We want to help customers get value from AI without losing control of their data, says Ali Elkortobi, CEO of infoCorvus

Andorra could become one of the growth hubs for a new technological era, where data and artificial intelligence are not managed in a chaotic way but instead are governed by strict oversight and clear rules. This view comes from a veteran of the global data management industry, Ali Elkortobi, who has spent over 40 years working in the Swiss banking industry, at Oracle in the United States, and on major international IT projects. According to him, the main challenge facing large companies today is not a lack of data, but rather its excess and disorder: information is fragmented, duplicated, and often poorly protected. This issue is becoming especially critical amid the rapid development of artificial intelligence. The expert believes that the future lies in an approach where data is first organized, cleaned, and secured, and only then used for AI systems. In addition, infoCorvus is considering opening a subsidiary in Andorra, with the aim of joining the private AI hub coordinated by the Andorran company ASC/Aitek, which is currently being established in the Principality, and of contributing to its visibility as a European showcase of digital sovereignty and sovereign AI. This initiative confirms Andorra’s growing attractiveness for technology players seeking an environment that combines innovation, security, and sovereignty. What in your own career made you feel that enterprise data management was still too fragmented and that the market needed a more unified solution? I have worked in data management for probably 40 years. I started in banking, worked on data modelling for banks, built banking applications in Switzerland. Then I joined Oracle, working on data migration, implementing databases and other challenging implementations. Oracle recognized that I was doing something special and asked me to go to the U.S., so I went to California. There, I headed engineering for data warehousing and ETL (transformation, loading, and extracting data)  for many years. Then I joined a new division around data archiving. I managed engineering for archiving databases, unstructured data, email, basically everything about archiving. This expanded my view on data profiling, discovery, retention, disposition, legal hold… I mean all aspects of data lifecycle management. After HP split and their software division was sold to MicroFocus, and later MicroFocus was acquired by OpenText, I was part of the executives who were let go. I then had offers to return to Oracle, but I also had the chance to join a former colleague who wanted to transform his services company into a product company. We decided to create a new company – infoCorvus. We found investors and built a system highly scalable, distributed, capable of handling high volumes of data with strong performance. Founders often build companies not only because they see an opportunity, but also because they’ve seen the same problems again and again. Was there a specific client challenge, failed transformation, or recurring gap in the market that pushed you to create infoCorvus? I’ve seen recurring problems for banks, insurance companies, and telecoms. I worked with big software vendors and had access to large enterprises. The problem is always how they do data integration – moving, transforming, and publishing data. Data never stays in one place; it’s always moved, transformed, and published to new user populations, either internally or externally. It is crucial to protect data, publish it to the right people, and understand what data you have. We built a data discovery module to give insights — what data exists, what it contains, and how it should be protected. We also included data augmentation, creating derived data for analysis. The goal of the platform is to ensure proper data manipulation, enhance data quality, and enable AI on governed data. ROAD is presented as a single control plane. What was the original strategic idea behind bringing all of those capabilities together instead of treating them as separate software categories? The strategic idea was actually very simple: data should be managed as one lifecycle, not as disconnected events. In most enterprises, governance is treated separately from migration. Migration is treated separately from ingestion. Ingestion is treated separately from archiving or application retirement. But in reality, all of those are part of the same operational continuum. When a company modernises its architecture, moves to cloud, retires a legacy application, prepares data for analytics, or exposes data to AI, it is dealing with the same underlying issue: how to control, move, understand, secure, and operationalise data. That’s why we built ROAD as a single platform rather than a collection of point solutions. And we also wanted to make that available not only to the largest enterprises, but to the broader mid-market and upper enterprise layer as well. So, the original strategic insight was this: if data is one enterprise reality, then the control layer around it should also be one. A lot of the AI market still talks first about speed and acceleration, and only later about governance and control. Your message seems to reverse that logic. Why do you believe governed data has to come first if companies want to make AI truly useful? Many companies are attracted to AI and want to extract value quickly. They hire developers and give them access to build AI applications, but often ignore security and governance. We believe that even if AI is accessing data, company policies and security must come first – who can access what, and who can see what. We built a Governed AI Bridge that treats AI like a human user: with roles, privileges, and controlled access. AI can access the data it’s allowed to, and we dynamically protect sensitive data while serving it. Your Governed AI Bridge is especially interesting because it seems to sit between enterprise data control and AI execution. Do you see it mainly as a protective layer – or as the beginning of a new architecture for trustworthy enterprise AI? I see it as both – but more importantly, I see it as the beginning of a new architecture. At one level, yes, it is a protective layer. It ensures that when AI accesses enterprise data, it does so according to the same permissions, roles, and controls that would apply to a human user. But it is more than just a security wrapper. What we have really built is a governed interaction layer between AI and enterprise data. That means AI does not need to know whether the data is coming from. ROAD can serve that data dynamically, just in time, and in a controlled way. At the same time, we can apply dynamic protections, for example, masking specific fields if AI is not supposed to see them. So yes, it protects. But more importantly, it creates a new enterprise pattern: AI should interact with data through a governed mediation layer, not through uncontrolled direct access. I think that is going to become a very important architectural principle for trustworthy enterprise AI. Under this new data dictionary, is it possible to tokenise it? Yes. ROAD can act as an MCP server with a semantic layer around the data. You can add your own semantic layer, enrich metadata, and provide knowledge about the data. This allows us to tokenise data if needed. The platform serves metadata and knowledge to LLMs or agentic AI, helping AI find and use data correctly. So, tokenisation is fully supported through this semantic and metadata layer. Andorra is proposing an initiative to create a technology hub focused on sovereign cloud, sovereign conversational AI, agentic no-code application development, and operational intelligence. How important is data readiness before launching such a platform, and how could you contribute to this effort? ROAD can prepare, clean, and serve data to these systems. It can enrich metadata, apply semantic layers, and deliver knowledge in real time to LLMs or agentic AI. Essentially, ROAD provides the governed, prepared data foundation, while the Andorran company ASC / Aitek can use that data for sovereign applications, agents, workflows, and operational intelligence. ROAD helps enterprises clean, govern, migrate, and safely expose their data, while a platform like ASC / Aitek could sit higher in the value chain – turning that governed data into sovereign applications, agents, workflows, and business action. In highly regulated sectors – finance, government, healthcare, and critical infrastructure- do you believe the next competitive advantage will come not from AI alone, but from the combination of strong data governance and sovereign execution environments? That’s a really great question. And yes, I do believe that. I think a lot of companies are going to make mistakes first. They’re very attracted by AI, and they want to move quickly. But they’re not always thinking about security, governance and sovereignty And they’ll likely realise that only after problems start appearing. The truth is: we’ve already dealt with these issues before. In business intelligence, data lakes, data warehouses, and data marts, we’ve been solving the same governance questions for more than 30 years. AI is powerful doesn’t mean we should forget all of that. We need to marry the two: the power of AI and the discipline of governance and security. We want to help customers get value from AI without losing control of their data. From your perspective, what do large enterprises still underestimate the most: the burden of legacy systems, the regulatory risk of exposing data to AI too early, or simply the fact that most AI initiatives fail because the underlying data architecture was never truly ready? If you walk into almost any data centre and look at the data landscape, you’ll find that more than 50% of the data is dormant. It often comes from legacy systems, historical systems, accumulated applications or duplicated processes. And there is usually a lot of redundancy, inconsistency and poor visibility One of the key pillars inside ROAD is the ability to deal with all of the data an organisation has. The idea is to look at data as a whole, to classify it and to help companies finally understand and manage all the dormant data they’ve been carrying for years. This is especially important because many companies are now moving from on-premise environments to the cloud. But simply doing a “lift and shift” is often a bad idea. Because the garbage you had on-premise becomes the garbage you now pay for in the cloud. And in many cases, it becomes even more expensive. A lot of companies don’t fully realise that over time, cloud costs can become significantly higher. Yes, you move from CapEx to OpEx, but if you look at the long-term operational cost over 3, 4, or 5 years, you may actually end up spending much more. So, the real question is: “Do you really want terabytes and terabytes of data in the cloud?”. That can become a very expensive mistake, sometimes even a business-critical one. That’s why we focus on helping customers to understand what they have, minimise unnecessary data, bring order to the environment and optimise the full lifecycle of data. Because ultimately data is value. Data is currency. That’s how we see it. infoCorvus is planning a technological partnership with ASC, an Andorran platform for sovereign cloud and artificial intelligence. This collaboration would be based on clear complementarity: on one hand, infoCorvus would handle data governance, migration, and secure exposure; on the other hand, ASC would transform this data into sovereign applications, intelligent agents, business workflows, and operational intelligence. If you were to design an ideal partnership model, how would you allocate responsibilities? This is a very strong partnership model because it avoids overlap and creates a very clean architecture. I think that kind of architecture will become increasingly important as enterprises move from “AI experimentation” to real operational deployment. AI does not become enterprise-ready when the model gets smarter. It becomes enterprise-ready when the data underneath it becomes governed, secure, understandable, and usable.The post We want to help customers get value from AI without losing control of their data, says Ali Elkortobi, CEO of infoCorvus first appeared on All PYRENEES.