AI at Chubb

Chubb Limited is a global property and casualty insurer that serves multinational corporations, mid-size and small businesses, and high-net-worth individuals across dozens of countries. In its December 2025 investor presentation, the company reported $90 billion in total capital, an AA rating from S&P, and $56.4 billion in trailing total premium revenue, ranking seventh among global insurers by revenue and first by income-per-revenue margin according to Chubb’s peer analysis.

Chubb has framed digital transformation as a structural reset of how it operates, underwrites risk, and scales growth, targeting automation of 85% of major underwriting and claims processes within three to four years. The company has also disclosed that it employs more than 3,500 engineers and has built engineering hubs in Mexico, Greece, India, and Colombia to support the shift.

This article examines two internal AI use cases: automated submission intake and underwriting, and AI-driven claims document processing.

Intelligent Underwriting Intake — Reducing the time and manual effort required to move a submission from broker inbox to bound policy.

AI-Enabled Claims Document Processing — Accelerating first contact with claimants by automating the extraction and triage of claims documents.

We begin by examining how Chubb applies AI to compress underwriting cycle times.

Intelligent Underwriting Intake

Underwriting intake has historically been a manual bottleneck: brokers submit risk information in inconsistent formats, and underwriters spend hours re-keying data, checking appetite, and assembling loss runs before they can price an account.

Chubb has identified intake, rating, and pre-underwriting as core targets of its digital enablement program, alongside claims and service, because these functions account for a large share of administrative expense across the 70% of the organization affected by the transformation. Faster, cleaner intake also matters competitively: Chubb’s North America wholesale unit alone processes more than 540,000 submissions a year, and volume at that scale makes manual triage increasingly uneconomical.

Chubb’s Global Platform ingests submissions from agents and customers, along with third-party data, and then uses large language models to extract and enrich the data before it reaches an underwriter. Predictive analytics prioritize quotes and support catastrophe modeling, while automation handles rules-based steps such as appetite checks, loss-run review, and schedule-of-value preparation without human touch. Chubb’s investor materials describe this as combining “intelligent intake,” connected operational data, and embedded AI across the underwriting workflow, rather than a single-point tool bolted onto an existing process.

For underwriters, the change shows up as pre-digested work rather than raw data entry:

Underwriters now receive pre-filled risk data along with a system-generated recommendation on the likelihood of binding, shifting their time toward judgment calls rather than data assembly.

Pre-underwriting steps that once required manual review — appetite checks, loss runs, schedule-of-value preparation — now run with no human touch.

Rating data flows directly into administrative systems, which Chubb says is raising no-touch rates across the intake pipeline.

This use case is past the pilot stage in North America and is in the earlier stages internationally. Chubb’s Global Platform handles 45,000 submissions every month, and the company reports that North America cycle times have dropped from 24 hours to 2 hours, with EMEA still in early stages but already showing handle-time gains.

Endorsement cycle time has fallen from 22 days to 8 days, with a target of under 1 day in 2026. Chubb is candid that this is uneven progress rather than a finished system: automated rate generation is expected to reach only 20% no-touch by year-end for North America commercial lines property and casualty and financial lines, and several markets, including UK and France middle market and Mexico auto and small commercial, are listed as active build targets rather than completed rollouts.

Screenshot from Chubb’s December 2025 investor presentation outlining the company’s digital-enablement strategy across underwriting, service, and claims operations, including AI-driven intake, automation, and claims processing initiatives. (Source: Chubb)

AI-Enabled Claims Document Processing

Claims handling is where underwriting discipline meets customer experience, and speed to first contact is a major driver of both. Chubb reports that average first contact on claims submitted via email has fallen from 24 hours to 3 hours through automation initiatives.

Chubb’s claims systems extract and enrich incoming claims documents, including email submissions, using AI-enabled ingestion and enrichment capabilities deployed across its business processes. On top of that extracted data, Chubb applies predictive models to forecast claims severity and recommend the next-best adjuster action, and routes a feedback loop of claims signals back to underwriting so pricing and risk selection reflect emerging loss patterns rather than stale historical data alone.

For claims adjusters and the underwriters who depend on their output, the practical changes include:

Incoming claims documents are automatically processed, extracted, and enriched before an adjuster opens the file, removing manual sorting and data entry.

Adjusters receive a system-suggested next action and an early severity estimate rather than starting each file from a blank assessment.

Underwriting teams gain a continuous feedback channel from claims outcomes, which Chubb positions as part of a broader effort toward more consistent, better-informed underwriting and claims decisions with faster cycle times.

Screenshot from Chubb’s 4D analytics briefing, highlighting reported business outcomes from the company’s claims analytics platform. (Source:Chubb)

This is one of Chubb’s more mature AI deployments by volume. In North America, the company’s claims systems processed, extracted, and enriched more than 3 million claims documents in 2025 at an 85%+ no-touch rate, and the average first contact time for claims submitted by email has dropped from 24 hours to 3 hours. That said, Chubb’s own materials caution that broader AI and analytics adoption across the company remains experimental in aggregate, with the company expecting adoption to surge industry-wide over the next two to three years — meaning document-level automation has scaled, while more advanced, judgment-heavy claims automation is still ahead.

This analysis examines the following lessons enterprise leaders can draw from Chubb’s AI adoption:

Anchor AI Investment to Cycle-Time Metrics – Chubb ties every use case to a measurable before-and-after time metric — hours, days, or no-touch percentage — which keeps the initiative accountable to operations rather than treated as a general technology upgrade.

Build a Closed Feedback Loop Between Claims and Underwriting – Routing claims severity signals back into underwriting turns AI from a cost-cutting tool into a pricing and risk-selection input, compounding value beyond the initial efficiency gain.

Sequence Regional Rollouts Deliberately – Chubb scaled North America first and treats EMEA and other regions as earlier-stage builds, an approach that lets the company learn from a mature deployment before committing capital elsewhere.