AI Strategy Becomes the Top Question on P&C Earnings Calls

S&P Global Market Intelligence headlined its Q1 2026 P&C earnings recap "Strong results, competition, AI dominate agendas." Prior recaps organized around underwriting results, reserve development and premium growth, with AI in a late paragraph on operational efficiency.

The quantitative confirmation is in CB Insights' data: earnings-call mentions of Anthropic rose 199% quarter over quarter to 233, narrowing the gap with OpenAI's 395 to its smallest yet.

Key Takeaways

• 199% quarter-over-quarter growth in Anthropic earnings-call mentions to 233, against OpenAI's 395, driven by AIG, Travelers and Allianz CEOs naming the vendor directly.
• 370,000 submissions processed with a 55% reduction in time to quote and roughly 40% more binding is AIG's disclosure, and it is the only carrier reporting production volume alongside financial outcome.
• 1.5 combined ratio points is Chubb's stated run-rate expense saving target, alongside 85% process automation and a roughly 20% headcount reduction over three to four years.
• $1.5 billion annually is Travelers' disclosed technology spend, the largest among top-10 P&C carriers, with no metric yet tying it to a specific financial outcome.
• Zero questions were asked on any top-15 carrier call about AI regulatory compliance cost, despite the Colorado AI Act taking effect June 30, 2026.

Why the Topic Moved to the Front

The shift is partly about AI and partly about what else there was to ask.

Q1 2026 was the industry's best underwriting quarter in 25 years, with roughly $22 billion of underwriting gain, so the Q&A bandwidth that normally goes to reserve adequacy and catastrophe losses was available. Analysts spent it on deployment timelines, governance structures and measurable outcomes.

The vendor-mention data shows the same movement from the other side. Anthropic mentions rose 199% quarter over quarter to 233 against OpenAI's 395, driven on the insurance side by three CEOs naming the vendor directly: AIG's Peter Zaffino on multi-year model use, Travelers' Alan Schnitzer on the relationship behind a 10,000-employee deployment, and Allianz's Oliver Bate using it as a benchmark for assessing his own company's capability. Outside the top carriers, Hub International reported deploying Claude across more than 20,000 employees with early productivity gains of 85% in targeted use cases.

The disclosure expectation has inverted with it. Two years ago an AI question on a Chubb or Travelers call would have been unusual. Now silence on AI in prepared remarks is the thing that gets noticed.

Four Carriers, Four Kinds of Disclosure

The disclosures differ less in ambition than in what they commit the carrier to being measured against.

Disclosure Tier	Carrier	Key Q1 2026 Disclosure	Accountability Level
Tier 1: Production metrics	AIG	370K submissions processed, 55% faster quoting, 88% AI-adjuster agreement, 29.3% expense ratio	Measurable, trackable quarter to quarter
Tier 2: Structural commitments	Chubb	Global Claims role tied to AI, 85% automation target, 1.5 CR points savings, 20% headcount reduction	Multi-year targets with organizational changes
Tier 2: Governance framework	Progressive	AI Strategy Council, 3-5 year mandate, agentic AI for direct channel "easy policies"	Board-level oversight, directional not numeric
Tier 3: Infrastructure deployment	Travelers	10,000 Anthropic assistants, $1.5B tech budget, TravAI platform for 30,000+	Input metrics (spending, headcount), no output metrics
Tier 3: Qualitative positioning	Allstate	ALLIE platform codes one-third of software, exploring AI to reduce expenses	Product-level proof points, limited financial guidance
Tier 4: Transparency pioneer	Hartford	Published first Algorithmic Impact Assessment (February 2026), AI assistant in underwriting	Governance leadership, limited production disclosure

AIG reported production alongside results. AIG Assist delivered a 30% increase in quoted submissions, a 55% reduction in time to quote and roughly 40% more binding in Lexington Middle Market Property, having processed over 370,000 submissions across eight lines. Claude aligned with professional adjusters 88% of the time out of the box, and agent autonomy moved from under an hour with Claude 2.0 to around 30 hours. The expense ratio came in at 29.3%, down 120 basis points, against a calendar-year combined ratio of 87.3%.

Chubb committed structurally rather than numerically, naming Kevin Rampe Global Claims Officer on April 9, 2026 in a newly created role explicitly linked to AI execution. His path from Global Compliance Officer through General Counsel to Head of North America Claims indicates the company is treating deployment as a governance problem. The December 2025 targets behind it: automate 85% of major underwriting and claims processes, reduce headcount roughly 20% over three to four years, deliver about 1.5 combined ratio points of run-rate savings.

Progressive built the governance layer first. CEO Tricia Griffith, who also holds the title of AI Director, announced an AI Strategy Council with a three-to-five-year mandate, the first board-level AI body at a top-five personal lines carrier. Travelers disclosed the largest deployment, nearly 10,000 Claude assistants with more than 30,000 employees on the TravAI platform, against more than $1.5 billion of annual technology spend.

The distinction matters where an expense assumption is set. AIG's 29.3% is realized and in the filing basis. Chubb's 1.5 points is a target, worth roughly $600 million to $700 million of annual underwriting income on its premium base, contingent on a plan running three to four years. Travelers guided to a roughly 28.5% expense ratio for 2026, consistent with historical ranges and showing no AI inflection, on a deployment less than four months old at quarter close. Three carriers, three different evidentiary standards, all described as AI progress.

What Nobody Asked

The tier structure ranks disclosure specificity, and specificity is not the same thing as deployment maturity. Across every top-15 carrier call, the questions ran in one direction.

Analysts pressed on financial attribution, asking AIG whether the 29.3% expense ratio was sustainable or partly a Corebridge separation effect, and asking CNA about the pace and impact of its initiatives. They pressed on governance structure, treating Progressive's Council and Chubb's Rampe appointment as leading indicators of follow-through. They pressed on vendor strategy, probing Travelers on splitting Anthropic for engineering and OpenAI for customer-facing claims, and AIG on Palantir Foundry lock-in.

Nobody asked what any of it costs to comply with. The Colorado AI Act takes effect June 30, 2026 and the NAIC's 12-state AI evaluation pilot was running concurrently, and neither appeared in a question on any of those calls. Nor did third-party vendor concentration, which the same CB Insights data implies is increasing: mentions consolidating around two providers is the adoption signal and the concentration signal at once.

That asymmetry is what makes the tier map a disclosure ranking rather than a capability ranking. A carrier reporting a 40% binding lift is reporting the numerator of an AI business case whose denominator has not been asked for, and the carrier with the largest disclosed spend and no attributed outcome sits lower in the tier than one with attributed outcomes and a smaller disclosed spend. Grant Thornton found 24% of carriers confident of passing a governance review within 90 days, which is the cost side of the same programs, and it has not yet reached an earnings call.