Skip to sources
Time Machine

Insurance Underwriters

Scrub through 336years of this role's history, from when it first emerged, through every wave of technology that reshaped it, to the cited projections for where it's heading next.

2026drag to travel through time
1700172517501775180018251850187519001925195019752000now
Country
2026
Known today as Insurance Underwriters (BLS SOC 13-2053)
Latest actual · 2024
127K
BLS OEWS May 2024 employment for insurance underwriters. Sourced from O*NET which reflects the same BLS establishment-survey figure. Employment grew from the 2014 trough of 104,942 to a peak around 2021-2022, driven by specialty commercial lines expansion, the emergence of cyber insurance as a major product, and strong demand for complex-commercial underwriters who can work alongside AI tools rather than being replaced by them. AI-powered straight-through processing handles 60-80% of personal-lines submissions without human review (Insurance Journal 2025), so future headcount contraction is projected to concentrate in routine processing roles rather than specialty underwriting.
Latest actual · 2024
$79,880
Source: BLS-OEWS
Each dot is a cited figure over time; the dotted line only links them (values between aren't measured). Hollow dots are estimates.
Tools of the era

The tools that defined the work

Select an era to see how it reshaped the work.

  • Blackboard ledger + physical inspection (Lloyd's market and early American fire underwriting era)

    The first insurance underwriters had no tools beyond pen, paper, and personal knowledge. At Lloyd's Coffee House the risks were literally written on blackboards: cargo manifest, destination port, route, season, captain's reputation. The underwriter assessed the risk entirely from personal experience of maritime trade, weather patterns, and the trustworthiness of the people involved. In American fire insurance, early underwriters would walk the premises being insured, inspecting building construction, proximity to neighbors, and the availability of water. The Friendly Society's "Hand in Hand" mark -- a plaque placed on buildings it insured -- was both a marketing device and a practical tool: it told fire brigades (which were often insurance-company brigades) which buildings to prioritize saving.

    Ledger workPaper recordkeeping
  • Actuarial tables + mortality schedules + medical examination (life insurance scientific era)

    The mid-19th century brought the first scientific tools to underwriting. Edmund Halley's 1693 survival table was the conceptual foundation, but it was James Dodson's 1755 demonstration that age-based premiums could make life insurance viable for any applicant that transformed the practice. The Equitable Life Assurance Society, founded in London in 1762, was the first insurer to systematically use mortality tables to price policies. By the mid-1800s, American life insurers were employing company physicians to examine all applicants -- a physician's letter of assessment was required for any policy above a minimal amount. The underwriter worked from two tools: the actuarial table (which set the premium range for a given age and health class) and the physician's report (which placed the applicant into a class). Fire underwriting developed parallel rate-manual tools: large rate manuals produced by organizations such as the National Board of Fire Underwriters (1866) specified the premium for any building type, construction quality, and occupancy class.

    Effect on the work

    Actuarial tables expanded the underwriting market enormously: any literate clerk could look up a rate once the tables were established, reducing the cognitive barrier to entry. This drove the creation of large insurance company workforces where underwriting clerks applied standardized rules to a high volume of applications rather than exercising individual judgment on each one.

    Work toolChanging equipment
  • IBM punched-card tabulation + paper rating manuals (clerical processing era)

    Metropolitan Life Insurance was among the first carriers to adopt IBM punched-card tabulating equipment -- renting IBM machines for $225,000 per year for actuarial and policy-record applications. By the 1940s, large insurers ran their policy records on tabulating machines: premium billing, lapse tracking, and mortality statistics were among the first functions mechanized. For the underwriter, the key shift was organizational: the volume of applications processable by a carrier exploded, requiring large clerical underwriting departments to handle intake, rating, and policy issuance. Travelers Insurance installed one of the first IBM mainframes in the insurance industry in the early 1960s, automating batch rating for personal auto and homeowners policies. Underwriters worked from thick paper rate manuals and applied the tables manually, with tabulating machines handling the record-keeping and billing rather than the underwriting decision itself.

    Effect on the work

    IBM tabulation and early mainframe computing expanded the insurance market rather than contracting the underwriting workforce. More policies could be processed per carrier, driving employment growth rather than displacement. The workforce grew from roughly 35,000 in the 1920s to an estimated 55,000 by the early 1940s and substantially more by the 1960s.

    Work toolChanging equipment
  • Mainframe rating systems + personal computers + early automation (digital rating era)

    The 1970s and 1980s brought mainframe-based rating systems that automated the mechanical application of rate tables: a clerk could enter a vehicle year, make, model, driver age, and ZIP code and receive a computed premium in seconds, replacing the manual table lookup that had dominated for decades. Applied Systems, founded in the early 1980s, was one of the first companies to build dedicated software for insurance agency management and rating -- their tools spread from large carriers down to independent agents by the late 1980s. For life underwriting, the 1980s AIDS crisis drove a significant expansion in medical testing requirements, adding new data inputs to the underwriting process. Preferred-rate programs emerged as carriers recognized that healthy applicants deserved lower premiums, requiring underwriters to segment risk more finely. Personal computers arrived in insurance company underwriting departments in the mid-1980s, beginning the shift from batch mainframe processing to interactive desktop rating.

    Effect on the work

    Mainframe rating automation did not significantly reduce underwriting headcount in this era -- instead, it expanded the volume of business each underwriter could process, supporting premium growth without proportional headcount growth. The workforce remained above 160,000 through 1990, then began a slow decline as productivity gains outpaced premium volume growth.

    Mainframe processingComputerized records
  • Rules-based expert systems + early automated underwriting (decision-engine era)

    The mid-1990s saw the first wave of technology that genuinely displaced underwriting labor rather than amplifying it. Rules-based expert systems -- software encoding underwriting guidelines as if-then decision trees -- began handling personal auto and homeowners applications automatically. If an applicant met the carrier's filed guidelines, the system issued the policy without human review. If it fell outside, it was routed to a human. The Y2K crisis of the late 1990s had an ironic effect: carriers faced with replacing legacy systems largely chose to patch rather than rebuild, delaying the modernization of underwriting platforms but also leaving early expert systems in place longer than intended. By 2000, carriers processing personal lines at scale reported that a significant share of clean applications were being issued without human touch -- the first true straight-through processing. For individual life insurance, simplified-issue products that required no medical exam (only a questionnaire and a prescription database check) began replacing full medical underwriting for policies below $500,000.

    Effect on the work

    This era produced the first measurable underwriting headcount decline. From roughly 162,000 in 1990, insurance underwriter employment fell to an estimated 147,000 by 2000 and continued declining through the 2000s. Each new generation of automated underwriting software extended the share of submissions that could be decided without human judgment -- concentrating remaining human underwriting on the complex and non-standard risks that rule engines could not handle.

    Work toolChanging equipment
  • Straight-through processing + predictive analytics + insurtech platforms (STP era)

    The 2010s delivered the second, larger wave of underwriting automation. Cloud computing, mobile channels, and modern APIs made it possible to build insurance products that could originate, rate, and bind entirely without human involvement. Guidewire and Duck Creek, founded in the early 2000s, became the dominant policy-administration platforms for mid-size carriers; they incorporated automated underwriting engines as first-class features. Cape Analytics (founded 2014) used aerial and satellite imagery to score property risks at the address level, replacing the physical inspection for standard commercial and personal property accounts. Cytora (founded 2016) automated the triage of broker submissions for commercial lines, routing in-appetite risks to straight-through processing and flagging exceptions for human review. By the late 2010s, personal lines carriers were reporting 70-80% straight-through processing rates for clean submissions. The workforce declined from the 2000 level through 2014 (reaching 104,942) before rebounding as specialty commercial lines -- particularly cyber insurance, which emerged as a major product after the 2013-2015 breach wave -- required large numbers of experienced human underwriters that no automated system could replace.

    Effect on the work

    Employment reached its post-2000 trough at 104,942 in 2014, then grew steadily as specialty lines expansion offset the continued automation of personal-lines processing. The net effect was a workforce bifurcation: carriers reducing personal-lines headcount while simultaneously hiring for commercial, cyber, and specialty roles.

    Work toolChanging equipment
  • AI-powered underwriting platforms + large language models (agentic AI era)

    Generative AI entered insurance underwriting from two directions simultaneously. Large language models (ChatGPT, Claude) gave underwriters tools to parse complex submission documents -- lengthy engineering surveys, D&O prospectuses, contractor qualification packages -- at a fraction of the previous time. Akur8 deployed machine learning pricing engines that could recalibrate rate models from live loss experience in hours rather than weeks. Hyperscience's intelligent document processing automated the extraction of structured data from unstructured broker submissions. By 2025, McKinsey was projecting that 90% of pricing and underwriting tasks for personal and small-business insurance would be automated by 2030, with AI-driven adoption rates rising from 14% of carriers in 2024 toward 70% by 2028. Agentic AI architectures -- where specialized agents handle intake, risk profiling, pricing, compliance review, and decision orchestration as an autonomous pipeline -- were being piloted at leading carriers by 2025-26. The surviving human underwriter role concentrates on exactly what these systems cannot do: bespoke broker negotiation, coverage-term judgment for novel risks, regulatory filing expertise across 50 state jurisdictions, and the relationship capital that determines whether a specialty broker routes their best business to your carrier or a competitor.

    Effect on the work

    BLS projects a 3% decline in insurance underwriter employment from 2024 to 2034 (from 127,000 to approximately 123,700), a much smaller decline than Frey and Osborne's 1999% computerization probability predicted. The occupation is proving more resilient than headline automation projections suggested, but the composition is shifting: routine-processing roles are declining while specialty and complex-commercial roles are growing, requiring more education, more experience, and materially better judgment.

    Work toolChanging equipment
Projection cone · present → 2034

What credible sources project

Scrub the slider past now to anchor each scenario on the scrubber. The spread is the range of futures credible sources project for this role.

Employment outlook
Projected change in the number of people doing this work.
BLS National Employment Matrix 2024-34
2034
-3%
BLS Employment Projections industry-occupation matrix, 2024-34 cycle. Projects -3% employment change for 13-2053 -- from 127,000 (2024) to approximately 123,700 (2034), a loss of roughly 3,300 positions. BLS classifies this as a declining occupation against an all-occupations average of +4%. The BLS methodology models continued expansion of automated underwriting software in personal and small-commercial lines as the primary headwind, with specialty commercial demand as a partial offset. About 8,200 openings per year are projected due to retirements and turnover, so the occupation remains active even while shrinking.
AI task exposure
Share of the role’s tasks that researchers estimate AI can do. This is a measure of task exposure, not a forecast of jobs lost.
Frey and Osborne (2013) -- "The Future of Employment"
2033
99%
of tasks
Gaussian-process classifier on O*NET task features. Frey and Osborne placed insurance underwriters at 99% probability of computerization -- the highest risk category in their 702-occupation study. The bottleneck analysis found no meaningful barriers in perception and manipulation, creative intelligence, or social intelligence for a generalist underwriting role: rules can be encoded, ratings can be computed, and basic coverage decisions can be made algorithmically. This was a reasonable assessment of the tasks; what it did not model was the occupation's bifurcation response. The 99% figure represents exposure as a ceiling on the routine-processing segment; it has substantially materialized in personal lines STP. Actual employment from 2013 to 2021 grew 16% -- because specialty underwriting, which F&O did not separately model, expanded faster than personal-lines automation contracted.
McKinsey Global Institute -- insurance automation scenario (2025)
2030
90%
of tasks
McKinsey projects that by 2030, more than 90% of pricing and underwriting tasks for personal and small-business insurance will be fully automated. This is a task-exposure figure for the routine processing segment of the occupation, not a projection of total headcount loss. Specialty, complex-commercial, and large-account underwriting is explicitly excluded from the 90% scenario. McKinsey also projects agentic AI adoption in insurance rising from roughly 14% of carriers in 2024 to 70% by 2028, driven by autonomous underwriting pipelines that can handle intake, risk profiling, pricing, compliance review, and policy issuance without human involvement for qualifying risks.
Eloundou et al. -- "GPTs are GPTs" (2023/2024)
2028
55%
of tasks
GPT-4 task-by-task LLM exposure labeling on O*NET tasks. Insurance underwriters score in the high range for LLM exposure because the dominant tasks -- analyzing application information, evaluating risk potential, reviewing coverage recommendations, drafting policy documents -- are language-intensive and information-processing tasks where large language models have direct capability. The Eloundou measure captures LLM exposure specifically (as opposed to general automation from rules engines or computer vision), and it is additive to the existing automation from expert systems and STP platforms. The 55% figure represents the share of underwriting tasks with substantial LLM exposure, weighted by importance -- higher than the BLS projection implies, because LLMs could accelerate the automation of complex-document tasks that rules engines could not previously handle.
Today, in this role

What's shifting in the work right now

The historical view above shows how this role has moved. This is the present-day detail: which AI tools are picking up which tasks, where the edge still is, and the natural directions this work can grow.

What's changing in your day

Three parts of your work where AI is already doing real lifting, and what stays yours.

AI is sitting alongside you hereProcess and extract risk data from broker submissions, applications, and supplemental documents (loss runs, financial statements, engineering reports) using Hyperscience intelligent document processing

Process and extract risk data from broker submissions, applications, and supplemental documents (loss runs, financial statements, engineering reports) using Hyperscience intelligent document processing; validate AI-extracted fields for completeness and accuracy before routing to pricing and coverage analysis workflows.[7]

Where your edge is

Focus quality-control effort on the highest-impact fields (limits, deductibles, schedule-of-values accuracy, prior-loss pattern) where AI extraction errors have largest underwriting consequence; develop systematic spot-check protocols rather than reviewing 100% of output, freeing time for coverage analysis.

AI is sitting alongside you hereReview and validate AI-generated property risk assessments: for new commercial property submissions, interpret Cape Analytics aerial-imagery risk scores (roof condition, exposure class, proximity hazards) alongside physical inspection reports

Review and validate AI-generated property risk assessments: for new commercial property submissions, interpret Cape Analytics aerial-imagery risk scores (roof condition, exposure class, proximity hazards) alongside physical inspection reports; apply local market knowledge and coverage-specific adjustments before binding or quoting.[8]

Tools picking this up
Where your edge is

Develop deep knowledge of the local building stock, regional catastrophe exposure, and loss history patterns that aerial imagery scores cannot fully capture; build the skill to identify when Cape Analytics scores require override because of recent renovations, protective systems, or misclassified construction types.

AI is sitting alongside you hereReview AI-generated auto damage assessments from Tractable for commercial fleet or personal auto submissions: confirm AI damage estimates align with loss run history and vehicle schedules

Review AI-generated auto damage assessments from Tractable for commercial fleet or personal auto submissions: confirm AI damage estimates align with loss run history and vehicle schedules; use outputs to adjust experience-rating modifications and set appropriate deductible structures for fleet accounts.[9]

Tools picking this up
Where your edge is

Develop fleet underwriting specialization — large commercial auto accounts involve complex driver profiles, vehicle schedules, and safety programs that require experienced judgment beyond damage-photo analysis. Tractable accelerates data intake; your value lies in the exposure analysis and program structure.

Where this role is heading

Natural next steps for someone with your foundation: not exits, evolutions.

A direction you could grow

Financial Managers

Senior underwriters with management experience and a specialty book of business are well-positioned for Underwriting Manager or VP-Underwriting roles — formal Financial Manager positions that oversee the carrier's underwriting function, set appetite guidelines, manage the underwriting team, and own the portfolio's profitability. As AI absorbs routine submission processing, carriers are investing in experienced leaders who can govern AI-assisted workflows, maintain broker relationships at the senior level, and set strategic appetite direction. The transition leverages existing expertise without requiring a credential-heavy pivot.

What you'd add
What it takesSome new skills to pick up
Share this year
Drops anyone you send it to straight into 2026.
Preview card
Part of Business & Finance · see all 32roles →
Different role?

See the same long-arc view for your own profession.

Browse the directory by industry, or search by title or SOC code. New roles ship every few weeks. Every profile cites every claim.

Browse all roles

The data behind this timeline

On record since1700
Latest tracked employment127,000 (US, 2024)
Latest median pay$79,880 (2024)
Outlook-3% by 2034 (BLS National Employment Matrix 2024-34)
View all 28 cited data points
YearUS employmentMedian annual paySource
18908,500n/aESTIMATE
192035,000n/aESTIMATE
194355,000n/aESTIMATE
1961n/a$5,500BLS-HISTORICAL-BULLETIN
1990162,000$28,000ESTIMATE
2000147,000$43,000BLS-OEWS
200396,890$47,330BLS-OEWS
200496,110$48,550BLS-OEWS
200598,970$51,270BLS-OEWS
200699,430$52,350BLS-OEWS
200798,920$54,530BLS-OEWS
200898,690$56,790BLS-OEWS
200998,430$57,820BLS-OEWS
201095,350$59,290BLS-OEWS
201192,840$60,830BLS-OEWS
201291,810$62,870BLS-OEWS
201392,540$63,780BLS-OEWS
2014104,942$64,220BLS-OEWS
201589,960$65,040BLS-OEWS
201691,650$67,680BLS-OEWS
201789,910$69,760BLS-OEWS
201896,040$69,380BLS-OEWS
2019100,050$70,020BLS-OEWS
2020101,790$71,790BLS-OEWS
2021107,690$76,390BLS-OEWS
2022105,900$76,230BLS-OEWS
2023101,310$77,860BLS-OEWS
2024127,000$79,880BLS-OEWS
Embed this timeline on your site

Free for any site. Paste this where the timeline should appear; it stays interactive, every datapoint stays cited, and it sets no cookies on your page. How embedding works

<iframe src="https://futurehistory.earth/embed/13-2053"
  width="100%" height="430" style="border:0"
  title="Insurance Underwriters, a Future History timeline"
  loading="lazy"></iframe>

See all roles in Business & Finance