The insurance industry is caught in an AI gold rush, but not everything that glitters is fit for actuarial purposes.
According to AKUR8, as agentic AI shifts from hype to real-world deployment, pricing teams face a decision that could shape the next decade of their work: choose a tool designed for the job, or embed an opaque black box at the centre of their most heavily regulated process.
InsurTech firm AKUR8 recently talked about agentic AI for actuaries, what it is and why it matters.
Actuarial work carries a uniquely demanding standard that few other professions must balance. First is regulatory accountability, as pricing, reserving and capital models are submitted to regulators and signed off by certified professionals, meaning unexplainable outputs carry legal and compliance consequences.
Second is financial materiality: a flawed model does not simply generate a poor report, it can misprice risk and trigger adverse selection across an entire book of business. Third is the explainability standard, as actuaries must justify and defend every assumption in their models. A black box result is unacceptable, however accurate it may appear.
Understanding how AI arrived at this point helps clarify the stakes. Predictive AI moved actuaries beyond Excel, with multivariate models forecasting loss ratios, churn and underwriting tiers. Akur8 led this transformation in insurance, replacing legacy tooling with ML-powered pricing that paired statistical rigour with full transparency.
Generative AI then expanded capability from prediction to production, generating reports, code and analysis from natural language. However, regulators and practitioners have flagged its opaque decision-making and unpredictability, leaving it vulnerable to errors, hallucinations and misuse, a fundamental barrier for professionals whose outputs carry regulatory weight.
Agentic AI represents the next leap. These systems plan, decide and execute multi-step tasks with a degree of autonomy, shifting from answering questions to completing missions.
According to McKinsey’s State of AI 2025 report, 62% of organisations are at least experimenting with AI agents, while 23% report scaling agentic systems within at least one business function. Yet no more than 10% of respondents in any given function say their organisations are scaling agents broadly.
The danger with the wrong agentic AI is compounding error. Without human checkpoints, a flawed assumption made early can propagate through a workflow, affecting model structure, variable selection and ultimately the rates reaching policyholders. Layer on regulatory requirements around explainability and filing documentation, and an agentic AI without actuarial grounding becomes a liability rather than a tool.
Done properly, agentic AI should free actuaries from execution while they retain judgment. That includes automating tedious rating structure updates, generating reports from reusable templates, guiding formula building, flagging variable issues before modelling, working across languages, and leveraging competitor rate filings as a strategic baseline.
McKinsey’s data shows AI agent adoption in insurance scaling fastest in risk, legal and compliance (16%) and knowledge management (16%), while product development and strategy, where actuaries have the greatest impact, sit at just 0-2%. Akur8 argues that gap reflects a lack of fit-for-purpose tools rather than irrelevance.
Its Akur8 Agents run directly on the company’s platform infrastructure, grounding outputs in its ML-native pricing foundation, with prompts and data never reused for training, every action visible and nothing changing without approval.
Read the full AKUR8 post here.
Stay ahead of the decisions shaping InsurTech. Subscribe to FinTech Global’s newsletter for strategic intelligence and early insight trusted by industry leaders.
Copyright © 2026 FinTech Global
