Compliance Studio - behind the system

The development of Compliance Studio originated from a recurring observation in regulatory practice. Across more than 200 interviews with Regulatory Affairs and Quality Assurance professionals in medical devices, IVD, pharmaceuticals, and biotechnology, a consistent pattern emerged.

The central challenge was not access to regulatory information, nor a lack of tools. Instead, it concerned the ability to determine and maintain a correct regulatory scope for a given product across multiple jurisdictions.

This distinction proved critical.

Observed problem

In most organisations, regulatory scope is not established as a stable construct. It is continuously reconstructed through interpretation of regulations, internal documentation, and accumulated experience.

While this approach can function operationally, it introduces structural limitations. Interpretations vary between teams and individuals, decisions are not always traceable, and maintaining consistency over time becomes increasingly difficult. Most importantly, uncertainty persists regarding whether the defined scope is complete and still correct.

These constraints were observed regardless of organisational maturity.

Research approach

Rather than addressing symptoms at the level of workflows or document management, the development effort focused on the regulatory landscape itself.

Over a three-year period, more than 20,000 regulatory instruments were collected, reviewed, and structured. Particular attention was given not only to individual regulations, but to how they relate to one another across jurisdictions and how they are implemented in practice.

This work revealed a fundamental issue: regulatory frameworks are not organised for operational use. They are distributed across legal layers, dependent on product characteristics, and subject to variation in local implementation. As a result, determining applicability requires understanding both individual regulations and the system they form.

System design

Based on these findings, the system was designed to treat regulatory scope as a structured and governed entity.

This required first transforming regulatory content from document-based sources into structured data. Regulations were classified, decomposed, and connected, allowing them to be interpreted within a defined framework rather than in isolation.

In parallel, relationships between jurisdictions were explicitly modelled. Supranational regulations, national implementations, and regulatory counterparts were mapped in a way that made alignment and divergence visible and computable.

Only after this structure was established did it become meaningful to introduce a reasoning layer.

Use of AI

AI is applied within the system, but its role is deliberately constrained.

Rather than generating conclusions from unstructured regulatory text, AI operates on top of the structured regulatory model. This allows it to interpret relationships, assess applicability based on defined product attributes, and evaluate the impact of regulatory changes within a controlled context.

This distinction is essential. Without an underlying structure, AI tends to produce outputs that are difficult to verify and reproduce. By contrast, when applied to structured data, outputs can be made consistent, explainable, and traceable.

AI is therefore not used to replace regulatory judgement, but to support reasoning within clearly defined boundaries.

A more detailed description of this approach, including its limitations and appropriate use, is provided in our AI literacy material.

Implementation

The implementation of this system required iterative validation of both data structures and regulatory relationships. Classification models, mapping logic, and system architecture were refined continuously to ensure that outputs remained aligned with regulatory reality.

The resulting system comprises a substantial technical foundation, integrating structured regulatory data with a controlled reasoning layer.

Operational outcome

From a user perspective, interaction with the system is intentionally simplified. A product is defined through a set of attributes, after which the system determines applicable regulations, structures them across selected markets, and maintains that scope over time.

Regulatory updates are no longer treated as isolated events. Instead, they are interpreted in relation to the existing regulatory scope, allowing their relevance and potential impact to be assessed within context.

Discussion

The approach differs from conventional regulatory tools, which typically focus on document retrieval, monitoring, or workflow support. While these functions remain useful, they do not address the underlying problem of defining and maintaining regulatory scope as a structured entity.

By contrast, treating regulatory scope as a governed system introduces a different operational model. It reduces reliance on manual reconstruction, improves traceability, and provides a more stable basis for handling regulatory change.

Conclusion

The complexity of regulatory frameworks has not decreased. However, the ability to structure and connect regulatory data changes how that complexity can be managed.

By establishing a structured regulatory foundation and applying a constrained reasoning layer, it becomes possible to define and maintain regulatory scope with greater consistency over time.

This forms the basis of Compliance Studio.

Explore Compliance Studio