Insights
Practitioner writing on actuarial work, data engineering, AI and model validation — and the operating layer that ties them together.
IFRS 17 implementation is finished. IFRS 17 analytics has barely started. The insurers that treat phase two as a software upgrade rather than a reporting refresh will discover, slowly and expensively, that the standard's value is in the explanation — not the journal entry.
Triangle methods aren't broken. They're just leaving information on the table. The reserving frontier is not about replacing chain-ladder — it is about getting at the claim-level signal aggregate cells throw away, without losing the audit trail.
A quiet renaissance is underway in the actuarial profession, and most of us are too busy reconciling cashflows to notice. The work that has crowded out actuarial thinking for three decades is exactly what AI is best positioned to take over.
The most interesting AI work in actuarial right now is not in pricing. It is in the model office — and specifically in the slow, repetitive, evidence-heavy world of ALM and solvency, where the gap between what an analyst spends a quarter doing and what an agent can credibly automate is the widest in the firm.
Cyber is the hardest actuarial modelling problem in insurance today, and the reason is not technical. It is that almost every assumption underwriting the traditional toolkit — stable threat environment, independent losses, credible history, fixed wording — is now wrong at the same time.
A mortality basis built only by extrapolating the last twenty years of history is, in 2026, an assumption about the future that the future has stopped supporting. The shift in mortality and morbidity work is no longer cosmetic — it is structural.
A pricing model that wins on out-of-sample deviance but cannot be explained to the pricing committee is not a better model. It is a worse one. Interpretability is not a cosmetic feature in insurance pricing — it is part of model fitness.
Some risks are too large, too correlated or too uncertain for private insurance to carry alone. Pretending otherwise has produced two decades of widening protection gaps in South Africa. Pooling is the structurally honest response to the kinds of risk that ordinary insurance cannot price into affordability.
The most common mistake actuarial teams are making with AI agents in 2026 is choosing the most autonomous pattern they can defend, when most of the productivity is in the most boring one they can ship. Five deployment patterns, ranked smallest-first.
A 1-in-100-year event estimated from historic experience may no longer have the same frequency under changed climate conditions. That single sentence, taken seriously, breaks most of the climate work being done in insurance today — and points at the actuarial discipline that is going to replace it.
