AI and the End-to-End Claims Process: Options, Vendors and the State of Play, Mike Daly

AI and the End-to-End Claims Process: Options, Vendors and the State of Play

The Strategic Landscape

The insurance industry is past the question of whether to apply AI to claims — the conversation has shifted to how fast and how deep. A majority of claims professionals have identified improving processing efficiency and reducing claims cycle time as core goals, and after proof-of-concept work in 2025, insurers are now ready to scale AI fully into production for 2026. The prize is substantial: generative AI alone is estimated to unlock a $100 billion benefit opportunity for P&C insurers through reducing loss-adjusting expenses by 20–25% and claims leakage by 30–50%.

Yet my observation about FNOL automation is that most carriers have barely scratched the surface. Many carriers currently operate four to seven separate systems across policy administration, claims management, billing, underwriting, fraud detection, and multiple third-party vendor applications, which creates integration drag that limits how far AI can reach.

The real opportunity lies in AI threading through the entire claims lifecycle and in parallel with similar projects in other parts of the insurers’ value chain. Differing goals, teams, and organisational separation are a drag on real innovation that requires a strong direction and strategy plus change management coming from the CEO and C-Suite. It’s not all about Agentic AI/Digital Workers either as I explain below.

Despite the top decile of insurers experimenting more I suggest that there is still a big gap between PR and actual progress with advances adopted once exhaustive testing and confidence that compliance, security and full auditability are achieved.

A last comment in this section. I still feel that all the options I describe below are more product and insurer centric rather than designed from the customer backwards. That requires a cultural change and I look forward to seeing a new entrant to the market that enables that approach for insurers.

The Full Claims Lifecycle — Where AI Can Intervene

1. Pre-Loss / Pre-FNOL (Proactive) This is an emerging frontier. During large-scale natural disasters, agentic AI systems can track changes, identify at-risk properties, initiate early claims notifications, and begin outreach before claims are formally filed. Duck Creek has articulated this well: as a hurricane makes landfall, agentic AI can autonomously monitor conditions, synthesise data, identify properties within the highest impact zone, and predict potential damage before even the first claim comes in.

2. FNOL and Intake This is where most current automation sits — chatbots, voice AI, and digital portals. But AI can go further than mere intake: triage, coverage verification, reserve-setting triggers, and fraud scoring can all happen at FNOL. In catastrophe scenarios, AI-assisted claims response times have reportedly fallen from roughly 30 hours to about 30 seconds for certain events.

3. Damage Assessment and Estimation This is one of the most technically mature areas. Computer vision and aerial/drone imagery are transforming field inspection. EagleView Assess uses autonomous drones to capture accurate and consistent property imagery, precise measurements, and AI-driven damage detection, streamlining workflows to expedite insurance claims and document storm-related damage. This has been formally integrated into Verisk's Xactimate and XactAnalysis platforms. Separately, the Tractable–Verisk collaboration enables the identification, classification and measurement of property damage through AI, with the potential to cut the time to settle a property claim from months to as little as one day.

4. Adjudication and STP (Straight-Through Processing) AI-powered claim management systems can process 70–90% of simple insurance claims in a straight-through manner, with claims decisions delivered in minutes rather than weeks. STP rate is now a primary productivity benchmark. Early adopters of Shift Technology's agentic claims platform report 3% lower claims losses, 30% faster claims handling, and a 60% overall automation rate.

5. Fraud Detection Predictive and graph-based AI now runs continuously across claims portfolios, scoring anomalies in real time at intake rather than post-settlement. Shift Technology built its reputation here before expanding to claims orchestration.

6. Complex Claims, Litigation and Negotiations The hybrid approach that developed in 2025 defines success in 2026: AI agents manage volume and precision while adjusters focus on tasks demanding negotiation — liability disputes, large commercial losses, bodily injury, and represented claimants. Agentic AI can surface relevant policy language, prior case precedents, and reserve recommendations to support adjusters rather than replace them.

7. Subrogation A chronically under-automated area. By identifying recovery opportunities, predicting liability, and initiating subrogation files, agentic AI improves accuracy and recovery rates without additional carrier cost.

8. CAT Events at Scale Global insured losses from natural catastrophes totalled $137 billion in 2024, trending toward $145 billion in 2025. For reinsurers and carriers managing CAT events, AI agents can ingest emails, PDFs, spreadsheets, and broker correspondence from cedents, automatically detect the document type, extract relevant claim fields — loss location, policyholder, reserve amount — and create a structured digital record, turning what was a weeks-long manual triage into near-real-time intelligence. Agentech's platform equips adjusters with over 200 AI "coworkers" and automated workflows to improve claims accuracy, accelerate processing, and scale during CAT events and daily claims alike.

The AI Architecture: Three Layers Working Together

No single AI model can meet every need — leading insurers are adopting multi-model AI architectures that support forecasting, automation, content generation, and intelligent orchestration while remaining transparent, governed, and usable by business teams. The three layers are:

Predictive AI — ML models for fraud scoring, reserve prediction, STP routing, litigation propensity
Generative AI — Document drafting, summarisation, coverage interpretation, customer communications
Agentic AI — Orchestrating multi-step workflows autonomously, escalating to humans at defined thresholds, operating across systems in real time

Allianz's "Project Nemo", launched in Australia in 2025, is a real-world example: an agentic AI solution using specialised, task-oriented agents that can independently plan, decide, and collaborate to complete multi-step workflows rather than just answering prompts, launched to automate low-complexity, repetitive claims and significantly reduce processing times.

The Vendor Landscape

Tier 1 — Full Core Platform Providers (Policy + Claims + Billing)

These are the dominant plays for insurers seeking an integrated enterprise platform. Guidewire, Duck Creek, and Applied Systems together generated $918 million in annual recurring revenue in 2024, commanding a combined 15.85% insurance software market share.

Guidewire (ClaimCenter) — The most widely deployed claims system globally for large P&C carriers. Cloud-native (Guidewire Cloud), strong analytics through Guidewire Analytics/Predict, broad partner marketplace (over 200 integrations). Best for large, multi-line P&C carriers needing full lifecycle integration and analytics. High implementation cost and long timelines are the key trade-offs. That leads to insurers evaluating specialist claims platforms to handle the complexity and breadth of line of business claims.
Duck Creek Technologies — Specialises in SaaS core systems, streamlining policy, billing, and claims for P&C clients, with a strong focus on open APIs and modular platform design. Widely regarded as more configurable and agile than Guidewire for mid-market carriers. Duck Creek has been explicit about agentic AI for CAT response.
Majesco — Delivers modern insurance software platforms with strengths in cloud transformation and embedded insurance solutions, enabling insurers to enhance product speed-to-market and automate core processes. Strong in both L&H and P&C. Recently acquired Research Corp to strengthen its mid-market P&C claims position.
Sapiens International — A significant global player, particularly strong in Europe, with its CoreSuite and IDITSuite. Stands out for its industry-specific focus and ability to handle large volumes of data, making it suitable for larger insurance companies with complex operations.
EIS — A cloud-native, API-first platform built for digital-first carriers and insurtechs. Gaining traction as a modern alternative to legacy Guidewire migrations.
Genasys- a PAS coe platform that is no code and woud help apply AI across the whole value chain

Tier 2 — Specialist Claims Platforms and Point Solutions

For insurers who want best-of-breed claims capability without replacing their entire core:

Snapsheet — A modern, cloud-native virtual claims platform. Strong in auto and property, with a growing agentic AI capability. Attractive for carriers wanting to modernise claims without a full core system swap.
Five Sigma — An AI-native claims management platform built specifically with ML at its core, rather than bolted on. Strong in commercial and specialty lines.
Origami Risk — Strong in commercial, self-insured, captive, and TPA use cases. Risk management and claims in a single integrated platform.
Shift Technology — Positioning itself as an AI layer that sits across claims systems. Shift Claims integrates seamlessly with existing claims and core systems as an AI layer, with AI agents and generative AI trained on the insurer's own processes working alongside claims teams to assess, triage, advise, and automate tasks or entire claims.
Spear Technologies (SpearClaims) — A modern platform emphasising accessible AI, allowing business users — not just IT — to configure and govern AI models directly within workflows.

Tier 3 — Specialist Data, Estimation and Ecosystem Providers

These are essential components in a modern claims architecture, often integrated with Tier 1 and Tier 2 systems:

Verisk / Xactimate / XactAnalysis — The dominant property estimating ecosystem in the US. Now integrating drone imagery (EagleView), AI damage detection (Tractable), and 3D modelling (Hover).
Tractable — Computer vision AI for auto and property damage assessment. Auto AI analyses photos to generate repair estimates; Property AI does the same for structural damage.
CCC Intelligent Solutions — The dominant platform in US auto claims, covering estimatics, repair management, total loss, and medical bill review. Deeply embedded in the collision repair network.
Solera / Audatex — CCC's main international competitor in auto claims estimatics. Strong in Europe and globally.
Cotality — Property data and catastrophe loss analytics, widely used for pre-bind risk and post-event CAT response.
Agentech — Purpose-built agentic AI for claims workflows, positioning itself as a "digital coworker" layer deployable across carriers, TPAs, and IA firms.

Tier 4 — Enterprise and Horizontal AI Platforms Entering Insurance

Salesforce (Financial Services Cloud + Agentforce) — Being adopted as a claims front-end and customer engagement layer, particularly where carriers want omnichannel service integrated with CRM. Salesforce is ranked as a top performer in P&C core insurance platforms.
Pegasystems / Appian — Low-code orchestration and decisioning platforms that can wrap legacy claims systems with intelligent automation.
ServiceNow — Being piloted in claims operations management and adjuster task orchestration.

Build vs. Buy vs. Extend

The strategic choice for insurers and TPAs is increasingly nuanced:

License a full platform (Guidewire, Duck Creek, Majesco) — Lowest long-term technology risk, but significant implementation cost (typically £5–50M+ for large carriers), multi-year timelines, and dependency on the vendor's AI roadmap.

License a specialist claims platform (Snapsheet, Five Sigma, Origami) — Faster to value, modern architecture, but requires integration investment with policy and billing systems.

Build AI on top of legacy — Fastest for proof-of-concept; highest long-term technical debt. Many carriers are doing this with Pega, Appian, or direct LLM integrations, but risk creating fragmented point solutions.

AI layer strategy (Shift Technology, Agentech) — An emerging model where AI orchestration sits above existing systems, preserving core system investments while adding intelligence. Attractive for carriers mid-transformation.

Enterprise Integration — The Broader Picture

The coming years will see insurers coalesce AI transformation efforts across underwriting, claims, policy servicing, and customer experience, expanding into actuarial analysis, compliance, finance, producer management, and talent development. This means claims AI cannot be designed in isolation — it must connect to:

Underwriting systems (claims experience feeding back to pricing models)
Customer/CRM platforms (single view of policyholder across service and claims)
Finance and reserving (real-time reserve adequacy)
Reinsurance bordereau (automated cession and loss reporting)
Fraud intelligence networks (cross-carrier data consortia)
Regulatory reporting (IFRS 17, Solvency II, state-level AI compliance)

The regulatory environment is also sharpening. The NAIC has released an AI Principles framework addressing transparency, explainability, and non-discrimination requirements, with state-level regulations regarding claims-specific AI still evolving.

Summary Assessment

The market is bifurcating: carriers who treat AI as a bolt-on to legacy processes are seeing modest FNOL gains, while those embedding AI deeply across the claims lifecycle — connecting assessment, adjudication, fraud, subrogation, and CAT response into a coherent architecture — are achieving structural cost and experience advantages. The vendors that will dominate are those offering genuine multi-model orchestration (predictive + generative + agentic) within governed, auditable workflows, rather than selling individual point solutions. For insurers and TPAs evaluating options, the critical questions are: how open is the API architecture for ecosystem integration, how mature is the vendor's agentic AI roadmap, and how will the platform scale during the increasingly frequent CAT surge periods that now define the industry's most critical operational moments.