Quarter covered: January – March 2026 · Global scope · Published March 2026
Something is shifting in Q1 2026 in the world of insurance with regard to AI adoption, as is the case with many industries. The conversation about AI globally has stopped being about whether to adopt, and started being about what you are willing to restructure to make it work. Less than three months into the new year, insurers have announced real agentic deployments, broker stocks have moved on ChatGPT news, and hiring has been paused in many geographies as decision makers wait to see what AI would replace. The pilot era is over. This issue covers three fronts where the action is happening: claims automation, underwriting transformation, and the slow erosion of traditional distribution.
Agentic claims handling moves from pilot to product
Travelers USA's February launch of a fully voice-based AI claims agent — built with OpenAI — is the most concrete signal yet that 2026 belongs to agentic AI.
In February, Travelers announced its AI Claim Assistant: an autonomous voice service that handles inbound auto damage claims end to end — gathering information, answering policy questions, helping customers decide whether to file, triggering photo uploads, scheduling repairs, and reserving rental cars, all without a human on the line. The service was developed with OpenAI, and early feedback was described as overwhelmingly positive. The carrier has signalled it will extend the capability to other lines of business.
This is not a chatbot. It is an autonomous agent executing a workflow end to end. The distinction matters because it changes the economics: the question is no longer how many call centre staff you need to handle volume, but which claims interactions require a human at all.
Meanwhile, Munich Re's primary unit ERGO announced plans to cut approximately 1,000 positions in Germany, citing AI's growing capacity to handle simple, repetitive telephony and operations tasks. Both stories are part of the same trend; one carrier just said it more directly than the other.
GenAI and agentic systems accounted for 68% of new deployments in Q4 2025, with claims leading all categories at 37%. What changed in Q1 2026 is that "deploying" stopped meaning a controlled pilot and started meaning a live customer-facing product.
"AI is starting to run processes rather than just support them. Claims is first because it is complex, data-heavy, and measurable."
— Annabel Ayles, Co-CEO, Evident (March 2026)
The underwriting desk is being redesigned, not just sped up
AI adoption in underwriting is set to jump from 14% today to 70% within three years, per Accenture's survey of 430 senior underwriting executives across 11 countries — published in early 2026.
The underwriting transformation story in Q1 2026 is less about new announcements and more about scale. What was a controlled experiment in 2025 is being pushed into production. The headline numbers are striking: AI has reduced standard underwriting decision times from three to five days down to approximately 12 minutes, while maintaining 99.3% accuracy in risk assessment. For complex policies, processing times have dropped 31% with accuracy improving 43%.
Allianz's BRIAN — a generative AI guidance tool for underwriters — saved approximately 135 working days in information gathering in its first year of deployment. Sixfold, working with Generali, reported 90% underwriter adoption and 50% faster turnaround. These are operational results from deployed systems, not vendor roadmaps.
The more important question in Q1 2026 is whether underwriting is being augmented or restructured. The evidence points to both. Routine submissions — clean risks, standard profiles — are increasingly handled with minimal human touch. The underwriter's role is shifting toward complex cases, appetite decisions, and broker relationships: work that requires judgment and context that AI cannot yet replicate. The gap between insurers that have made this shift and those still evaluating it is widening fast.
BCG estimates underwriting can capture 36% of total AI value in insurance — more than any other single function.
The buying journey is being rewired before carriers notice
In February 2026, broker stocks — Aon, Willis Towers Watson, Arthur J. Gallagher — fell sharply after news of ChatGPT-integrated personal lines insurance apps. Markets are pricing in what carriers are still running working groups to discuss.
The structural shift in distribution is not theoretical. In Q1 2026, three platforms — Insurify, Tuio, and Experian — launched ChatGPT personal lines apps directly in OpenAI's app directory. Munich Re's Next Insurance already allows customers to purchase and bind commercial policies through an AI chatbot with no human agent involved. The buying journey for standardised products is being disintermediated in real time.
BCG's January 2026 analysis frames what comes next: by 2030, AI assistants will compare policies, handle inquiries, and execute purchases autonomously on behalf of consumers. The insurer's first competitive challenge will not be winning the customer — it will be getting into the shortlist surfaced by the customer's AI agent. Insurers whose products and pricing are not structured for machine-readable comparison will be excluded before a human ever enters the conversation.
The uncomfortable reality: roughly 80% of policies are still sold through traditional channels. Agentic distribution will not eliminate brokers or agents quickly. But it will compress their role in low-complexity, price-sensitive lines — and analysts at BofA estimate $15 billion in insurance commissions are tied to transactions now at real risk of AI disintermediation.
A customer asks Apple Intelligence: "What's the best auto insurance for me?" The agent queries 50 carriers, shortlists 3. Your brand is not there — not because your pricing is wrong, but because your pricing is not machine-readable.
— Scenario analysis, ai-risk.co (closer to Q4 2026 than Q4 2030)
Closing note
The narrative to resist is the one where AI in insurance is a gradual, manageable transition that incumbents will navigate at their own pace. The Q1 2026 evidence suggests something more disruptive: carriers cutting call centre headcount while deploying agentic systems, and consumers beginning their buying journeys inside AI assistants rather than search bars. None of this is irreversible. But the window to respond is shorter than most long range plans assume.
