Chubb Unifies Global Claims Under One AI-Driven Mandate: Why the Rampe Appointment Changes the Math

Chubb named Kevin Rampe Global Claims Officer on April 9, 2026, giving one executive responsibility for claims across 54 countries. He keeps the North America Claims role he has held since 2021 and reports to both Evan Greenberg and John Keogh.

Read against the automation target Chubb published in December, the appointment is a governance decision rather than a technology one. Rampe's career is regulatory, not engineering, and that is the constraint the 85% target actually runs into.

Key Takeaways

• One officer now owns claims across 54 countries, reporting to both the CEO and the COO at Chubb Group level, which puts the claims function in the room where capital is allocated.
• The automation target is 85% of major underwriting and claims processes, touching 70% of the organization within three years, with headcount expected to fall roughly 20% over three to four years.
• The financial prize is about 1.5 combined ratio points. On roughly $56 billion of annualized net premiums written, that is approximately $840 million a year.
• Claims typically runs 35% to 45% of a P&C carrier's operating expenses, so a proportional share of the target puts claims automation near 0.6 combined ratio points, roughly $336 million.
• The Q1 2026 combined ratio of 84.0% against 95.7% is mostly a catastrophe comparison, with cat losses of $500 million against $1.64 billion and the prior year carrying $1.47 billion of California wildfire losses.

A Governance Appointment, Not a Technology One

Rampe's route to the job is the substance of the announcement. He joined Chubb in 2005 as Global Compliance Officer, then served as General Counsel of North America and Global Deputy General Counsel before taking Head of North America Claims in 2021. Before Chubb he was First Deputy Superintendent of the New York State Insurance Department and First Assistant Counsel to Governor George Pataki.

That is the profile you appoint when the hard problem is deploying a standardized workflow across 54 regulatory environments, not when the hard problem is model accuracy. Chubb already employs more than 3,500 engineers globally with hubs in Mexico, Greece, India and Colombia. Keogh's framing in the announcement was that claims "is the fundamental promise of what we sell."

The dual reporting line is the operative detail. Rampe reports to Greenberg and Keogh at the Group level and to Juan Luis Ortega in his North America capacity, which gives the claims function direct access to the executives controlling strategic direction and capital allocation.

The financial position gives him time. Chubb reported Q1 2026 P&C underwriting income of $1.79 billion on an 84.0% combined ratio, with core operating EPS of $6.82, up 85.2%.

Metric	Q1 2026	Q1 2025	Change
P&C Combined Ratio	84.0%	95.7%	Improved 11.7 pts
Current AY ex-Cat CR	82.1%	N/A	+9.8% UW income
Net Premiums Written (Total)	$14.0B	$12.6B	+10.7%
P&C NPW	$11.7B	$10.9B	+7.2%
Life NPW	$2.29B	$1.72B	+33.1%
Net Investment Income	$1.7B	$1.55B	+9.5%
Core Operating ROE	14.0%	N/A	N/A
Catastrophe Losses	$500M	$1.64B	-$1.14B

What 85% Buys, and Where It Lands

The December 2025 investor presentation set the numbers Rampe now owns: 85% of major underwriting and claims processes automated, 70% of the organization touched within three years, run-rate expense savings worth roughly 1.5 combined ratio points, and headcount down about 20% over three to four years. On approximately 43,000 employees that is around 8,600 positions.

Sizing the prize is straightforward arithmetic. Q1 net premiums written of $14.0 billion annualize to roughly $56 billion, so 1.5 combined ratio points is approximately $840 million a year. Claims operations typically account for 35% to 45% of a P&C carrier's operating expenses depending on line mix and LAE allocation. At 40%, a proportional share puts claims at roughly 0.6 combined ratio points, about $336 million.

Claims may carry more than its share, because it is higher-touch than underwriting, has more transactions per premium dollar, and duplicates process across regions that global pricing models already standardize. Underwriting is the more centralized function to begin with.

The reason the organizational change had to come first is that the automation depends on inputs the regions control. Claims data from a personal lines loss in Bangkok, a commercial property claim in London and a workers' compensation case in New York arrive in different formats with different coding schemes and different regulatory fields. A triage model trained on North American structures produces unreliable output on the others, and only a global mandate can force the schema.

The same holds for the technology stack. The 2016 merger with ACE combined two global carriers each carrying its own claims systems, and regional leaders making independent vendor decisions add integration work that the automation then has to absorb.

For a reserving actuary the visible consequence is loss adjustment expense. If FNOL intake, document processing and initial reserve setting automate across a large share of claims, allocated LAE per claim declines measurably and flows into reserve estimates. Faster closure without lower average severity would offset part of that through earlier recognition, which is why the metric to watch is LAE per claim by line rather than aggregate expense ratio. AIG is pursuing the same expense result through line-of-business embedding and Travelers through function-specific agents, both faster to a first result and neither carrying a 54-country consistency requirement.

The 85% Is Touches, Not Dollars

The target's shape is what limits it, and the limit sits exactly where the appointment was aimed.

Claims activities automate at very different rates:

• FNOL intake and triage. High potential. Structured capture, initial severity scoring and routing run with little intervention on straightforward claims.
• Document ingestion and extraction. Moderate to high. Police reports, medical records and repair estimates parse well, with accuracy varying by jurisdiction and format.
• Reserve estimation. Moderate. Severity models set initial case reserves on high-frequency, low-severity claims; complex and litigated claims still need examiner judgment.
• Coverage determination. Low to moderate. Policy language interpretation and exclusion analysis across 54 regulatory frameworks is the part current systems handle least evenly.
• Settlement and payment. High for small claims, where automated offers and straight-through payment below a threshold are already deployed across the industry.

So 85% almost certainly describes the share of claims touches, not the share of claims dollars. A carrier can automate 85% of the activities in a personal auto claim while keeping human oversight on effectively all of its complex commercial liability and environmental claims. The expense reduction per automated touch varies by line and by geography, which is why the 1.5-point target is a run-rate estimate rather than a schedule.

Centralization also concentrates the execution risk it was created to manage. Fragmented regional structures fail regionally. A single global claims design that produces inconsistent results across jurisdictions fails everywhere at once, and the 54-country footprint that makes standardization valuable is the same thing that makes a standardization error expensive.

The cushion funding the patience is partly cyclical as well. The move from a 95.7% combined ratio to 84.0% is driven mainly by the catastrophe comparison, $500 million against $1.64 billion, with the prior-year quarter carrying $1.47 billion of California wildfire losses. The current accident year ex-catastrophe ratio of 82.1% is the durable figure, and it is the one an automation programme has to improve on.