About Cammelot

The Cammelot project began as an independent research initiative to explore the systemic boundaries of artificial intelligence within public infrastructure; it has since evolved into a comprehensive simulation framework that uses real-world data to test how different technological architectures affect the delivery of healthcare in the Netherlands.

The Scope of the Project

Cammelot is a digital representation of a Dutch town consisting of forty-five citizens, three general practitioners, and one hospital. The health outcomes within the simulation are determined by actual disease prevalence data from the RIVM, while the demographics and financial costs are based on CBS mortality tables and NZa tariff schedules. The 16-bit visual style is a deliberate choice intended to focus the user’s attention on the underlying patterns of the system rather than the aesthetic details. In this environment, citizens move through their daily lives and interact with healthcare providers; when the system cannot process patients efficiently, the simulation records the resulting health decline and mortality events.

Technical Architecture and Lineage

The design of Cammelot is informed by the “Generative Agents” research published by Park et al. (Stanford/Google, 2023), which demonstrated how AI agents could simulate complex social behaviors in a virtual town. Cammelot adapts this “cognitive loop” of perceiving, reflecting, and planning for a specific healthcare context. Instead of open-ended social interactions, these agents are driven by epidemiological data and regulatory constraints.

The project involved running 100 simulations of the current healthcare status quo and 100 simulations of a system overhaul where AI acts as the primary connective tissue. By comparing these two regimes, we can observe the impact of technology on variables such as specialist wait times, doctor burnout, and the equitable distribution of care across different age groups.

About the Author

I am Simone Cammel, and I have worked in Big Tech for the last six years. Because of my professional background in the technology sector, it should be assumed that I have a bias toward the potential of artificial intelligence. This is the reason I built Cammelot as a “falsification machine” to see if a data-driven model could prove those biases wrong.

I believe that the current discussion regarding AI in healthcare is often focused on the wrong objectives. Many see AI as a way to increase the speed of existing tasks, such as writing emails or taking notes; however, I believe the real potential lies in restructuring how entire systems operate. This simulation is an attempt to be honest about which problems AI can solve and which ones it might worsen.

Open Source Commitment

Transparency is a core requirement for this project, and therefore all code and simulation data are available for public review. If you find a flaw in the logic or a mistake in the statistical analysis, I encourage you to share your findings. The project is designed to be a collaborative effort where the parameters can be adjusted by anyone who wishes to test a different hypothesis.

Primary Data Sources

The simulation parameters are drawn from the following public Dutch healthcare datasets to ensure the results are grounded in reality:

• CBS, Centraal Bureau voor de Statistiek - population demographics and baseline mortality tables
• RIVM, Rijksinstituut voor Volksgezondheid en Milieu - age-stratified disease prevalence and comorbidity risk factors
• NZa, Nederlandse Zorgautoriteit - healthcare tariffs and legal waiting time standards (Treeknorm)
• LHV, Landelijke Huisartsen Vereniging - administrative workload and capacity of general practitioners
• IZA, Integraal Zorgakkoord - transformation targets for system interoperability and staffing

Contact and Collaboration

If you have a research question, a critique of the model, or an interest in collaboration, you can reach out via the following channels: