The Split: How AI Divides the Insurance Workforce
Insurance job openings fell by half in the year through March 2026, dropping from 281,000 to 138,000 average monthly postings, the lowest level in a decade, Claims Journal reported. The decline isn't driven by a weak economy or falling premiums. "AI automation is absorbing claims processing, back-office operations, and customer-service functions that previously required large teams," the journal noted. Yet a Q1 2026 study by The Jacobson Group and Aon found roughly half of carriers plan to increase headcount over the next year. Automation cuts back-office jobs while tech, claims, and specialty underwriting still hire.
The paradox defines the industry now: traditional roles are contracting while AI-native technical and product roles surge, forcing legacy carriers and startups into a high-stakes battle for hybrid talent — professionals who speak both risk and code.
Global AI spending in insurance reached roughly $15 billion in 2025, with projections clustering around 27 percent annual growth through 2035. North America accounts for nearly half that outlay. Asia-Pacific is accelerating fastest, above 33 percent annually, fueled by digital-first ecosystems in China and India. Europe holds about a quarter, pushed by Solvency II modernization and the EU AI Act's classification of underwriting as high-risk. Software takes more than half of revenue; services grow faster, near 39 percent, as implementation demand intensifies.
Capital flows tell the sharpest story. Carriers poured an estimated $6.8 billion into AI infrastructure upgrades in 2024 alone, Celent says. Zurich committed $1.8 billion over three years. Tokio Marine allocated roughly $1.5 billion in 2025. Travelers' annual tech spend surpasses $1.5 billion. Swiss Re reports insurers now direct 3 to 8 percent of cloud-linked IT budgets to AI development. Venture capital confirms the shift from experimentation to infrastructure: startups raised $40 billion in 2025 for generative AI in insurance. OpenAI secured $6.6 billion and xAI closed a $6 billion agreement in 2024. Wefox obtained a €170 million package for underwriting automation. Liberate closed a $50 million Series B for faster agent platforms.
The talent gap is structural. The U.S. industry faces 400,000 retiring professionals, creating an urgent need for automated backfilling. Accenture estimates automation could touch 71 of every 100 working hours in the UK sector. Lightcast counted 10,000 unique job postings for specific AI insurance skills by mid-2025. Ping An maintains a massive technical edge with 21,000 developers and 3,000 scientists dedicated to R&D, holding more than 53,000 patent applications by September 2024, second globally in generative model filings. Swiss Re holds 634 risk-related patents; State Farm filed 70 new AI patents in Q3 2024; UnitedHealth Group added 10 that quarter. GlobalData tracked over 100 startups deploying new models in 2024 while patent applications reached 259 in Q3. Zurich Insurance now runs 200-plus specific use cases across its commercial division.
The bifurcation is not a future scenario. It is the current operating reality.
Legacy Carriers: Billion-Dollar Bets on a New Workforce
Nationwide's $1.5 billion technology commitment through 2028 is the largest single AI bet any U.S. carrier has placed to date. The Columbus-based Fortune 100 mutual announced the figure in October 2025, framing it as a 20 percent increase over recent annual tech spend. One hundred million dollars per year is earmarked explicitly for AI initiatives; the rest goes to broader modernization. Since 2015 the company has already poured $5 billion into its technology stack: 34 billion annual API transactions, 21 petabytes of data, 4,000 active applications. That foundation makes the new investment executable rather than aspirational.
CEO Kirt Walker said the moment is "the next industrial revolution" and tied the spend to Nationwide's mutual structure, which he argues combines long-term vision with public-company agility. The 2025 results back the confidence: record sales and premiums of $73.2 billion, up 7 percent; net operating income of $4.3 billion, up 37 percent; total adjusted capital of $32.8 billion, the highest in its 100-year history. In July 2025 Nationwide also acquired Allstate Benefits' group health business, expanding its self-funded and stop-loss capabilities for small and mid-sized employers.
The carrier has moved decisively out of "experimentation mode," CTO Fowler said. Six AI initiatives are already scaled; 18 flagship use cases have been identified across businesses. Over 2,000 custom AI agents run in production. Productivity gains of 15 to 30 percent are being measured. Claims Log Notes automatically summarizes thousands of claim log entries each week, freeing adjusters to focus on empathy and judgment rather than documentation. Digital twins of insurance products refine risk modeling and pricing. An AI-enabled developer tool has cut legacy-code migration time in half. A pet-claims automation target aims to handle 80 percent of volume with one-quarter settled instantly. Farm and agribusiness claims review time is expected to drop 20 percent.
Workforce transformation runs in parallel. Roughly half of Nationwide's 22,000 employees use Microsoft Copilot and other AI tools daily; the target is 90 percent by 2026. The company has identified around 1,000 associates who have already made themselves materially more productive with AI. A "Future Work Center" curates learning paths layered with "learning sprints" and "future ready sprints." Engagement scores rose 9 percent in the most recent semi-annual survey. Governance is deliberately rigorous: a Blue Team and Red Team vet every deployment, and an approval process is designed "not to slow things down but to speed things up" by preventing rework.
Global IT spending is projected to surpass $6 trillion in 2026, Gartner says, and 86 percent of technology leaders say they are confident agentic AI will deliver ROI. Nine in ten are developing or rolling out agentic systems. The carrier's mutual structure lets it absorb upfront cost without quarterly-pressure distortion — a luxury publicly traded peers lack. But the signal is clear: the legacy playbook is no longer defensive. It is a capital-intensive offensive built on data infrastructure, governed deployment, and a workforce mandate to make AI everyday infrastructure.
Startups Skip the Line: AI-Native Brokerages and the Hybrid Talent Hunt
While legacy carriers pour billions into retrofitting mainframes, a parallel workforce shift is unfolding at the industry's edges. AI-native startups are not modernizing old workflows — they are skipping them entirely. Fernstone, a Y Combinator Fall 2025 company, operates with five people in New York City and manages risk for more than 500 businesses across construction, security, and the trades. Its pitch: insurance is a $1 trillion category that still runs on pen and paper, and the broker's job — calling, emailing, even faxing carriers for quotes — can be collapsed by an AI agent that tracks email threads, follows up instantly, and issues certificates of insurance in a 15-day average sales cycle. The company has raised $3 million and is now hiring its first dedicated sales representative, a founding SDR role that carries equity upside alongside the mandate to scale from hundreds to thousands of customers.
Corgi represents the carrier-side equivalent. Founded by Emily Yuan and Nico Laqua, the startup secured full carrier licensure in 2025 and closed a $160 million Series B at a $1.3 billion valuation, with capital earmarked for widening coverage, growing distribution, and further developing the AI systems that run underwriting, claims, and policy operations end to end. An earlier $108 million round followed regulatory approval to enter the market. Both companies share a structural advantage: they never hired the armies of adjusters, processors, and middle managers that legacy carriers now need to retrain or displace. Their org charts were born post-LLM, so every role assumes AI fluency as a baseline.
That baseline is precisely where the hiring war concentrates. Fernstone's model pairs AI agents with licensed brokers-in-the-loop — a hybrid role that demands both underwriting judgment and the ability to supervise, prompt, and audit autonomous workflows. Corgi's full-stack approach requires engineers who understand loss ratios and actuaries who can shape model outputs. Neither talent pool exists at scale. Y Combinator alone lists 31 insurance-focused startups in its current directory, each competing for the same narrow band of such hybrid professionals. Legacy carriers' multi-billion-dollar tech commitments only intensify the bid: they can outbid on cash, but the startups offer founding-team equity, zero technical debt, and a product velocity that monolithic IT departments cannot match.
The paradox sharpens here. Startups bypassed the workforce contraction hitting traditional brokerages — no layoffs, no reskilling programs, no union negotiations — yet they face a more acute scarcity. They need people who can build, sell, and regulate AI-native insurance products simultaneously. The market has not produced them. Universities do not teach this intersection. Carriers do not train for it. The only reliable source is poaching from the few peers who have already cracked the hybrid model, which means compensation bands are being set by the most aggressive bidder, not by industry benchmarks.
The New Role Map: What AI-Native Insurance Jobs Look Like
The insurance workforce is splitting in two. Generalist underwriters — the backbone of commercial lines for decades — are in serious decline, with an AI automation risk score of 77 percent. Meanwhile, specialists who can navigate complex commercial risks, emerging categories like cyber, and the models that now price standard policies are commanding higher compensation. The shift is measurable: insurance AI hiring rose 32 percent even as total workforce numbers fell 2.2 percent, industry reports show. Across 100 current job postings from leading U.S. insurers, 72 percent of AI, analytics, data science, product, and technology roles explicitly mention artificial intelligence.
| Emerging Role | Core Responsibility | Key Skills | Automation Exposure |
|---|---|---|---|
| Hybrid underwriter | Blend traditional judgment with model oversight, dashboard interpretation, cross-team collaboration | Analytics interpretation, broker negotiation, model awareness, data literacy | Low — augments rather than replaces |
| AI governance / compliance lead | Oversee model transparency, bias monitoring, override rules, data lineage | Regulatory knowledge, AI ethics, audit protocols, explainability | Low — net new function |
| Automation champion | Drive adoption inside underwriting teams, bridge technical and business users | Process design, change management, low-code tooling, stakeholder communication | Low — creates new workflow |
| Portfolio data specialist | Sit between product, pricing, and underwriting; build market-sensing mechanisms | Real-time monitoring, predictive modeling, portfolio optimization | Medium — transforms analysis |
| Exponential underwriter variants (trailblazer, pioneer, deal-maker, optimizer, detective) | Own digital workflows, design predictive models, explain decisions to brokers, monitor portfolio risk, investigate complex exposures | Technology supervision, data science collaboration, commercial storytelling, risk forensics | Varies by variant |
The hybrid underwriter has become the hiring benchmark. Eliot Partnership said of the profile: someone who uses sound judgement and understands market conditions, but also interprets model outputs, works comfortably with data teams, and sees how decisions affect portfolio results, broker relationships, and customer outcomes. Bain's analysis of roughly 5,000 job postings confirms rising demand for specialty underwriting expertise and broker relationship management — human skills that AI supports but cannot replicate. Negotiation, complex risk evaluation, and regulatory judgment remain stubbornly human.
The modern underwriter is expected to have that same holistic perspective. They also need to work confidently with colleagues from actuarial, engineering and analytics.
Hiring criteria have shifted. Many job descriptions still reflect underwriting as it looked ten or twenty years ago. Current postings prioritize curiosity, learning ability, confidence with data tools, and the ability to handle complex information while staying grounded in fundamentals. Potential now matters as much as experience. Some of the strongest hybrid underwriters come from analytics or multi-team project backgrounds, even with fewer years of underwriting tenure. A clear competency framework (technical capability, sound judgement, commercial understanding, collaboration skills, ability to learn new tools) helps interviewers evaluate consistently.
The skills gap is widening fast. PwC's 2025 Global AI Jobs Barometer found that skill requirements for roles exposed to AI are evolving 66 percent faster than those in other fields. Seventy-four percent of insurance CEOs report concern about digital skills shortages. Fifty-five percent of executives believe underwriting shortages will limit growth in the coming year. Entry-level roles (claims intake, policy processing, data entry) are disappearing, cutting off the traditional apprenticeship path where employees learned the business from the ground up. At one commercial P&C carrier, underwriting assistants now have more end-to-end responsibility and sit in a pipeline to higher-level positions. A large life insurer is re-creating its entry-level experience with simulations, rotations, and AI-assisted learning to bridge the gap.
Carriers are responding with structured upskilling. Underwriting academies, data literacy training, and structured rotations with analytics teams are becoming standard. Internal "automation champions" support adoption inside teams. Leaders map current and future skills, focus training on analytics interpretation and model awareness, update performance measures, and give underwriters time to collaborate with specialist teams. Without this, automation becomes a system upgrade rather than a genuine improvement in how underwriters work.
The salary divergence is already visible. Underwriters who specialize in niche markets or complex risk classes are earning more as their expertise becomes rarer. Generalist roles face compression. For technical talent, the market is fierce, Zero G Talent's board data shows:
| Role Type | Salary Band | Median |
|---|---|---|
| AI specialists at frontier labs | $210k–$550k | $395k |
| Data & platform roles at enterprise infrastructure | $140k–$318k | $250k |
Insurance carriers competing for the same hybrid profiles must calibrate offers accordingly.
Regulation's New Demands: The Compliance Roles That Didn't Exist Two Years Ago
The NAIC's Model Bulletin, adopted December 4, 2023, did not create new law. It clarified that existing insurance statutes (unfair trade practices, unfair discrimination, market conduct) apply with full force when the decision maker is an algorithm. As of July 2026, 25 U.S. jurisdictions have adopted that bulletin. California, Colorado, New York, and Texas have layered on their own insurance-specific AI rules. Colorado's SB26-189, effective January 2027, replaces its prior high-risk AI framework with a narrower automated decision-making technology regime; risk-assessment obligations kicked in January 2026, and cybersecurity audit certifications phase in for larger businesses starting in 2028.
The regulatory architecture is moving from guidance to examination. Twelve states are piloting the NAIC AI Systems Evaluation Tool through September 2026. The Tool gives examiners a structured framework to review insurer AI systems during market conduct exams. A revised version is slated for consideration at the NAIC's 2026 Fall National Meeting. Separately, the PPWG expects to expose a full revised draft of Model 672 for public comment by year end.
What examiners will ask is already documented. Regulators will want to know where AI is used, what personal information supports it, which vendors are involved, how models are tested, how unfair discrimination is assessed, how outputs are reviewed, and how consumers are informed or afforded review when AI affects decisions. The bulletin covers product development, marketing, underwriting, pricing, policy servicing, claims management, and fraud detection. The TPDM Working Group has narrowed its initial framework focus to pricing and underwriting, the highest-impact consumer touchpoints.
"AI is a tool used in underwriting, pricing, claims, fraud detection, and utilization management; it does not alter insurers' legal obligations, and existing state insurance laws apply regardless of whether decisions are made by humans, algorithms, or third-party vendors," the NAIC said in a March 2026 client alert via Crowell & Moring
That sentence is the hiring mandate. Carriers cannot outsource accountability. The registry the NAIC is building is not a licensure regime, but it signals heightened scrutiny of vendor governance, particularly where third-party models and datasets feed underwriting and pricing. Insurers remain on the hook for vendor diligence and management obligations.
The compliance burden is colliding with operational reality. Capgemini's 2026 World Property and Casualty Insurance Report found roughly 60 percent of insurers still at exploration or proof-of-concept stage. Forty-two percent did not use key performance indicators to measure whether their AI investments were working. Seventy-two percent of AI-related spending went to technology; only 28 percent to change management. Poor data quality was the most common barrier to wider AI use. Insurers with mature AI programs shared a pattern: central governance, clear responsibilities for employees, and cross-functional teams working toward shared performance targets.
Those patterns map directly to new roles. Carriers are posting for AI governance leads, model risk officers, algorithmic fairness analysts, and vendor AI compliance managers — titles that barely existed two years ago. The skill set is hybrid: insurance regulatory law, statistical testing, data lineage, and the ability to translate model behavior into examiner-ready documentation. Third-party risk teams are expanding to cover AI vendors specifically. Some carriers are embedding compliance engineers directly into AI product squads, a structure borrowed from fintech.
The White House's national AI policy framework, released in 2026, recommends congressional action to preempt "cumbersome state AI laws" and establish a federal standard. The NAIC opposes preemption, arguing that state regulators have built meaningful supervisory infrastructure (principles, interpretive guidance, examination tools) that a federal override would displace. Industry trade groups have pushed back on the Evaluation Tool, urging revisions to clarify that references to governance and risk-assessment frameworks do not create new regulatory requirements. That tension (state examination tools versus federal preemption pressure) means compliance teams must build for two regimes simultaneously.
For hiring, the implication is clear: the regulatory wall is not a future risk. It is a present staffing requirement. Carriers that treat AI governance as a narrow technology issue miss the broader regulatory exposure. A defensible program must answer basic questions: where AI is used, what data supports it, the vendors involved and under what contracts, whether the tool affects consequential decisions, how outputs are tested, how consumers obtain review, and how AI-related cybersecurity risks are identified and mitigated. Each question is a role, or a fraction of one. The market is pricing that scarcity accordingly.
What Comes Next: Upskilling, Untapped Risk, and the Career Ladder
The insurance industry sits on a staggering amount of unaddressed exposure. Deloitte estimates a US$183 billion global protection gap, a figure that reframes the entire hiring conversation. If AI unlocks even a fraction of that market, the job creation tailwind could dwarf the contraction in traditional roles. The question isn't whether insurance employment grows; it's whether the workforce can evolve fast enough to capture it.
PwC's Insurance 2030 analysis frames the stakes plainly: the stability carriers relied on for predictable risk pricing and consistent growth is disappearing, replaced by a succession of short-term crises. That volatility demands a workforce fluent in digital tools, adaptable by design, and diverse in thinking — exactly the profile Aon identifies as the industry's defining talent challenge. The World Economic Forum's 2025 Future of Jobs Report backs this: a majority of employers now rank skill gaps as the biggest barrier to business transformation, and an even larger share plan to prioritize workforce upskilling by 2030 in response to ongoing disruptions.
RGA's research quantifies the upside for both sides of the employment contract. For professionals, AI training correlates with bigger salary increases, faster promotions, and greater job security. For insurers, it unlocks productivity gains, cuts operational costs, and improves customer satisfaction. For both, it's a hedge against obsolescence in a rapidly digitizing industry. The framing matters: Bevaya.ai argues the industry should stop talking about "job preservation" and start treating automation as a natural catalyst for career evolution, giving employees a clear path to grow alongside AI rather than compete with it.
McKinsey's Claims 2030 work makes this operational. Claims organizations that succeed will take two steps now: define the roles of the future, then grow future-ready talent to fill them. Those that miss the mark will likely be left behind. The same logic applies across underwriting, distribution, and actuarial functions. The carriers and startups winning the talent war today (Fernstone, Corgi, Lemonade, the legacy giants pouring billions into tech) aren't just hiring for current openings. They're building talent pipelines for roles that didn't exist three years ago: AI governance leads, prompt engineers for vertical insurance models, hybrid underwriter-analysts who can validate model output against regulatory requirements.
Market projections vary ($63 billion by 2032, $114 billion by 2031, $176 billion by 2035), but the direction is unanimous. AI in insurance is a compounding growth story. The fault line that runs through every carrier's hiring plan now has a name: the hybrid underwriter. When Kirt Walker called this the next industrial revolution, he wasn't predicting the future — he was describing the reorganization already underway at Nationwide, at Fernstone, at every carrier and startup racing to turn protection gaps into written premium. The pie is expanding faster than any single company can eat it. The winners will be the organizations that treat upskilling as continuous capital expenditure, not a one-time training budget, and the professionals who treat AI fluency as a career-long discipline, not a certificate.
Working in AI? Zero G Talent tracks the openings: see every open Databricks role, browse AI jobs, openings at Anthropic, and the people building the field.