AI, regulation and technological sovereignty: building architectures for auditable systems
22 October 2026 | 12.00h to 12.55h
With the entry into force of new AI regulations in Europe, AI systems must be designed to incorporate regulatory compliance requirements into their very architecture. Aspects such as AI governance, data and model traceability, observability, explainability, risk management, and human oversight are becoming essential to ensure auditable, transparent, and trustworthy systems. In this session, we will explore real-world examples of how AI architectures are being designed and deployed to meet regulatory requirements in sectors such as pharmaceuticals, public administration, and insurance.
Participants:
-
Toni Manzano Doñabeitia, CSO and co-founder, Aizon
-
Eva Aymamí Gili, Digital employee & customer experience manager, ESTEVE
-
Mikel Urizar, Manager of the Generative AI Center of Excellence for Iberia and Latin America, Inetum
-
Marc Oliveras Xiol, Vice president of engineering and CTO at Captio, Emburse
Speakers:
TONI MANZANO DOÑABEITIA
CSO and co-founder, Aizon
Toni is the co-founder and CSO of Aizon, an AI SaaS company for the pharmaceutical and biotech industries. He is a board member of PDA Regulatory Affairs and co-chair of the Advanced Manufacturing PDA Interest Group. He teaches AI courses at the Universitat Rovira i Virgili (URV) and serves as a subject matter expert for the United Nations on AI in the life sciences. He is the author of numerous publications in the pharmaceutical field and holds a dozen international patents. Toni holds a degree in Physics, a PhD, a Master’s degree in Information and Knowledge Society, and a postgraduate qualification in Quality Systems for pharmaceutical manufacturing and process research.
AI applied to pharmaceutical manufacturing: when regulation is the main driver
The pharmaceutical industry faces constant tension between the pressure to adopt AI and the obligation to maintain the controls required by GxP regulations. Aizon has developed a platform that embeds the entire model lifecycle within an auditable framework: every data point has provenance, every prediction is traceable, and every model version is reproducible. The architecture can be deployed on on-premises infrastructure or European cloud environments, ensuring true technological sovereignty. Results include a 20–40% reduction in deviations, faster batch release, and up to a 50% reduction in validation effort, without compromising the safeguards that protect patients.
EVA AYMAMÍ GILI
Head of customer and employee experience, Esteve Healthcare
Eva Aymamí Gili has more than 20 years of experience in the technology sector. She is currently Digital Employee & Customer Experience Manager at ESTEVE, where she leads digital strategies aimed at improving the experience of both employees and customers. Her work focuses on optimizing the use of tools and processes for sales and marketing teams, driving operational efficiency and ensuring the successful adoption of digital solutions across the organization.
MIA, ESTEVE's AI agent portal
This presentation describes a corporate AI agent portal, accessible through Microsoft Teams and the digital workplace, that enables the creation and use of AI agents in a scalable and governed way. The project includes generic, contextual, and specialized agents, with role- and audience-based access control, as well as usage traceability. It also features a use case in Chemical Reagents and incorporates security, governance, and usage monitoring mechanisms, ensuring a controlled, efficient environment aligned with business needs.
MIKEL URIZAR
Manager of the Generative AI Center of Excellence for Iberia and Latin America, Inetum
He holds a degree in Management Information Systems Engineering from the Universitat Rovira i Virgili and has nearly two decades of experience in the development, implementation, management, and maintenance of complex systems, as well as in organizational and business development. He is currently Manager of the Generative AI Center of Excellence at Inetum for Iberia and Latin America. His role focuses on business transformation, process optimization, and the development of innovative solutions through the use of disruptive technologies such as generative AI.
Craft: IA Generativa sobirana i auditable
Every organization adopting generative AI faces the same tension: capture its efficiency without losing control of data, costs, and compliance. Too often that means vendor lock-in. Craft is a governance and orchestration framework that lets you own your AI ecosystem—whether you’re a regulated enterprise or a public administration. It runs fully on-premise or air-gapped, swaps any model behind a single gateway, and enforces compliance as code: automated bias and PII gates plus immutable audit logs, auditable by design. The result is sovereignty without sacrificing speed: new use cases in days, with continuous control of cost, risk, and regulation.
MARC OLIVERAS XIOL
Vice president of engineering and CTO at Captio, Emburse
Marc Oliveras is vice president of engineering at Emburse, where he leads mobile and AI development teams across the United States, Canada, Europe, and India. He specializes in turning artificial intelligence into tangible business value by accelerating product innovation, improving team efficiency, and delivering scalable solutions.
