Claims adjusters

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

• Physical inspection + paper ledgers1752–1900

  Physical inspection + paper ledgers (fire insurance era)

  The Philadelphia Contributionship's founding innovation was the pre-coverage inspection, sending a surveyor to examine a building before agreeing to insure it. When a fire did occur, the same network of directors and agents who had inspected the property initially was responsible for evaluating the loss. There was no distinct "adjuster" profession yet. Claims were settled through negotiation between the insurer's representative and the policyholder, with the representative's judgment shaped by the policy language, the physical evidence of damage, and whatever witness accounts were available. The tools were the same as any 18th-century commerce: quill, ledger, and the testimony of men who had been at the scene.

• Independent adjuster profession + NAIC…1871–1940

  Independent adjuster profession + NAIC model laws (post-Chicago Fire era)

  The Great Chicago Fire of October 1871 destroyed approximately 17,500 buildings and caused roughly $222 million in property damage: enough to bankrupt or severely impair many insurance companies. The insurers that survived discovered that their existing staff could not assess and close the volume of claims. For the first time, they hired independent professionals whose sole purpose was to travel to the loss site, measure the damage against the policy, negotiate a settlement, and close the file. The modern claims adjuster was born from catastrophe triage. Simultaneously, the NAIC, initially called the National Insurance Convention, was founded in 1871 directly in response to the insolvencies the fire exposed. Its first session adopted a uniform annual statement for insurer financial reporting and established reserve-deposit requirements. State insurance departments multiplied; licensing of adjusters became a regulatory question. By the early 20th century, most US states required some form of adjuster licensure. Workers' compensation laws (passed in Wisconsin and eight other states in 1911, spreading to many states by 1920) created an entirely new claims category: workplace injury, requiring medical documentation review, rehabilitation coordination, and disability assessment alongside the physical loss investigation that property adjusters had always done.

• Automobile era + post-war health insura…1940–1970

  Automobile era + post-war health insurance: mass-market claims

  Two structural shifts in the 1940s-1960s transformed the volume and character of claims adjusting. First, the post-war automobile boom made auto accident claims the largest single category of property-casualty losses. A fender-bender in 1935 was unusual; by 1960, the US had tens of millions of registered vehicles and tens of millions of auto claims per year. The auto adjuster (skilled in vehicle valuation, repair cost estimation, and liability determination) became the modal claims professional. Second, employer-sponsored health insurance expanded dramatically through 1945-1960 collective bargaining agreements (stimulated by the 1942 wage freeze that made benefits a substitute for cash compensation). By 1960, a large share of Americans had some employer-provided health coverage, creating an industrial-scale health claims operation at every major insurer. Health claims adjusting was more clerical and less investigative than property adjusting, a paper-matching process, but it was a major new employment category. The 1968 National Flood Insurance Program (NFIP) added another stream: federally backed flood claims that required licensed adjusters certified under FEMA standards.

• Mainframe claims systems + CCC Computing1970–1995

  Mainframe claims systems + CCC Computing (auto estimating)

  The 1970s introduced mainframe-based claims management systems at major insurers. These were not automation in the modern sense, human adjusters still made every decision, but they replaced paper-based tracking of claim status, reserve levels, and payment history with centralized databases. The operational impact was real: a claims supervisor who once managed files through a physical filing cabinet could now oversee fifty adjusters through a CRT terminal. For auto physical damage specifically, the arrival of computerized estimating systems changed the adjuster's daily workflow more substantially. CCC Information Services (founded 1980, later renamed CCC Intelligent Solutions) introduced the first computerized auto-damage estimating system, which standardized labor time and parts pricing across thousands of auto body shops. Instead of negotiating repair costs from scratch on each claim, an adjuster could produce an industry-standard estimate in minutes. Mitchell International (founded 1946) was a parallel provider. By the late 1980s, any adjuster who did not use a computer-based estimating system was at a professional disadvantage. The tool had not replaced the adjuster: it had become their most essential instrument.

• HIPAA EDI mandates + digital photo-esti…1996–2010

  HIPAA EDI mandates + digital photo-estimation + internet claim filing

  The Health Insurance Portability and Accountability Act of 1996 (HIPAA) mandated electronic data interchange (EDI) standards for health insurance claims: the first federal requirement for electronic claims submission at scale. By 2003, when the final HIPAA transaction rules took effect, health insurance claim processing had become largely automated for straightforward medical claims. The claims examiner's job in health insurance shifted from manually reviewing paper explanation-of-benefits forms to exception-handling the cases that automated adjudication flagged for human review. For every claim that used to require a human decision, ten now flowed through without one. In property and casualty insurance, digital cameras and then smartphones enabled photo-based damage documentation for the first time. An insurer could now ask a policyholder to photograph the damage and email the images rather than waiting for an adjuster to drive to the site. Early online claim filing (major P&C insurers began offering internet-based FNOL, First Notice of Loss, filing in the 2000s) reduced the administrative layer of the adjuster's work. Hurricane Katrina (2005) and Hurricane Ike (2008) demonstrated that the adjuster workforce could not surge fast enough to handle catastrophic event claims, driving accelerated investment in virtual inspection tools.

• Drone inspection + predictive analytics…2010–2015

  Drone inspection + predictive analytics + Guidewire ClaimCenter

  The 2010s brought three converging innovations. First, commercial drone inspection entered insurance following FAA exemptions for commercial drone use, by 2014, State Farm, USAA, and Allstate were piloting drone roof inspections for property claims, replacing the physical adjuster climb (and the associated safety risk and scheduling delay). A drone inspection could assess a storm-damaged roof in minutes. Second, cloud-based claims management platforms (Guidewire ClaimCenter, Majesco, Duck Creek) replaced the aging mainframe systems at mid-size and large insurers, enabling mobile-first workflows where adjusters could document a site inspection on an iPad and have the data instantly in the central claims system. Third, predictive analytics software began scoring claims for fraud risk, litigation propensity, and settlement value on intake, allowing claims managers to route the lowest-risk, most-routine claims toward fast-track settlement without full adjuster review.

• Lemonade founding2015–2020

  Lemonade founding (2015) + Tractable AI (2014): first end-to-end AI claims

  April 2015: Daniel Schreiber and Shai Wininger, tech entrepreneurs with no insurance background, founded Lemonade with the explicit thesis that insurance claims could be handled by AI agents, not human adjusters. The company received its New York insurance license in 2016 and launched to consumers later that year. Its claims model was built around two AI agents: Maya (handles policy setup and coverage questions) and Jim (processes and pays claims). For a qualifying renters or homeowners claim, the entire interaction could happen in seconds: the policyholder records a brief explanation on their phone, Jim cross-references the claim against the policy, runs anti-fraud checks, and initiates payment. No human adjuster reviews the file unless Jim flags an anomaly. In the same period, Tractable AI (founded 2014 in London by Alexandre Dalyac) was developing computer-vision models to assess auto-body damage from photographs. By 2017, Tractable was working with GEICO and other major US insurers, enabling digital claims assessment: the policyholder photographs the damaged vehicle, Tractable's model estimates the repair cost, and the insurer issues a payment or repair authorization without a human appraiser visiting the scene. In June 2021, Tractable raised funding that valued it at $1 billion (unicorn status). By 2023, SoftBank had invested an additional $65 million.

• Generative AI claim adjustment2020–present

  Generative AI claim adjustment: autonomous FNOL, LLM-assisted investigation

  Lemonade's IPO on July 1, 2020 marked the public market's first judgment on the AI-insurance model: shares opened against a $29 IPO price, valuing the company at roughly $3.8 billion. The listing gave the AI-claims model a benchmark and attracted imitators. COVID-19 simultaneously accelerated virtual-inspection adoption across the traditional insurance industry: adjusters who could not travel to flooded or fire-damaged properties due to lockdowns were replaced, permanently in many cases, by drone inspection, satellite imagery assessment, and policyholder self-documentation. By 2023, generative AI was entering the claims workflow at traditional insurers. Large language models could draft demand letters, summarize medical records, generate coverage analysis, and identify subrogation opportunities: tasks that previously required paralegal-level reading skill and hours of document review. At the same time, AI-powered fraud detection became sophisticated enough to flag suspicious claims patterns across large books of business without a Special Investigations Unit (SIU) investigator manually reviewing each file. The Lemonade model's 2024 scale: the company reported 2.9 million customers and $527 million in revenue. It continues to process a substantial share of its simpler claims without human adjuster involvement. For traditional insurers, the question was no longer whether to deploy AI in claims: it was how fast.

  Effect on the work

  BLS projects thousands fewer Claims Adjuster, Examiner, and Investigator positions by 2034: a decline from the 356,100 baseline. This is one of the steeper projected declines among major white-collar occupations in the BLS 2024-34 dataset. The BLS notes technology explicitly as a driver: "Computers and other technology are making it easier for adjusters to process information quickly, which makes them more productive." Productivity gains reduce the number of adjusters needed to handle a given volume of claims even as insured losses may grow.