Underwriting teams already have authority limits and peer review; train them to apply both to agent work. Teach how the agent reaches a quote or referral, calibrate on past referrals, shadow live submissions beside a certified underwriter for two weeks, then sign off. Approval authority for agent drafts should follow that sign-off, line by line.
What does an underwriting team review?
An underwriting agent can extract facts from submissions, check them against guidelines, price within authority and draft quotes, endorsements and renewal terms. What reaches underwriters: quotes for review before issue, referrals above the agent's authority or outside guidelines, renewals where loss experience has changed, mid-term adjustments, and risks the agent refused as outside appetite.
The head of underwriting owns the programme. A senior underwriter with a current certificate runs calibration and shadowing. Each underwriter's existing authority, the lines and limits they may write, stays in force; agent review is added to it, not substituted for it.
How does this build on how underwriters already work?
Underwriters are used to written authority and to having their files peer reviewed. That is a strong base. Two things are new.
The first is reviewing someone else's draft at speed. A quote the agent drafted looks complete; the underwriter must check its inputs, not just its total. Each rating factor carries a provenance stamp: sealed if it came from the insurer's own systems, signed if a named party such as a surveyor answers for it, declared if the applicant simply states it, external if it came from elsewhere. Training teaches underwriters to scan those stamps first.
The second is knowing why the agent stopped. A referral names the bound that decided: role scope, certificate, sources, the insurer's own limits or the platform guardrails. An underwriter who knows the agent referred because its certificate for that line has lapsed will decide differently from one who assumes the risk itself was unusual.
The steps
- Person (head of underwriting): lists decision types and red lines. Decision types: issue a drafted quote, decide a referral, agree renewal terms, approve an endorsement. Red lines might include: issuing a quote that relies on a declared fact where guidelines require evidence; deciding a referral above one's own authority; overriding an agent refusal on appetite without the line head.
- Person (underwriter): completes the learning on depth, the minimum chain, verdicts, provenance and reasons.
- Person (senior underwriter): runs calibration on 40 anonymised past referrals and quotes, chosen where experienced underwriters historically disagreed.
- Agent, depth 2: drafts quotes and refers live submissions in shadow mode; the trainee decides each in a shadow queue while the senior underwriter decides for real.
- Agent, depth 1: logs both decisions, the rating factors relied on and their provenance.
- Person (senior underwriter): debriefs twice a week, using the existing peer review format.
- Person (head of underwriting): signs off after two weeks at the bar, by line of business.
- Person (workspace admin): grants approval rights matching the underwriter's written authority and sign-off.
| Step | The agent does | Depth | Stays with a person |
|---|---|---|---|
| Fact extraction | Extracts and stamps provenance | 1 | Underwriter checks key factors |
| Quote within guidelines | Drafts | 2 | Underwriter reviews and issues |
| Referral above authority | Prepares and holds | held | Underwriter with authority decides |
| Renewal with changed loss experience | Drafts revised terms and holds | held | Underwriter decides terms |
| Risk outside appetite | Refuses | refused | Line head decides any exception |
What should calibration focus on?
Disagreement and inputs. Choose referrals where the team has split before: a risk with an improving loss record after a bad year; a property with self-declared protections; a renewal where the broker presents a new risk manager as reason for a rate reduction. Each underwriter decides alone with a reason, then the group compares.
Then reverse the exercise. Give the team quotes the agent drafted and ask them to find the weakest input in each. Underwriters who can name it quickly, with its provenance, are ready for shadowing.
Which controls matter, and what evidence do they leave?
- Authority alignment. Evidence: approval rights matching each underwriter's written authority, with certificate expiry.
- Provenance check on issued quotes. Evidence: the classes of the rating factors relied on.
- Independent shadow decisions. Evidence: trainee and senior decisions filed separately.
- Peer review of agent-drafted files. Evidence: the review record in the usual format.
- Append-only record. Evidence: every verdict and decision, for audit and later claims.
What should you measure?
Agreement with the senior underwriter by line and decision type. Red-line misses, which must be zero. Quotes issued and later found to rest on unchecked declared facts. Referral turnaround, which brokers notice. And for each line, how often the deciding bound is the agent's certificate, a sign that recertification may be due.
Where does it usually go wrong?
Teams review the premium and not the inputs. The total looks reasonable; the sprinkler declaration was never checked.
Authority in the workspace drifts from written authority. Keep them matched and review both together.
The agent's certificate for a line lapses after 90 days or a run that clears no depth drops it to depth 1, and the team is suddenly flooded with referrals. Track expiry by line and plan cover.
Other functions follow the same arc; compare training the IT team and training the operations team. New readers may find what underwriting is a useful primer. Readiness can be recorded on the reviewer onboarding checklist for the underwriting team and graded with the competency matrix for the underwriting team. The guide to training teams for AI agents sets the order across the business.
Where the Academy fits
Pryme Intelligence Academy's Operator track covers how a governed agent decides, human review and reading the workspace, with module tests set in the learner's own industry and role. Certified Pryme Operator includes a two-week practicum of real review decisions signed off by a Pryme assessor. An insurer can make referral authority in the workspace depend on a current certificate, line by line.
Questions people ask
Does an underwriter's written authority change with an agent?
It should not widen. The agent adds drafts and referrals to review; the underwriter's authority still sets what they may decide.
Can the agent issue quotes without review?
For simple, well-evidenced lines it may, once certified and within the insurer's limits, with underwriters sampling. Complex or high-value lines usually stay at depth 2.
Who decides when the agent refuses a risk as outside appetite?
The line head, if an exception is to be considered at all. A refusal reflects the insurer's appetite and limits, so overriding it is a policy decision, not an individual one.