Anthropic says agentic AI is pushing insurers to prioritize data access over legacy UX

• Key Insight: See why agentic AI is forcing insurtechs to prioritize data accessibility over legacy UX.
• Forward Look: Brace for AI to create new insurable risks for brokers and carriers.

Anthropic, the company behind Claude, has a front-row seat to the disruption that generative AI is causing to the insurance space. But it's not the only one bearing witness to a fundamental change in this market.

Insurtechs are challenged to bring a new value proposition in a world where

Read Digital Insurance's recent coverage on this topic and more:

AI coding tools shift build-vs-buy calculus for insurers

Agentic AI is redrawing the build-versus-buy line that has defined insurer-vendor relationships for decades, according to Anthropic applied AI architect Eoghan Scully. Carriers with large engineering teams are increasingly finding it economically viable to replicate SaaS functionality in-house, while others are layering agentic tools on top of existing vendor systems. The likely outcome for most carriers: using AI agents as an interface into vendor platforms, preserving those systems as data and records infrastructure. That framing gives insurtechs a clear strategic foothold — prioritizing data accessibility for AI agents over legacy UX. Still, 77% of financial services leaders report no measurable ROI from AI investments, per PwC, signaling the shift remains early-stage.Read more: Anthropic: Insurtechs will see AI recast their vendor role

AI boosts demand for human advisors, NY Life survey finds

New York Life's Wealth Watch survey of 2,278 adults found AI is reinforcing — not replacing — demand for human financial advisors. Twenty-seven percent of respondents said AI access increases their need for an advisor, rising to 40% among Gen Z. Only 24% of Americans are comfortable using AI for financial guidance, and 62% say AI tools don't understand their personal situations well. Carriers and distributors can capitalize on this dynamic by positioning advisors as AI-informed partners: 76% of respondents said they're comfortable with their financial professional using AI, particularly for administrative and organizational tasks.Read more: New York Life: AI use is fueling demand for advisors

How one insurer turns ChatGPT into a submissions channel

Employers Holdings launched a ChatGPT app in April that generates workers' comp submissions and indicative quotes from just six data points, leveraging APIs the company built in 2017. Daily submissions through the channel are growing, though executives say the goal is lead facilitation, not direct replacement of agents. The integration required minimal new development because existing API infrastructure connected directly to OpenAI. Carriers without that groundwork face a harder path — legacy system compatibility, not the APIs themselves, is the primary obstacle. Meanwhile, 29% of auto and home insurance customers already use AI for research, and consumer confidence in AI for insurance fell to 40% this year from 46%, underscoring agents' continued relevance.Read more: How Employers finds new business from within ChatGPT

AI mirrors internet era for agents, expert says

The AI-driven disruption feared by insurance agents mirrors the anxiety surrounding the internet's rise in the late 1990s — a transition that ultimately expanded agent opportunity rather than eliminated it. Robert Hartwig, director of the Risk and Uncertainty Management Center at the University of South Carolina, told a Travelers webcast audience that agencies positioning themselves to adopt AI tools stand to gain efficiency and improve client experience. Back-office automation will offload routine tasks, but the agent-client relationship remains central. Hartwig also noted that AI creates new insurable risks — and new revenue opportunities for brokers and carriers.Read more: AI won't wipe out insurance agents, risk expert says

Explainability gaps cost insurers AI ROI, SAS study finds

Embedding explainability and human oversight into AI workflows is now a competitive differentiator for insurers, according to a SAS/IDC study. Among insurance AI leaders, 59% report strong or high ROI on AI investments; laggards report none. The sector scored lowest on data quality and model governance (60%) and highest on responsible AI policy (70%). Insurers' early investment in governance frameworks positions them well, but scaling deployment without matching technical controls risks eroding both accuracy and consumer confidence. Agentic AI compounds the stakes: SAS CTO Bryan Harris notes error rates on complex tasks can exceed 25%, making domain-embedded oversight essential.Read more: Where insurance professionals lose trust in AI: SAS

NAIC AI evaluation tool targets claims approvals, model bias

The NAIC's AI Systems Evaluation Tool, piloted since March, is on track for potential adoption at the commissioners' November meeting — giving insurers a narrow window to assess their AI governance posture. The tool scrutinizes not just denied claims but also approved ones, as regulators flag over-approval as a solvency risk that distorts pricing. It also examines AI model bias, with regulators concerned that opaque, non-actuarial models could produce discriminatory outcomes undetected. Compliance officers should document AI system oversight, vendor relationships and drift-monitoring protocols now. Startups face particular pressure to demonstrate competency to regulators unfamiliar with non-traditional modeling approaches.Read more: NAIC to make insurers show AI use in claims and models

AI liability gaps widen as silent risks challenge coverage limits

Silent AI — AI embedded in existing products and processes without explicit policy recognition — is creating coverage gaps that underwriters, legal experts and executives are only beginning to quantify. Existing policies, including D&O and commercial financial institution coverage, are increasingly excluding AI-related losses, while dedicated AI liability products from Lloyd's syndicates and U.S. MGAs carry modest limits and hard aggregate caps that may fall short of actual exposure. As agentic AI shifts liability from outputs to autonomous actions, underwriters must map insureds' AI use cases — distinguishing base model providers from deployers — to accurately assess liability profiles. Significantly more primary and reinsurance capacity will be needed as AI scales into critical business processes.Read more: Why insurers are split on covering 'silent' AI risks

Only 10% of life insurers can match advisors to buyers, report finds

Half of consumers are comfortable using AI to research life insurance, but 85% want human advisor interaction at some point in the buying process — and half prefer advisors who share their demographic background. Fewer than 25% of carriers can facilitate that kind of matching. Meanwhile, one in four consumers abandon the purchase process due to technical jargon, affordability concerns and perceived irrelevance. Closing that gap requires building data infrastructure capable of turning customer intelligence into personalized, proactive engagement. Only 10% of carriers currently meet that standard, according to the World Life Insurance Report 2027 from Capgemini Research Institute and LIMRA, which surveyed 6,175 consumers across 18 countries.Read more: AI can prep buyers, but life insurers still miss advisor matches

Insurtechs: Credibility can't be 'growth hacked'

Flashy demos and rapid growth strategies won't substitute for underwriting results and earned trust, insurtech leaders warn. Andrew Engler of RockRose Risk advises guarding track records "religiously," noting that even advanced technology struggles to offset carrier losses or misuse of capacity. Sean Eldridge of Crosstie cautions against confusing novelty with value — buyers need solutions that integrate with legacy systems, meet compliance requirements and produce measurable results post-implementation. Paul Templar of VIPR Solutions adds that SOC 2 compliance and enterprise-grade operational infrastructure become non-negotiable when pursuing major clients. The consistent theme: deep specialization and customer focus outperform trend-chasing.Read more: Insurtechs: Credibility can't be 'growth hacked'

Cyber coverage gaps put 30M microbusinesses at risk

Nearly 30 million U.S. businesses operate without employees, yet their cyber exposures increasingly mirror those of large enterprises. Electronic funds transfer and social engineering losses frequently exceed $25,000 — often beyond what coverage sublimits address. Traditional homeowners and home-business endorsements are failing to keep pace with third-party platform risks from cloud outages, API failures and AI-enabled fraud. Carriers serving this segment need to treat cyber as foundational coverage rather than an add-on, proactively transition growing home-based operations to standalone commercial policies and expand incident response beyond financial reimbursement to include forensic support, extortion negotiation and streamlined claims reporting.Read more: Microbusinesses challenge insurers to rethink cyber insurance

Tech, underwriting roles hardest to fill as hiring plans rise

Nearly half of insurers (49%) plan to increase headcount over the next 12 months, but the Jacobson Group and Aon's Semi-Annual U.S. Insurance Labor Market Study signals that much of this hiring reflects backfilling rather than growth. Actuarial, technology and executive positions remain the hardest to recruit for — complicated further by low voluntary turnover, which shrinks the pool of actively searching candidates. Eighteen percent of companies report hiring has grown more difficult year over year. With automation driving staff reductions at 11% of firms, carriers expanding their talent pipelines need strategies that reach passive candidates, particularly in technical and leadership functions.Read more: Insurance needs to hire tech and underwriting roles

This roundup was created with AI assistance. A Digital Insurance editor reviewed each item before publication. Introductory bullet points created by AI with editorial review.