The situation
Scenario logic lived in actuarial models, exposure data lived in portfolio systems, and the judgement connecting them lived in email threads and spreadsheets.
Complexity was being transferred to the user
The models were sound. The problem was that an underwriter had to assemble context manually before comparing two scenarios — pulling exposure from one system, assumptions from another and expert commentary from correspondence.
That reconstruction work was invisible in every process map and consumed a meaningful share of the working day. It also meant two underwriters could reach different conclusions from identical inputs without either being wrong.
From model outputs to a decision environment
Discovery ran with underwriters, actuarial teams and the Head of Insight together rather than sequentially, which surfaced disagreements about what the numbers meant while they were still cheap to resolve.
Delivery covered multi-source data ingestion, pipeline automation and interactive scenario visualisation. The interface held assumptions, portfolio exposure and commentary in one traceable view, revealing detail on demand rather than by default.
Expert judgement remained visible
The system was built to make expert reasoning inspectable, not to replace it. An underwriter could see which assumptions drove a result and say why they disagreed — which is what makes a decision defensible in a regulated environment.
Keeping judgement in view rather than automating it away is also what made the platform a credible foundation for later analytical and AI-assisted work.