Notes
Interactive essays, not hot takes.
The 2026 AI shifts that actually matter for actuaries — each one retold with something you can play with, and an argument you can disagree with. No breathless futurism; distributions, incentives, and audit trails.
The harness raises the floor.
Why the same model is chaos for one team and a colleague for another. A reliability simulator you can flip switches on, the case that consistency — not capability — is the 2026 bottleneck, and why actuaries walk into this discipline with the mental furniture already installed.
The wiki that writes itself. New
Team knowledge doesn’t get lost — it decays. Karpathy’s LLM-wiki pattern read as the study know-how book that keeps itself true — every close-out becomes an ingest, every “how did we handle this last time?” a query; the agent files and lints, the actuary approves. With a knowledge-decay simulator — shared drive vs annual review vs continuous lint.
2026 for actuaries, in five shifts soon
The handful of things that actually changed this year underneath the noise — retold interactively, with a skeptic's scorecard for each.