Strengthening Model Governance with Modern Development Practices

As part of a wider engagement with a major UK insurer, Broadstone was supporting the review and enhancement of actuarial modelling processes. During this work, we identified that the client’s model development environment would benefit significantly from more formal source control and structured change management.

While source control was not the original scope of the engagement, we recognised that the existing approach created avoidable operational risk and inefficiency for the modelling team. The current approach led to:

• Limited transparency over how models evolved over time
• Inconsistent version control and documentation practices
• Reliance on informal or manual processes for validation and review
• Difficulty evidencing audit trails for internal governance and regulatory scrutiny
• Greater key-person dependency than was necessary for a business-critical modelling environment

We designed and led the introduction of a more controlled and transparent model development framework, centered on GitHub-based source control and modern development practices. This was a proactive recommendation identified during delivery, rather than a pre-defined project requirement.

Introducing structured version control

We recommended and drove the adoption of GitHub-based source control across the insurer’s team, drawing on our practical understanding of how much this would improve control, collaboration and efficiency.

Embedding controlled change processes

We reshaped the model change process so that governance was built directly into development, rather than being applied retrospectively after technical work had already been completed. This moved the team away from informal processes towards a structured, auditable development approach

Enhancing capability through AI-enabled tooling on subsequent engagements

The initial source control work was not AI-led. However, once stronger governance foundations were in place, Broadstone subsequently helped enhance development capability through AI-enabled tooling and modern developer workflows.

Improving documentation and auditability

We also improved the transparency of model artefacts and development decisions, making it easier for the client to evidence what had changed, why it had changed, and who had reviewed it providing clear linkage between model logic, assumptions and implementation.

The introduction of source control and structured development practices delivered a step change in both governance and efficiency. It also created a future-ready platform for the controlled adoption of AI-enabled development tools.