Antonella Serine (KLA Digital), a technological promise under pressure
Artificial intelligence automates, accelerates, and optimizes processes, sometimes at a speed that exceeds human analytical capacity. But behind this promise lies a growing concern: how can we maintain control over systems capable of making decisions, sometimes without human intervention? In regulated sectors, this question is no longer theoretical. It becomes operational, legal, and strategic. It is within this critical space that Antonella Serine, founder of KLA Digital, operates. Her ambition is clear: to enable companies to use AI at scale without losing control over what it does or the ability to prove it.
A journey at the heart of complex systems
KLA Digital’s positioning does not stem from abstract intuition, but from a career built through contact with organizational complexity. At Accenture, Antonella Serine participated in structuring technology projects for major European companies. There, she discovered that a project’s success depends less on the technology itself than on its ability to integrate into existing, often constrained processes. She continued at BNP Paribas Cardiff, in a highly regulated environment, where she worked on digital transformation challenges. This experience revealed a major gap between ambition and reality: transformation only becomes tangible if it is controllable, measurable, and compliant. Finally, her time with the Buenos Aires government, where she managed the Education Ministry’s budget, reinforced this requirement for accountability. Every decision must be explained, every action must be justifiable. From these experiences emerged a structural conviction: in complex organizations, technology is only valuable if it is measurable, explainable, and controllable.
The blind spot of enterprise AI
For the past two years, companies have been massively experimenting with AI agents capable of chaining tasks, analyzing data, and making simple decisions. Yet a blind spot persists. Companies know how to deploy models and automate processes, but they know far less about how to continuously supervise what AI actually does. This lack of visibility creates concrete risks: uncontrolled decisions, errors difficult to detect, absence of evidence in case of audit, unclear accountability. In a context marked by the entry into force of the EU AI Act, this vulnerability becomes critical. Now, it is no longer enough to use AI—companies must be able to demonstrate how it works, at all times.
KLA Digital: a real-time governance layer
KLA Digital positions itself as a response to this governance deficit. The platform is not an additional artificial intelligence, but a control layer that integrates above existing systems. It allows AI to operate while making its actions visible, controllable, and traceable. This approach rests on a simple logic: not to slow down automation, but to frame it to make it usable in demanding environments.
Framing action: integrated control points
The first function consists of framing the action of AI agents. The platform introduces control points directly into automated processes. At certain key moments, human validation is required, particularly when the risk level increases. This mechanism allows for adjusting the degree of AI autonomy. A simple task can be fully automated, while a sensitive decision requires human intervention. This hybrid model rests on a central idea: automation should not be total, but progressive and controlled.
Continuous observation: measuring beyond results
The second function concerns real-time supervision. KLA Digital does not merely measure final results but also observes anomalies, deviations, and risky situations. “Near-misses,” often invisible, become valuable indicators for anticipating incidents. This capability transforms AI into an observable system, capable of being analyzed and continuously improved. It answers a question rarely asked in companies: what actually happens between two automated decisions?
Documenting and proving: the logic of the “Evidence Room“
The third function is that of proof. KLA Digital generates detailed traces of all actions performed by AI agents: decisions made, data used, human validations, and results obtained. This information is organized in a space called the “Evidence Room,” designed to produce documentation usable in case of audit or regulatory review. Within the framework of the EU AI Act, this capability becomes strategic. It transforms a regulatory obligation into a management and credibility tool.
Concrete uses in high-risk environments
KLA Digital primarily targets sectors where stakes are particularly high, such as banking, insurance, and pharmaceuticals. In these environments, automation is essential for processing large volumes, but the risk of error is critical. Use cases are concrete: customer file analysis, anti-money laundering (KYC/AML), risk management, operational support, processing of sensitive health data. In each of these areas, the question is not only to go faster, but to do it right and be able to demonstrate it.
A strategic shift: from performance to accountability
KLA Digital’s approach reveals a deeper evolution in AI adoption. Until recently, a system’s value was evaluated based on its performance. Today, a new criterion emerges: accountability. A powerful but uncontrollable AI becomes a risk. A governed AI becomes a strategic asset. This shift redefines companies’ priorities and raises a central question: will the ability to govern AI become more decisive than the ability to develop it?
Between innovation and regulation: a new balance
The debate often pits innovation against regulation, as two contradictory forces. On one side, the need to move fast. On the other, the obligation to secure. KLA Digital proposes a different approach: integrating regulation from the design phase. This logic transforms compliance into a competitive advantage, enabling accelerated deployments while controlling risks. It nevertheless requires organizational evolution, with strengthened collaboration between technical teams, business units, and compliance functions, as well as a culture of proof and transparency.
A new function within the company?
The emergence of solutions like KLA Digital could give rise to a new function within companies, at the intersection of technology, legal, and operations. A function dedicated to AI governance, capable of managing these complex systems. This perspective raises several questions: who should bear this responsibility, do companies have the necessary skills, and how can we prevent this governance layer from becoming a brake on innovation?
What the KLA Digital case reveals
The project led by Antonella Serine highlights an often underestimated reality. The challenge of artificial intelligence is not solely technological. It is organizational, cultural, and regulatory. It is no longer just about designing intelligent systems, but about building reliable, explainable, and auditable systems. From this perspective, governance is no longer a constraint, but a condition of existence.
Conclusion: regaining control without giving up innovation
KLA Digital is part of a fundamental dynamic: reconciling technological power with human control. By offering a real-time governance layer, the platform enables companies to move from experimentation to industrialization of AI. But beyond the solution, a question remains: are companies ready to face what their AI actually does? Governing means making visible, and making visible sometimes means questioning.
