LO
Insurance underwriter — tasks, one by one
The unit of analysis is the task, not the job title. Each one below carries its direction, whether the judgement rests on evidence or on platform inference, the reasoning, and what it does not establish.
Every task on this page#
Underwriting standard applications
Automating✓ Evidence-backedAccepting or declining straightforward applications — healthy applicants, small sums, standard small-business risks — against the insurer's rules.
Rules engines and models now read the application, pull in the data they need and decide the straightforward cases without an underwriter. Prudential plc's filing for 2024 says around 74% of its new business policies were processed through auto-underwriting capabilities, alongside 96% submitted electronically.
The filing does not say which products or markets the figure covers, how the rest were handled, or whether any underwriting roles changed.
Complex and large risks
Being augmented≈ Platform inferenceApplicants with medical history, large sums insured, unusual properties and commercial risks that need reading, questions and judgement.
Software summarises medical records, inspection reports and financial statements and flags what matters. Deciding whether to take an unusual risk, and on what terms, is a judgement the insurer holds a named underwriter responsible for.
This rests on what underwriting software can do, not on a measurement of how complex cases are now handled.
Setting terms, exclusions and loadings
Being augmented≈ Platform inferenceChoosing the exclusions, extra premiums and conditions that make a risk acceptable, within the insurer's guidelines.
Rating models suggest a price and standard terms. Tailoring them to a case that does not fit the model, and knowing when a loading is fair to the applicant, stays with the underwriter.
This rests on what rating and pricing tools can do, not on a measurement of how often underwriters now change their output.
Referrals, agents and brokers
Still human-led≈ Platform inferenceHandling the cases the rules refer out, answering agents' and brokers' questions, and explaining a decline or a loading.
Automation sends the cases it cannot decide to people, and those are the ones where an agent or broker wants an explanation or a negotiation. Explaining a decision so that it holds, and adjusting it when new information arrives, is a conversation.
This rests on how referral and broker work is done, not on a measurement of how many cases are now referred.
Maintaining the rules the machine decides by
New task≈ Platform inferenceWriting and tuning automated underwriting rules, checking their decisions, and watching for unfair or drifting outcomes.
When most applications are decided by rules, the underwriter's knowledge moves into the rules: which answers trigger a referral, which data sources to trust, and whether the machine's decisions stay within the insurer's guidelines and the regulator's expectations.
No source yet measures how many underwriters now work on rules and models rather than cases; this is an inference from the direction of the work.