Rohan.
← All posts
1 min read

Why actuaries should care about agentic AI

Agentic AI moves models from answering questions to completing tasks. Here is what that shift means for pricing and reserving work.

A traditional model answers a question you pose. An agent decides which questions to ask, calls tools, and works toward a goal you set. For actuarial work, that is a meaningful change in how software fits into the day.

A small example

Consider a pricing check you run every month. An agent can pull the exposure data, run the model, compare against last month, and flag what moved — before you open the workbook.

price_check.py
def price_check(portfolio):
    exposure = load_exposure(portfolio)
    result = run_pricing_model(exposure)
    delta = compare_to_prior(result)
    return flag_material_moves(delta, threshold=0.05)

The actuary stays in control of judgement and sign-off. The agent removes the mechanical steps in between.