Exclusive on-site program (starting at 14:00 CET)
We will host an in-person program for registered on-site attendees, featuring senior representatives and thought leaders from the German public sector and healthcare ecosystem. The sessions focus on practical implementations of confidential computing and secure AI in regulated environments, providing insights into real-world deployments and compliance-aligned architectures.
The event also provides a unique opportunity to engage directly with experts in confidential computing, including speakers and sponsoring organizations.
Location: Amplifier Berlin (Gustav-Meyer-Allee 25, 13355 Berlin)
Pre-event talks: confidential computing & AI in action
See abstract
Registers form the digital backbone of modern public administration in Germany – from trade and vehicle registration registers to professional and environmental registers, they provide the foundation for public administrative services. Their modernization was formally mandated by the Register Modernization Act (RegMoG).
In practice, however, today's register landscape consists of outdated, heterogeneous structures that are costly and complex to operate.
This is precisely where the Online Register Platform (ORP) comes in: it is an open, reusable architectural and software foundation that simplifies and accelerates the setup and operation of modern registers. It additionally incorporates the so-called RaaS (Register as a Service) architectural guidelines.
Confidential Computing enables the logically tenant-separated operation of registers across many (cloud) runtime environments. Through hardware-backed Trusted Execution Environments, sensitive data can be processed in encrypted form, while simultaneously a trusted remote attestation ensures secure data exchange between the various authorities. The result is a modular, privacy-compliant register infrastructure that both respects federal requirements and significantly accelerates digital administrative processes."
See abstract
Primary care practices in Germany face a dilemma: the demand for AI-supported assistance in documentation, knowledge retrieval, and administrative tasks is substantial, yet the use of conventional cloud-based AI services is incompatible with medical confidentiality obligations and GDPR requirements.
Already, 45% of Germans use AI chatbots for health-related questions, and 15% of practices have begun implementing AI solutions. However, operating proprietary AI infrastructure is typically neither economically nor technically viable for individual practices.
HÄVG Hausärztliche Vertragsgemeinschaft AG is connected to more than 16,000 general practitioners through the General Practitioner–Centered Care model (Hausarztzentrierte Versorgung, HZV), collectively serving approximately 10 million insured patients. In a feasibility study, HÄVG demonstrated how Confidential Computing can address this challenge.
In collaboration with Edgeless Systems (Privatemode AI), an architecture was developed that enables AI processing within a trusted execution environment. The solution is aligned with the gematik specification “Healthcare Confidential Computing.” Data remains cryptographically protected both in transit and during processing. Access by cloud or AI providers is demonstrably excluded.
Based on this architecture, an “HZV AI Chat” was developed as a demonstration application and deployed to selected practices. The application includes, among other features, privacy-secure web search with human-in-the-loop oversight and quality-assured clinical guideline research.
The solution integrates specialized medical LLMs developed by Fraunhofer IAIS, optimized for the German healthcare system. Benchmarks demonstrate significant performance advantages compared to generic open-weight models.
The presentation outlines the technical architecture, practical use cases, and initial findings from the ongoing evaluation, and discusses the transferability of this model to other regulated industries.
See abstract
The software development lifecycle is being fundamentally reshaped by AI — but isolated copilots and bolt-on assistants only scratch the surface. GitLab Duo Agent Platform (DAP) takes a different approach: purpose-built AI agents that operate across the entire DevSecOps workflow, deeply integrated into the platform developers already use every day.
In this session, you'll learn what the GitLab Duo Agent Platform is, how it goes beyond simple code completion to orchestrate tasks across planning, coding, testing, security, and deployment. We'll explore the architecture behind DAP, look at real-world use cases, and demonstrate how teams can build and leverage custom AI agents to automate complex workflows — all within a single, unified platform. Whether you're a developer, platform engineer, or engineering leader, this talk will give you a clear picture of where agentic AI in DevSecOps is headed and how to get started today.
