Angles by Hirak Chatterjee

Most carriers are still pricing AI like SaaS — fixed seats,…

Wed, 27 May 2026 09:44:38 GMT

Most carriers are still pricing AI like SaaS — fixed seats, predictable renewals. Agentic workloads don't work that way. The cost unit is consumption, not license, and it scales with usage in ways finance teams have no model for yet.

P&C carriers running claims triage, subrogation, or underwriting agents need consumption-based cost models before scaling. A single complex claims workflow can amplify token usage 55x versus a simple query. Budget overruns aren't a risk — they're the default outcome without architecture guardrails.

The first AI casualties in enterprise aren't human jobs —…

Wed, 27 May 2026 09:44:38 GMT

The first AI casualties in enterprise aren't human jobs — they're the 2021-vintage AI point solutions that carriers bought during the last wave. The real obsolescence risk sits inside existing vendor contracts, not in future AI replacing staff.

P&C carriers running narrow-trained ML models for fraud detection, first notice of loss triage, or document extraction should audit those vendor vintages now. Foundation models handle the same tasks with broader generalization and no per-category retraining. Contracts signed in 2022-2023 may be locking carriers into pre-LLM architecture at post-LLM prices.

P&C carriers aren't building data centers, but they operate…

Wed, 27 May 2026 09:44:38 GMT

P&C carriers aren't building data centers, but they operate in the same communities bearing the costs. The backlash is a leading indicator of regulatory and reputational risk that the industry hasn't priced into its AI governance posture. Being seen as responsible AI users — not just deployers — is now a competitive differentiator with regulators and policyholders.

State insurance regulators are watching public sentiment on AI closely. Carriers that can demonstrate explainable, auditable AI decisions in claims and underwriting — and communicate that clearly — are better positioned as regulatory scrutiny tightens. The trust deficit in AI is now a material risk in a regulated industry.

P&C carriers are pricing cyber insurance on a model that no…

Mon, 18 May 2026 03:04:12 GMT

P&C carriers are pricing cyber insurance on a model that no longer exists. The assumption that offensive capability requires scarce human expertise is gone. Every actuarial table built before Mythos is now a liability, not an asset.

P&C carriers writing cyber lines need to reprice immediately. The frequency and severity assumptions underpinning current policy structures were built for a pre-Mythos threat landscape. This is not an incremental adjustment.

P&C carriers are about to gain access to enterprise-grade…

Mon, 18 May 2026 03:04:12 GMT

P&C carriers are about to gain access to enterprise-grade AI at a fraction of current costs. The efficiency race China was forced into is now a pricing weapon that will structurally compress what carriers pay for AI inference — and what they can charge for AI-enabled services.

Carrier technology budgets built around current AI inference costs will look overstated within 18 months. The more urgent question is whether procurement and vendor governance frameworks can handle models from non-US jurisdictions given regulatory and data residency constraints.

The 'deploy AI and see what happens' era is over.

Mon, 18 May 2026 03:04:12 GMT

The 'deploy AI and see what happens' era is over. The model labs have concluded that AI value capture requires embedding people inside customers to redesign workflows top-down — exactly the pattern of mainframe-era computing. P&C carriers waiting for AI to self-deploy are misreading the moment.

For P&C carriers, this means AI transformation is not a technology project. It is an organisational redesign effort that requires executive mandate, data readiness work, and sustained human-led implementation — more like a Guidewire delivery than a SaaS rollout.

The data centre insurance market is growing fast, but the…

Mon, 11 May 2026 02:31:15 GMT

The data centre insurance market is growing fast, but the accumulation risk is growing faster. When multiple large data centres are clustered within a 20-mile radius — as is occurring in Northern Virginia and Abilene, Texas — a single natural catastrophe becomes a correlated loss across multiple policies and potentially multiple carriers who may not know they share the same geographic exposure. CAT accumulation tools built for conventional property risk were not designed for this geometry.

Property underwriters writing data centre risk need to treat geographic clustering as the primary accumulation lens, not individual site characteristics alone. The question for every data centre submission is not only "what is this building's risk profile?" but "what is our total exposed value within 20 miles of this location across all programmes and all lines?" Carriers who cannot answer the second question are accumulating risk they cannot measure — and therefore cannot price, limit, or reinsure correctly.

The actuarial community now has a statistically credible…

Mon, 11 May 2026 02:31:15 GMT

The actuarial community now has a statistically credible dataset for autonomous vehicle safety across a meaningful exposure base. The debate is no longer theoretical — it is a pricing question. The first carriers to build rate filings that appropriately reflect Waymo's actual loss experience will define the AV insurance pricing baseline that every other carrier either adopts or argues against for the next decade. First-mover advantage in rate filing is a durable structural benefit in a regulated market.

Personal auto underwriting teams that are not already developing AV-specific rate filings are behind the credibility curve. The dataset is available, the statistical power is sufficient for actuarial conclusions, and the regulatory filing landscape for AV-specific products is still open enough for early movers to shape it. The carriers who wait for AV fleet size to become material will be pricing against competitors who have already built and filed AV-specific products backed by 170 million miles of real data.

The trailblazer model is not replicable by adding…

Mon, 11 May 2026 02:31:15 GMT

The trailblazer model is not replicable by adding technology spend. It requires an organisational architecture decision — specifically, the elevation of AI from a departmental capability to an enterprise mandate with board-level visibility and cross-functional accountability. Carriers currently running AI as a portfolio of innovation projects are not on a path to trailblazer status regardless of how many projects they complete. The projects are not the problem.

CEOs and boards of carriers in the 90% majority should ask one direct question: is AI currently an enterprise mandate with board-level accountability and cross-functional ownership, or is it a CIO portfolio item? If the answer is the latter, the organisational architecture change needed to close the trailblazer gap starts with that governance decision — not with the next technology investment. The sequence matters.

The gap between tool acceptance and decision-making trust…

Mon, 11 May 2026 02:31:15 GMT

The gap between tool acceptance and decision-making trust is the most important consumer signal P&C executives have received this year. Customers are not opposed to AI — they are opposed to invisible AI authority. The product design and communication implications are significant and specific: carriers need to make human oversight legible, not merely present. "A human reviews every decision" in fine print is not legible oversight.

Customer experience roadmaps that treat rising AI acceptance scores as permission to accelerate autonomous decision-making are misreading the data. The architecture that builds long-term trust is one where AI is visibly in a supporting role — serving customers faster, with the human decision point explicit and accessible. Carriers who design for visible human oversight will outperform on the retention metrics that matter over a three-to-five year horizon.

The 21% revenue gap between AI trailblazers and the rest is…

Mon, 04 May 2026 00:58:31 GMT

The 21% revenue gap between AI trailblazers and the rest is not primarily a technology gap. Capgemini found that P&C carriers commit 72% of their AI investment to technology and infrastructure, and only 28% to change management — including employee capability, workflow redesign, and leadership alignment. The carriers spending most of their AI budget on technology and wondering why returns are limited have misdiagnosed the constraint.

Before the next AI investment decision, CFOs and CEOs should require three things that most carriers currently cannot provide: a named executive owner accountable for AI outcomes (not the CIO alone), a defined set of business metrics that will determine whether the investment succeeded, and a line connecting AI initiative results to business unit P&L. These are prerequisites for return, not outputs that emerge naturally from deployment.

The ownership ambiguity is the more critical finding.

Mon, 04 May 2026 00:58:31 GMT

The ownership ambiguity is the more critical finding. Unclear ROI is a measurement problem with a known solution. Unclear ownership is a governance failure — and technology investment cannot solve it. An AI programme without a named executive owner, defined accountability, and agreed success criteria will not generate returns regardless of the quality of the models deployed.

The governance question is not where the AI sits technically — it is who is accountable for what it produces commercially. Boards should be asking: who is the named executive responsible for AI outcomes, what are they being measured on, and how does that connect to the firm's financial results? If the answer involves committees, shared ownership, or "the CIO and the business working together," the governance architecture is not yet fit for purpose.

A 60–99% reduction in quote-to-bind time is not an…

Mon, 04 May 2026 00:58:31 GMT

A 60–99% reduction in quote-to-bind time is not an incremental efficiency gain — it is a structural change to what it means to underwrite commercial P&C at volume. Carriers who respond this fast can handle submission volumes that are operationally impossible for carriers running current processes. The market dynamic that follows is consolidation of distribution toward the fastest, most consistent responders — because brokers optimise for execution certainty, not relationships alone.

Commercial lines underwriting managers should examine their own submission flow data before evaluating AI-assisted triage investments — specifically the correlation between response time and bind rate, and the proportion of profitable business lost to processing speed. For most carriers, the business case for underwriting AI is already present in their own declination and lapse data. It is rarely being read that way.

The 72/28 split produces a predictable and observable…

Mon, 04 May 2026 00:58:31 GMT

The 72/28 split produces a predictable and observable outcome: technically capable AI that operationally underperforms because the people using it have not been equipped, the processes around it have not been redesigned, and the managers overseeing it have not been trained to govern it. Technology is the easier problem. The change management constraint is where most P&C AI programmes stall — and where almost no carrier is investing proportionately.

Any AI investment submission to a board should now include a change management budget alongside the technology budget, with named accountability for adoption outcomes. A 70/30 technology-to-change-management allocation is more likely to generate returns than the current industry norm. Finance leaders who are not asking this question of their technology teams are approving investments without the component that determines whether they work.

Consumer acceptance is rising, but it is acceptance of AI…

Mon, 20 Apr 2026 01:44:22 GMT

Consumer acceptance is rising, but it is acceptance of AI as a supporting actor, not a lead. The carriers interpreting the improved approval ratings as a green light for autonomous decision-making are misreading the signal. Tolerance is for AI that helps people faster; resistance is for AI that replaces the human moment in high-stakes transactions. Those are different product design briefs.

Customer experience strategies built around autonomous AI for claims filing, policy renewal, or cancellation will encounter trust resistance that data alone will not resolve. The design question for 2026 is not how much AI to deploy in customer-facing roles — it is how to make human oversight visible and credible when AI is involved. Carriers who make that human oversight legible will outperform those who optimise for autonomy on the metrics that drive long-term retention.

The industry is treating these exclusions primarily as a…

Mon, 20 Apr 2026 01:44:22 GMT

The industry is treating these exclusions primarily as a coverage question for commercial policyholders. The more immediate issue for P&C carriers is the inverse: the same CGL policies carriers use to protect their own operations are being rewritten to exclude AI-related claims. A carrier deploying AI at scale across claims, underwriting, and customer service may be creating liability exposure that its own insurance programme no longer covers.

Risk management and legal teams need to audit their own insurance programmes against the CG 40 47/48 language — specifically whether carriers' own CGL, E&O, and D&O policies have adopted equivalent exclusions, and whether the carrier's AI deployments would fall within those exclusions as a claimant. This is not theoretical: the carriers most exposed are those who deployed AI rapidly in 2024–2025 without a parallel review of their own coverage position.

CAT modelling has historically operated on annual or…

Mon, 20 Apr 2026 01:44:22 GMT

CAT modelling has historically operated on annual or quarterly cycles. AI-enabled real-time exposure monitoring changes the operating model for the CAT function itself — from periodic analysis to continuous portfolio surveillance. That is an organisational and governance change as much as a technology one. A CAT team still structured around quarterly exposure reviews cannot operationalise real-time data even if the platform delivers it.

CROs and CAT managers at property-heavy carriers should assess whether their CAT governance model — reporting cadence, decision rights, underwriting authority thresholds — is structured to act on real-time exposure signals. A platform that delivers live data into a quarterly decision process produces limited value. The competitive advantage of real-time CAT capability is only realised when the organisational structure is redesigned to use it.

The risk transfer architecture for AI liability is being…

Mon, 20 Apr 2026 01:44:22 GMT

The risk transfer architecture for AI liability is being constructed by reinsurers and specialty markets without meaningful primary carrier input. The definitions being established now — how AI errors are categorised, what indemnification triggers look like, how performance warranties relate to underlying policy coverage — will shape the primary market's options for the next decade. Carriers not engaged in this conversation will be price-takers when the market matures.

Chief Underwriting Officers and Chief Risk Officers at primary carriers should be in active dialogue with their reinsurance partners about how AI risk is being structured at the treaty level. The early conversations are happening now, and the conventions being established are not yet hardened. Carriers who engage now can influence the architecture; those who wait will inherit it.

Insurance policy language always outpaces the litigation…

Mon, 20 Apr 2026 01:44:22 GMT

Insurance policy language always outpaces the litigation that interprets it. Carriers rushing to adopt broad AI exclusions are writing language that will be tested in courts that have not yet developed AI coverage doctrine. The exclusions may be broader than carriers can defend, or narrower than intended — the ambiguity is a litigation factory. The carriers who assumed the new ISO forms resolved their AI exposure have not yet seen the claim that tests that assumption.

Underwriting and claims teams adopting the CG 40 47/48 wordings should work with coverage counsel now to establish internal position papers on how the exclusions will be applied in specific claim scenarios — before litigation forces the issue. The carriers who have done this work will respond to the first significant AI coverage dispute from a position of preparation; those who have not will define their position in the adversarial context of a claim, at the worst possible moment.

AI leadership in P&C is not a three-year strategy horizon.

Tue, 31 Mar 2026 02:17:08 GMT

AI leadership in P&C is not a three-year strategy horizon. BCG's framing is that the compounding advantage of early deployment — in data quality, model refinement, and organisational capability — creates competitive distance that later entrants cannot close by outspending. The window they reference is not metaphorical.

Boards and executive teams still treating AI as a technology investment category need to reframe it as a competitive positioning decision with a closing deadline. The BCG finding suggests the question is no longer whether to invest, but whether the current pace and scale of investment is sufficient to remain in a defensible market position by the time the next renewal cycle runs.