Which AI agent skills matter most for underwriters?

Deciding the referrals an agent holds is the underwriter's most decisive skill, because referrals are where the unusual risks sit. Ranked after it: weighing risk information by where it came from, pricing within authority, watching accumulations, reviewing drafted wordings, and feeding corrections back so the agent refers fewer clean risks.

What does an agent take on in underwriting?

Underwriters assess proposals, gather information, price risks, set terms and wordings, and manage their portfolio within authority. An agent can read submissions, extract and check data, pull external information, apply underwriting rules and quote standard risks at depth 3 within the insurer's own limits.

Anything outside appetite or authority is referred. The underwriter decides referrals, writes non-standard terms and keeps the portfolio view the agent does not have.

Which skills matter most, in order?

1. Deciding referrals. Each referral arrives with the rule that triggered it and the agent's summary. The underwriter must read past the summary to the facts that matter, and decide to accept, load, exclude or decline.
2. Weighing information by source. The proposal form is declared by the applicant. A survey report is signed by a named surveyor. Claims history from the insurer's own system is sealed. Data scraped from the web is external. An underwriter weighs a risk differently depending on which of these carries the key facts.
3. Pricing within authority. The agent proposes a price. Under the EU AI Act, risk assessment and pricing for life and health insurance are listed as high-risk. In any line, the underwriter must be able to explain a loading and stay within their own authority.
4. Watching accumulations. An agent quoting one risk at a time may not see that ten properties on the same street, or several policies on the same supply chain, are building an accumulation. The underwriter keeps that view.
5. Reviewing wordings. The agent drafts endorsements and exclusions from the sealed wording library. The underwriter checks that they say what was intended and do not conflict with the base policy.
6. Banking corrections. When an underwriter accepts a referral as clean, a short note explains why, so the rule can be refined.

What does the agent do, and what stays with the underwriter?

What does deciding a referral look like?

Take a commercial property insurer. The agent refers a small warehouse because the declared construction type triggers a rule. The summary says "combustible cladding declared; outside appetite".

The underwriter opens the file. The declaration is the broker's, declared. A signed survey from the previous year shows the cladding was replaced with non-combustible panels. The underwriter asks the broker to confirm, receives an updated declaration, and accepts the risk at standard terms. The correction note asks the builder to compare declared construction with any signed survey on file before referring. Clean risks like this one stop clogging the referral queue.

Collections and IT roles look at agents from very different positions; compare skills for collections agents and skills for IT administrators. Where agents quote alone, see the skills that matter when agents act alone on routine work.

How should underwriting teams build these skills?

Replay last quarter's referrals with outcomes hidden and compare decisions across the team. Map who may approve what in the skills matrix template for an AI agent team. For market specifics, see underwriter training in insurance in Ireland. The full map is in AI agent skills by role.

Where the Academy fits

Underwriters who decide referrals and approve pricing fit the Operator track at Pryme Intelligence Academy, with module tests set in insurance and in the underwriting role. Senior underwriters who help design the agent's rules may go on to the Builder track, which ends in a capstone agent certified in Attest to depth 2 or higher.