WELCOME
The EU LDT AI Notebook enables cities and technical teams to design, develop, and operationalise artificial intelligence models within a unified and scalable environment.
By combining interactive notebooks, model lifecycle management, and workflow orchestration, it allows users to move seamlessly from experimentation to deployment, leveraging data, federated learning, and shared assets across the EU LDT Toolbox.
THE CHALLENGE
Cities across Europe increasingly rely on data to inform decisions, but transforming that data into actionable intelligence remains complex.
AI development is often fragmented across tools, environments, and teams, making it difficult to manage model lifecycle, ensure reproducibility, or deploy solutions at scale. Additionally, integrating datasets, validating models, and operationalising them within real-world use cases requires specialised expertise and infrastructure that many cities lack.
The EU LDT AI Notebook addresses this gap. It provides a structured and interoperable environment where models can be developed, tested, validated, and deployed within a single framework, fully connected to the EU LDT ecosystem.
HOW IT WORKS
The AI Notebook connects the full AI lifecycle in a structured flow, from development to deployment and reuse.
It enables a continuous loop where models are created, tested, validated, deployed, and improved based on performance and feedback.
It supports cities and technical teams in exploring a catalogue of AI models, creating workflows through interactive notebooks, and running model training or inference in a consistent way.
A model project represents the full lifecycle of an AI model, including its development history, configurations, datasets, and performance metrics. It ensures traceability, version control, and reproducibility of AI assets.
Experiments organise multiple workflow executions with different parameters, enabling comparison of results and identification of the most suitable model. KPIs and metrics are tracked to evaluate performance.
Validated models are deployed as inference interfaces, enabling real-time or asynchronous predictions. These services can be integrated into applications, simulations, or external systems.
INTEGRATION WITH THE EU LDT TOOLBOX
→ Input
Receives datasets from the EU LDT Use Cases & Scenarios, trained models from Federated Learning, and user authentication via Identity Management.
→ Output
Produces models and exposes inference services for use in applications and predictions.
The AI Notebook operates as the AI execution layer of the EU LDT Toolbox, enabling cities to transform data into
predictive and actionable intelligence.
WHY IT MATTERS
Cities need to move from data availability to actionable intelligence, but often lack the infrastructure and tools to operationalise AI effectively.
See how cities make better decisions.
KEY CAPABILITIES
Interactive environments for building, testing, and iterating models with full flexibility.
Versioning, metadata management, and traceability through the Model Registry.
Execution of training pipelines, KPI tracking, and comparison of results across experiments.
Deployment of models as APIs with synchronous or asynchronous execution.
Integrate with external applications and other LDT tools (Data Platform, Marketplace, Federated Learning).
Ensure compliance through validation workflows and approval processes before model publication.
WHO USES IT
LOOKING FOR DOCUMENTATION?
Access the right resources based on your role and what you need to achieve.
Are you a policy, strategy or innovation lead?
Understand how the EU LDT Toolbox supports strategic planning, policy design and impact measurement.
Explore key concepts, frameworks and real use cases to make informed decisions.
Are you part of a technical or implementation team?
Deploy, integrate and extend the Toolbox in your city environment.
Includes architecture, configuration guides, APIs and source code to support implementation.
New to the Toolbox? Start with the core concepts to understand how everything connects.
STANDARDS AND COMPLIANCE
The AI Notebook is mainly grounded in MIM 2, MIM 3, MIM 10, with MIM 4 and MIM 8 reinforcing interoperability and deployment flexibility across the ecosystem.
JOIN OUR COMMUNITY
Connect with the EU LDT ecosystem and stay updated on releases, pilots and collaboration opportunities.