Why insurance, and what the measured returns cover
Insurance is consistently ranked among the industries with the most to gain from this technology (McKinsey, BCG; see the references behind the integration phases). The reason is the raw material: submissions, loss runs, wordings, broker email and claims files are the document work an LLM reads well.
What has actually been measured is narrower than that ranking suggests, and the gap between the two rows below is the whole strategy. Assistive work pays: drafting, summarizing and extracting typically save 20–50% of the time on document-heavy tasks. Reported results for fully automated use cases are mixed. The pattern that wins is AI drafts, a person decides.
Figure
What the measured returns actually cover
Time saved on document-heavy work. The evidence supports assistance, and does not yet support handing the task over.
Assistive · AI drafts, a person decides
Fully automated · no person in the loop
Drawn from measured carrier and vendor results. The second row is empty rather than estimated: reported outcomes for full automation vary too widely to state as a range, and that is itself the finding.
Which use cases pay, ranked by measured value over implementation risk, is set out on Costs & value; anything that influences underwriting, pricing, claims or customers needs the controls in Governance before it goes near a decision. The return is not spread evenly across the industry, though, and the rest of this page is about where it concentrates.
The leverage is in the hours before the judgment
Personal lines and standardized small-commercial products increasingly run on structured data, rules, and straight-through processing. Specialty and other large-complex accounts stay bespoke, document-heavy, and expert-driven: manuscript wording, layered exposures, broker negotiation, loss runs, engineering reports, and exposure schedules all need account-level judgment (Capco, which also puts 30–40% of underwriter time in administrative work).
The variable that moves is account size, not the specialty label. Accenture's survey of 434 P&C underwriters found that the share of time going to non-core and administrative work falls as accounts get larger, from 47% on books averaging under $10,000 in premium to 35% above $250,000, while the share going to risk analysis and pricing climbs from 22% to 33% (Accenture, 2021 P&C Underwriting Survey). Read against a specialty book of 50 respondents, the specialty and commercial splits are near-identical; it is the premium size of the account that separates them, and specialty is where the large-complex accounts sit.
Figure
As accounts grow, less of the month goes to administration
Share of an underwriter's time on each kind of work, at the two ends of the account-size range. The administrative share falls; the risk-analysis share rises.
Source: Accenture, 2021 P&C Underwriting Survey: shares of time, not hours per account; only the published endpoints of the account-size range are shown.
Two cautions on that figure. It measures shares of a month, not hours per account, so it says how underwriter time is divided rather than how much of it a single account consumes, and no public source we could find gives the absolute number. It is also five years old, and Accenture's later work reports incremental improvement rather than a different shape.
What the shares do establish is that even on the largest accounts a third of the work is still administrative, and most of it lands before any risk judgment starts. A submission can wait hours or days while emails and attachments are read, keyed, validated, and made decision-ready. Hiscox described manual extraction in one London Market workflow as taking up to three days (Hiscox); Arch reported a two-to-three-day intake-to-decision cycle before automating submission intake (Ivans/Arch). The capacity cost compounds: Deloitte reports that many specialty insurers process only 20–30% of the submissions they receive, because manual reentry and disconnected workflows cap how much of the flow a team can look at (Deloitte).
That last number is the argument in one line. The constraint is not that underwriters judge too slowly; it is that they never see most of the book.
Figure
Most of the submission flow is never evaluated
Manual reentry and disconnected workflows cap how much of the flow a team can look at; the rest goes unread.
Source: Deloitte: many specialty insurers process only a fifth to a third of the submissions they receive.
| Personal / standardized small commercial | Specialty / large-complex accounts | |
|---|---|---|
| Account shape | High volume, repeatable questions, structured data, filed products | Lower volume, bespoke terms, layered or unusual exposures, unstructured broker packs |
| Human role | Algorithms handle routine cases; people manage exceptions | Experienced underwriters interpret risk, negotiate terms, and retain pricing/bind authority |
| Best automation target | Straight-through decisions on simple risks | Make the account decision-ready: ingest, extract, enrich, check appetite, surface precedents, draft the memo |
| Why agents pay | Reduce transaction cost at scale | Return scarce expert hours on every account and let the same team evaluate more of the submission flow |
Which is why triage and multi-agent stacks pay here first
A triage agent stops expert time being spent on incomplete or out-of-appetite submissions. A multi-agent stack runs independent work in parallel: one worker extracts exposures, another checks guidelines, another enriches external data, another compares pricing and portfolio constraints, and an orchestrator assembles the decision-ready view. The underwriter still makes the consequential judgment, and no longer spends the first part of every account moving information between systems.
Figure
A triage stack turns the submission into a decision-ready view
Independent workers run in parallel; the orchestrator assembles; the underwriter's judgment closes the loop.
The measured results follow that shape: AIG reports 2–5x faster underwriting; Markel reports a 113% productivity uplift; Arch cut intake-to-decisioning by 70%; and Hiscox reduced one bounded specialty quote workflow from three days to three minutes. These are carrier- or vendor-reported results rather than universal benchmarks; the source passages are collected on the evidence page.
Figure
Reported results on specialty-shaped work
Carrier- and vendor-reported, not universal benchmarks; they show the shape of the gain, not a promise.
Sources: AIG · Markel · Arch (Ivans) · Hiscox; passages on the evidence page.
Filing freedom is a speed advantage, not a governance exemption
Every major carrier agentic deployment of the past 18 months began in E&S lines, and the reason is structural: surplus-lines carriers do not file rates and forms, so they can iterate at a speed admitted lines cannot match. Speed-to-quote is becoming the visible competitive weapon, and the timeline shows how quickly that gap opened.
The exemption is narrower than it looks. Surplus-lines status exempts a carrier from rate and form filing. It does not exempt it from AI governance: the NAIC model bulletin applies to all licensed insurers, and unfair-discrimination, unfair-trade-practices, and claims-practices law applies to AI-assisted decisions exactly as to human ones. Treat filing freedom as a reason to move first, not as a reason to skip the controls in Governance.
Where this goes next
The phased plan built on this case is the integration phases, whose competitor evidence and gated phases assume specialty economics throughout. The cost side, and what thin claims data does to the credibility of automated model search, is on Costs & value.