to help users generate questions, surface insights, and interpret patterns without having to start from a blank page.”
What that really means is an evolution of self-service analytics: natural language interaction, automated insight generation, and AI assistants built right into the workflow. “There is a shift away from traditional patterns in which IT teams built pre-defined dashboards for business users to consume. Instead, AI is changing insight discovery into a more conversational and interactive experience. That matters because many users are still excluded by the technical demands of traditional analytics tools. Augmented analytics lowers the barrier. Instead of requiring users to know exactly which dashboard or visualization will help them answer their question, the platform itself can propose relevant views, explain patterns, and guide exploration. That opens analytics to a wider audience than self-service alone could reach.”
It also points toward a convergence between analytics platforms and AI platforms. “As agentic AI capabilities mature, users are unlikely to care which underlying service is generating an answer, a visualization, or a recommendation. What will matter is that the experience feels unified, trustworthy, and responsive.”
Easier ways for more people to work with data
To give Joel the last word, “Enterprises still need dashboards and robust reporting, but they also need flexible data products, stronger governance, and easier ways for more people to work with data. The lake house, organized through medallion principles and enriched by AI, is emerging as the architecture that can support all three. The challenge now is less about whether these capabilities exist and more about how quickly organizations can put the right foundations in place to use them well.”
Learn More
Connect with Joel in the Oracle Analytics and AI Community to hear more about his work as an Oracle ACE and Analytics advocate.